Final Report Summary - ANGIOGENESIS AND HBP (Glycosylation of angiogenic factors by the hexosamine biosynthetic pathway (HBP), a nutrient sensor with a novel metabolic-signaling role in angiogenesis?”)
The hexosamine biosynthetic pathway is deemed a nutrient sensing pathway due to the requirement of amino acids (glutamine), carbohydrates (glucose), lipid metabolism (acetyl- CoA), nucleotides (uridine) and energy (ATP) to yield UDP-acetylglucosamine (UDP-GlcNAc), the major product used in glycosylation. Two types of glycosylation modification on proteins exist: N- and O-glycosylation. N-glycosylation is a necessary regulatory step of many proteins to ensure correct folding, transport and activity whilst O-glycosylation acts as regulatory signal that can modify proteins and impart an effect within minutes, similar to phosphorylation and can compete with phosphorylation on certain sites. These two types of glycosylation allow a short term and long term regulatory link to nutrition.
Blood vessels are lined by quiescent endothelial cells (ECs) that maintain vessel integrity ensuring blood flow and thereby oxygen and nutrient supply. In hypovascularized, hypoxic tissue or in disease (e.g. in growing tumors) they become activated and sprout to (re)vascularize the tissue. An intriguing apparent paradox is that in healthy perfused vessels, i.e. in nutrient-rich conditions, ECs are quiescent, yet they sprout in nutrient-poor conditions. This work was carried out to investigate whether the HBP may co-regulate EC growth versus quiescence, by integrating nutrient availability with pro-angiogenic or pro-quiescence signals.
To start with, I examined the major enzymes of the HBP; GFAT1, GFAT2 and UAP1. GFAT1/2 is the first step that adds the glutamine to glucose to yield glucosamine, this vital step is deemed the rate-limiting step of the pathway. UAP1 adds an uridine to the synthesized GlcNAc to give UDP-GlcNAc, the final product of the HBP pathway. Using genetic manipulation I found that little effect was observed in proliferation or viability of human umbilical cord ECs (HUVECs) when GFAT1 or GFAT2 was decreased whereas inhibition of UAP1 expression resulted in massive decreases in proliferation and glycolysis, and in EC sprouting in an in vitro spheroid sprouting assay (figure 1). These intriguing results have led us to start the generation of EC- specific GFAT1 and UAP1 conditional knock-out mice (VECAD promoter driven Cre).
To gain further insights into the role of GlcNAc in angiogenesis, GlcNAc was applied to endothelial cells under a number of different nutritional conditions. Whilst there was a degree of increase in EC proliferation and vessel sprouting (in vitro spheroid sprouting assay) (figure 2), the outstanding observation was that when GlcNAc was supplemented into EC media that contained no glucose or glutamine, a subset of ECs were able to survive, without media changes, for a number of days (maximum observed = 35 days) in a quiescent state (G0 observed by FACs analysis) which could be brought back into a proliferative state (figure 3). To ensure that the GlcNAc was not being catabolised itself to act as an energy store, 13C-GlcNAc was added to ECs and cell extracts were analyzed by mass spectroscopy to see whether any other metabolites were labelled, which they were not (data not shown). One hypothesis was that these cells survive due to GlcNAc regulated changes in autophagy. Autophagy is the catabolic process of degrading cellular compartments during times of stress, such as nutrient starvation, to provide energy. A major component of the autophagic pathway is ATG5, which when inhibited/ genetically decreased leads to inhibition of autophagy. By this method, GlcNAc treated HUVECs in glucose/ glutamine free media were unable to survive in an autophagy compromised system (figure 4).
Next, the role of O-glycosylation in angiogenesis was investigated. Firstly, O-glycosylation was examined during EC changes from a quiescent phenotype to a proliferative one. I found that there was a significant decrease in OGT and OGA expression (qPCR) and O-GlcNAc modified proteins (Western Blot) in proliferative cells versus quiescent ones (figure 5). To further investigate this we used pharmacological inhibitors. However, all attempts to use published pharmacological O-glycosylation inhibitors failed to decrease O-GlcNAc, as observed by western blot (Figure 6 and data not shown). Genetic knockdown of OGT, the enzyme that modifies proteins with the UDP-GlcNAc, decreased EC proliferation and in vitro sprouting but failed to significantly reduce vessel growth in a vessel specific OGT knockout mouse model (Figure 7). In vitro knock down also resulted in forcing ECs into a senescent state (as observed by beta-galactosidase) at G0 (FACs analysis) but also enlarged their cellular phenotype (as observed by cell-mask staining techniques) (figure 8). To investigate whether OGT controlled players of the autophagy pathway, thus linking the previously observed increased survival of ECs in GlcNAc supplemented media, blotting was carried out on OGT knock down ECs. Two key players were found to be differentially regulated when OGT levels are reduced; LC3B, the lipidated form of LC3 and a major constituent of the autophagosome that is used by western blot as a read out of autophagy activity, and ULK1, involved in a protein complex that initiates autophagy. Upon OGT ablation, LC3B is increased and ULK1 is decreased (figure 9). This is an intriguing result as it would be expected that these protein expressions would be symbiotic given their involvement in the same pathway. Further work needs to carried out to elucidate whether this differential regulation increases or decreases autophagy and what is regulating these paradoxal effects. One possibility is the regulation by O-GlcNAcylation of upstream targets known to link energy/ nutrient stress and autophagy, such as mTOR and AMPK. One hypothesis is that these targets are not deemed energy stressed when O-glycosylated but when this site is lost, either in nutrient starvation or OGT knock-down, then these sensors activate autophagy in an LC3B dependent/ ULK1 independent mechanism.
A unique type of O-linked glycosylation is important for Notch signaling, a downstream pathway of VEGF signaling, and involved in EC quiescence/proliferation. Both the Notch receptors and ligands are modified first with a fucose group by the enzyme protein-O- fucosyltransferase (Pofut). This is then extended by the addition of a GlcNAc, mediated by the fringe proteins. This modification is known to be necessary for receptor ligand interaction. Currently there are no other known targets of the Fringe proteins. Preliminary data indicates an obvious proliferation defect in ECs upon genetic inhibition of radical fringe (one of the three known Fringe isoforms) and Pofut (figure 10). Whilst all three Fringe isoforms are expressed in ECs, it is intriguing that there is a selective proliferative advantage for Radical fringe over the other two. Further investigations are required for whether there is preferential activity on target proteins by the three different fringes.
Lastly, N-glycosylation has been examined by use of pharmalogical inhibitors. N-glycosylation, compared to O-glycosylation that involves only two enzymes, is highly regulated with over 30 enzymes synthesising and adding the oligosaccharides (with GlcNAc as the base unit) to the N- glycosylation motif on specific proteins. The work was carried out mostly on VEGFR2, the major receptor of angiogenesis and highlighted as having 18 N-glycosylation motifs. Examining of VEGFR2 under various nutrient conditions revealed that removal of glucose, but not glutamine, reduced its glycosylation status within 24 hours, which could be rescued by the addition of GlcNAc (figure 11). The first step of N-glycosylation begins with the transfer of a GlcNAc (from UDP-GlcNAc) to a dolichol phosphate group, a step that is inhibited by tunicamycin. Complete deglycosylation of VEGFR2, by tunicamycin, resulted in inhibition of VEGFR2 phosphorylation and downstream activation (as observed by western blot), decreased EC proliferation and reduced in vitro sprouting (figure 12). Similar results are observed by 2-DG, however, given that it is a glucose analogue and inhibits all pathways that use glucose, any results obtained could not specifically be associated with glycosylation (data not shown). Once the N-glycan core has been synthesized, it is further processed by glucosidases and mannosidases to form complex N-glycans that are then added to the proteins. It has been previously considered that only the highly glycosylated VEGFR2 could be phosphorylated, however, kifunensine, an inhibitor of mannosidase I, and swainsonine, an inhibitor of mannosidase II, was able to partially reduce the degree of glycosylation of VEGFR2 but did not inhibit VEGFR2 phosphorylation, EC proliferation or in vitro sprouting (figure 13 and not shown).
The current hypothesis is that tip cells, migrating in hypoxic and nutritional poor areas, will have reduced nutrient uptake resulting in decreased HBP flux. This flux change will decrease VEGFR2 N-glycosylation, Notch GlcNAcylation and O-glycosylation status of stress sensor proteins that activate the formation of the autophagosome, influencing the cell’s responsiveness to VEGF, Notch signaling and stimulation of autophagocytosis respectively.
Future work will involve the elucidation of the exact mechanism by which autophagy is regulated by glycosylation and how this regulates EC survival in a glucose/ glutamine free environment. To identify proteins that are O-glycosylated in ECs, work has been carried out using a GlcNAc derivative that contains an azide group (GlcNAz). This was applied to ECs for 24 hours and then proteins modified with the GlcNAz were isolated and will be run on mass spec for identification. For more direct information regarding potential stress sensor targets of O- glycosylation, tagged AMPK and mTOR will be expressed in ECs and then precipitated followed by blotting for O-GlcNAc. Further work will then follow from these methods depending on targets highlighted.
Blood vessels are lined by quiescent endothelial cells (ECs) that maintain vessel integrity ensuring blood flow and thereby oxygen and nutrient supply. In hypovascularized, hypoxic tissue or in disease (e.g. in growing tumors) they become activated and sprout to (re)vascularize the tissue. An intriguing apparent paradox is that in healthy perfused vessels, i.e. in nutrient-rich conditions, ECs are quiescent, yet they sprout in nutrient-poor conditions. This work was carried out to investigate whether the HBP may co-regulate EC growth versus quiescence, by integrating nutrient availability with pro-angiogenic or pro-quiescence signals.
To start with, I examined the major enzymes of the HBP; GFAT1, GFAT2 and UAP1. GFAT1/2 is the first step that adds the glutamine to glucose to yield glucosamine, this vital step is deemed the rate-limiting step of the pathway. UAP1 adds an uridine to the synthesized GlcNAc to give UDP-GlcNAc, the final product of the HBP pathway. Using genetic manipulation I found that little effect was observed in proliferation or viability of human umbilical cord ECs (HUVECs) when GFAT1 or GFAT2 was decreased whereas inhibition of UAP1 expression resulted in massive decreases in proliferation and glycolysis, and in EC sprouting in an in vitro spheroid sprouting assay (figure 1). These intriguing results have led us to start the generation of EC- specific GFAT1 and UAP1 conditional knock-out mice (VECAD promoter driven Cre).
To gain further insights into the role of GlcNAc in angiogenesis, GlcNAc was applied to endothelial cells under a number of different nutritional conditions. Whilst there was a degree of increase in EC proliferation and vessel sprouting (in vitro spheroid sprouting assay) (figure 2), the outstanding observation was that when GlcNAc was supplemented into EC media that contained no glucose or glutamine, a subset of ECs were able to survive, without media changes, for a number of days (maximum observed = 35 days) in a quiescent state (G0 observed by FACs analysis) which could be brought back into a proliferative state (figure 3). To ensure that the GlcNAc was not being catabolised itself to act as an energy store, 13C-GlcNAc was added to ECs and cell extracts were analyzed by mass spectroscopy to see whether any other metabolites were labelled, which they were not (data not shown). One hypothesis was that these cells survive due to GlcNAc regulated changes in autophagy. Autophagy is the catabolic process of degrading cellular compartments during times of stress, such as nutrient starvation, to provide energy. A major component of the autophagic pathway is ATG5, which when inhibited/ genetically decreased leads to inhibition of autophagy. By this method, GlcNAc treated HUVECs in glucose/ glutamine free media were unable to survive in an autophagy compromised system (figure 4).
Next, the role of O-glycosylation in angiogenesis was investigated. Firstly, O-glycosylation was examined during EC changes from a quiescent phenotype to a proliferative one. I found that there was a significant decrease in OGT and OGA expression (qPCR) and O-GlcNAc modified proteins (Western Blot) in proliferative cells versus quiescent ones (figure 5). To further investigate this we used pharmacological inhibitors. However, all attempts to use published pharmacological O-glycosylation inhibitors failed to decrease O-GlcNAc, as observed by western blot (Figure 6 and data not shown). Genetic knockdown of OGT, the enzyme that modifies proteins with the UDP-GlcNAc, decreased EC proliferation and in vitro sprouting but failed to significantly reduce vessel growth in a vessel specific OGT knockout mouse model (Figure 7). In vitro knock down also resulted in forcing ECs into a senescent state (as observed by beta-galactosidase) at G0 (FACs analysis) but also enlarged their cellular phenotype (as observed by cell-mask staining techniques) (figure 8). To investigate whether OGT controlled players of the autophagy pathway, thus linking the previously observed increased survival of ECs in GlcNAc supplemented media, blotting was carried out on OGT knock down ECs. Two key players were found to be differentially regulated when OGT levels are reduced; LC3B, the lipidated form of LC3 and a major constituent of the autophagosome that is used by western blot as a read out of autophagy activity, and ULK1, involved in a protein complex that initiates autophagy. Upon OGT ablation, LC3B is increased and ULK1 is decreased (figure 9). This is an intriguing result as it would be expected that these protein expressions would be symbiotic given their involvement in the same pathway. Further work needs to carried out to elucidate whether this differential regulation increases or decreases autophagy and what is regulating these paradoxal effects. One possibility is the regulation by O-GlcNAcylation of upstream targets known to link energy/ nutrient stress and autophagy, such as mTOR and AMPK. One hypothesis is that these targets are not deemed energy stressed when O-glycosylated but when this site is lost, either in nutrient starvation or OGT knock-down, then these sensors activate autophagy in an LC3B dependent/ ULK1 independent mechanism.
A unique type of O-linked glycosylation is important for Notch signaling, a downstream pathway of VEGF signaling, and involved in EC quiescence/proliferation. Both the Notch receptors and ligands are modified first with a fucose group by the enzyme protein-O- fucosyltransferase (Pofut). This is then extended by the addition of a GlcNAc, mediated by the fringe proteins. This modification is known to be necessary for receptor ligand interaction. Currently there are no other known targets of the Fringe proteins. Preliminary data indicates an obvious proliferation defect in ECs upon genetic inhibition of radical fringe (one of the three known Fringe isoforms) and Pofut (figure 10). Whilst all three Fringe isoforms are expressed in ECs, it is intriguing that there is a selective proliferative advantage for Radical fringe over the other two. Further investigations are required for whether there is preferential activity on target proteins by the three different fringes.
Lastly, N-glycosylation has been examined by use of pharmalogical inhibitors. N-glycosylation, compared to O-glycosylation that involves only two enzymes, is highly regulated with over 30 enzymes synthesising and adding the oligosaccharides (with GlcNAc as the base unit) to the N- glycosylation motif on specific proteins. The work was carried out mostly on VEGFR2, the major receptor of angiogenesis and highlighted as having 18 N-glycosylation motifs. Examining of VEGFR2 under various nutrient conditions revealed that removal of glucose, but not glutamine, reduced its glycosylation status within 24 hours, which could be rescued by the addition of GlcNAc (figure 11). The first step of N-glycosylation begins with the transfer of a GlcNAc (from UDP-GlcNAc) to a dolichol phosphate group, a step that is inhibited by tunicamycin. Complete deglycosylation of VEGFR2, by tunicamycin, resulted in inhibition of VEGFR2 phosphorylation and downstream activation (as observed by western blot), decreased EC proliferation and reduced in vitro sprouting (figure 12). Similar results are observed by 2-DG, however, given that it is a glucose analogue and inhibits all pathways that use glucose, any results obtained could not specifically be associated with glycosylation (data not shown). Once the N-glycan core has been synthesized, it is further processed by glucosidases and mannosidases to form complex N-glycans that are then added to the proteins. It has been previously considered that only the highly glycosylated VEGFR2 could be phosphorylated, however, kifunensine, an inhibitor of mannosidase I, and swainsonine, an inhibitor of mannosidase II, was able to partially reduce the degree of glycosylation of VEGFR2 but did not inhibit VEGFR2 phosphorylation, EC proliferation or in vitro sprouting (figure 13 and not shown).
The current hypothesis is that tip cells, migrating in hypoxic and nutritional poor areas, will have reduced nutrient uptake resulting in decreased HBP flux. This flux change will decrease VEGFR2 N-glycosylation, Notch GlcNAcylation and O-glycosylation status of stress sensor proteins that activate the formation of the autophagosome, influencing the cell’s responsiveness to VEGF, Notch signaling and stimulation of autophagocytosis respectively.
Future work will involve the elucidation of the exact mechanism by which autophagy is regulated by glycosylation and how this regulates EC survival in a glucose/ glutamine free environment. To identify proteins that are O-glycosylated in ECs, work has been carried out using a GlcNAc derivative that contains an azide group (GlcNAz). This was applied to ECs for 24 hours and then proteins modified with the GlcNAz were isolated and will be run on mass spec for identification. For more direct information regarding potential stress sensor targets of O- glycosylation, tagged AMPK and mTOR will be expressed in ECs and then precipitated followed by blotting for O-GlcNAc. Further work will then follow from these methods depending on targets highlighted.