Glucosamine-Mediated Hexosamine Biosynthesis Pathway Activation Uses ATF4 to Promote "Exercise-Like" Angiogenesis and Perfusion Recovery in PAD

Abstract

Background: Endothelial cells (ECs) use glycolysis to produce energy. In preclinical models of peripheral arterial disease, further activation of EC glycolysis was ineffective or deleterious in promoting hypoxia-dependent angiogenesis, whereas pentose phosphate pathway activation was effective. Hexosamine biosynthesis pathway, pentose phosphate pathway, and glycolysis are closely linked. Glucosamine directly activates hexosamine biosynthesis pathway.

Methods: Hind-limb ischemia in endothelial nitric oxide synthase knockout (eNOS^-/-) and BALB/c mice was used. Glucosamine (600 μg/g per day) was injected intraperitoneally. Blood flow recovery was assessed using laser Doppler perfusion imaging and angiogenesis was studied by CD31 immunostaining. In vitro, human umbilical vein ECs and mouse microvascular ECs with glucosamine, L-glucose, or vascular endothelial growth factor (VEGF₁₆₅a) were tested under hypoxia and serum starvation. Cell Counting Kit-8, tube formation, intracellular reactive oxygen species, electric cell-substrate impedance sensing, and fluorescein isothiocyanate dextran permeability were assessed. Glycolysis and oxidative phosphorylation were assessed by seahorse assay. Gene expression was assessed using RNA sequencing, real-time quantitative polymerase chain reaction, and Western blot. Human muscle biopsies from patients with peripheral arterial disease were assessed for EC O-GlcNAcylation before and after supervised exercise versus standard medical care.

Results: On day 3 after hind-limb ischemia, glucosamine-treated versus control eNOS^-/- mice had less necrosis (n=4 or 5 per group). Beginning on day 7 after hind-limb ischemia, glucosamine-treated versus control BALB/c mice had higher blood flow, which persisted to day 21, when ischemic muscles showed greater CD31 staining per muscle fiber (n=8 per group). In vitro, glucosamine versus L-glucose ECs showed improved survival (n=6 per group) and tube formation (n=6 per group). RNA sequencing of glucosamine versus L-glucose ECs showed increased amino acid metabolism (n=3 per group). That resulted in increased oxidative phosphorylation (n=8-12 per group) and serine biosynthesis pathway without an increase in glycolysis or pentose phosphate pathway genes (n=6 per group). This was associated with better barrier function (n=6-8 per group) and less reactive oxygen species (n=7 or 8 per group) compared with activating glycolysis by VEGF₁₆₅a. These effects were mediated by activating transcription factor 4, a driver of exercise-induced angiogenesis. In muscle biopsies from humans with peripheral arterial disease, EC/O-GlcNAcylation was increased by 12 weeks of supervised exercise versus standard medical care (n=6 per group).

Conclusions: In cells, mice, and humans, activation of hexosamine biosynthesis pathway by glucosamine in peripheral arterial disease induces an "exercise-like" angiogenesis and offers a promising novel therapeutic pathway to treat this challenging disorder.

Keywords: angiogenesis; endothelial cells; glucosamine; oxidative phosphorylation; peripheral arterial disease; reactive oxygen species.

Figure 1.. In vitro, under hypoxia and serum starvation (HSS), ECs treated with glucosamine vs. control have less apoptosis, improved survival and hypoxia dependent angiogenesis without activating VEGFR2-Akt-eNOS-glycolysis or PPP.

A. Western blot for HUVECs treated with L-glucose (5mM), glucosamine (5 mM) or glucosamine (5 mM) in the presence of OGT inhibitor (OSMI-1 10 μM) under HSS conditions for 2 hours. B. CCK-8 survival assay for HUVECs treated with L-glucose (5mM), glucosamine (5 mM) or glucosamine in the presence of OGT inhibitor (OSMI-1 70 μM) (labeled as glucosamine + OGT inh) under HSS conditions for 24 hours. C. Western blot for HUVECs treated with L-Glucose (5 mM), glucosamine (5 mM) or glucosamine (5 mM) + OGT (O-GlcNAc transferase) inhibitor for 18 hours under HSS conditions checking the levels of cleaved caspase 3, total caspase 3 (to assess apoptosis) and PCNA (to assess proliferation). GAPDH was used as loading control D. Quantification of cleaved caspase 3/caspase 3 and PCNA/GAPDH in C. Statistical analysis was done using Kruskal-Wallis test. E. Tube formation for HUVECs treated with L-glucose (5mM), glucosamine (5 mM) or glucosamine in the presence of OGT inhibitor (OSMI-1 70 μM) under HSS conditions. Scale bar is 100 μM. F. Quantification of number of tubes in E. Statistical analysis was done using Ordinary one-way ANOVA. G. Western blot for HUVECs treated with L-glucose (5mM), glucosamine (5 mM) for 6 hours under HSS conditions. H. Quantification p-VR2/VR2, p-Erk/Erk, p-AKT/AKT, p-eNOS/eNOS. Statistical analysis was done using unpaired two tail t-test. I. qPCR for HUVECs treated with glucosamine (5 mM) or L-Glucose (5 mM) for 6 hours under HSS conditions checking the expression of PFKFB3 and G6PD. Statistical analysis was done using unpaired two tail t-test. J. CCK-8 survival assay for microvascular ECs isolated from eNOS KO mice treated with L-glucose (5mM), glucosamine (5 mM) or glucosamine or VEGF165a (100 ng/mL) under HSS conditions for 24 hours. Statistical analysis was done using Ordinary one-way ANOVA. K. Tube formation for microvascular ECs isolated from eNOS KO mice treated with L-glucose (5mM), glucosamine (5 mM) or VEGF165a (100 ng/mL) under HSS conditions. Scale bar is 100 μM. L. Quantification of number of tubes in K. Statistical analysis was done using Ordinary one-way ANOVA. All data represent mean ± SD. *, p-value <0.05, **, p-value <0.01, ***, p-value <0.005, ****, p-value <0.001

Figure 2.. Glucosamine activate the hexosamine biosynthesis pathway (HBP) and improves hypoxia dependent angiogenesis and reduces necrosis without activating glycolysis or PPP.

A. Western blot for ischemic GA muscles (1 day post HLI) from male Balb/c mice treated with IP 600 μg/g glucosamine Vs control for 30 minutes where O-GlcNAcylation is assessed using O-GlcNAc antibody to ensure glucosamine reaches the ischemic muscles. GAPDH was used as housekeeping protein to ensure the loading is equal between the two groups. B. Experimental design showing the daily injection of glucosamine (600 μg/g) to Balb/c mice after hind limb ischemia surgery (HLI). C. Laser doppler perfusion imaging (LDPI) for Balb/c mice treated with either glucosamine or control shows glucosamine improves blood follow recovery starting at day 7. D. Quantification of the LDPI readings in C (ischemic/non-ischemic). Statistical comparison between the two groups was done using multiple comparisons Two-way ANOVA. E. Immunofluorescent images for gastrocnemius (GA) muscles at day 21 after HLI stained with CD31 antibody, blue is DAPI (nuclear staining) and red is CD31 (ECs marker). F. Quantification of CD31 density from E. Statistical comparison between the two groups was done using unpaired two tail t-test. G. Immunofluorescent images for gastrocnemius (GA) muscles of Balb/c mice treated with glucosamine vs. control for 7 days post HLI where tomato lectin 594 (TL594) was injected 15 minutes before muscles collection. TL594 is red and DAPI is blue. H. Quantification of the relative change in TL594 intensity in G. Statistical comparison between the two groups was done using unpaired two tail t-test . I. qPCR for CD31 isolated from Balb/c mice treated with control or glucosamine 3 days post HLI checking the expression of PFKFB3 and G6PD normalized to actin beta. Statistical comparison between the two groups was done using unpaired two tail t-test. J. Western blot for CD31+ cells isolated from balb/c mice treated with control or glucosamine 3 days post HLI. K. Quantification of PFKFB3 and G6PD protein expression from K normalized to GAPDH. Statistical comparison between the two groups was done using unpaired two tail t-test. L. Necrosis score for eNOS knock out mice day 3 post. HLI treated with IP glucosamine (600 ug/g/day) or control. Statistical analysis was done using contingency analysis and Fisher’s exact test which indicates that the glucosamine treated group had a higher percentage of scores of 0 compared to the control group (p < 0.05. *, p-value <0.05, **, p-value <0.01, ***, p-value <0.005, ****, p-value <0.001

Figure 3.. Glucosamine increases ATP production through activation of glutamine mediated mitochondrial oxidative phosphorylation to support glucosamine induced hypoxia dependent angiogenesis.

A. Relative change in total ATP level in HUVECs treated with glucosamine ( 5 mM) vs. L-glucose (5 mM) for 24 hours of HSS conditions. Statistical comparison between the two groups was done using unpaired two tail t-test. B. Volcano blot for RNA seq from HUVECs treated with glucosamine (5 mM) vs. L- glucose (5 mM) for 6 hours under HSS conditions where X axis is the Log2 fold change and y-axis is the –Log10 for the p-value between the groups. N=3/group. C. Heatmap for the expression of metabolic genes of HUVECs treated with glucosamine vs. L- glucose under HSS conditions from RNA seq in A. D. Glycolytic stress test for HUVECs treated with L-glucose (5mM), glucosamine (5mM) or VEGF165a (100 ng/mL). E. Quantification of glycolysis, glycolytic reserve and glycolytic capacity. Statistical comparison between the groups was done using One-way ANOVA. F. Schematic showing glutamine through GFPT1, ASNS and GPT2 can fuel Krebs cycle to generate NADH and FADH2 to drive oxidative phosphorylation. G. Mitostress seahorse assay for HUVECs treated with L-glucose (5mM), glucosamine (5mM) in the presence or absence of glutamine. H. Quantification of basal respiration, maximal respiration and mitochondrial ATP production of G. Statistical comparison between the groups was done using One-way ANOVA . I. Tube formation for HUVECs treated with L-glucose, glucosamine or VEGF165a in the presence or absence of oligomycin under HSS conditions. Scale bar is 50 μM. J. Quantification of the number of tubes in I. Statistical comparison between the groups was done using One-way ANOVA. *, p-value <0.05, **, p-value <0.01, ***, p-value <0.005, ****, p-value <0.001

Figure 4.. Glucosamine vs. VEGF165a is associated with less ROS, via activating serine biosynthetic pathway, and better ECs barrier function.

A. CM-H2DCFDA assay to measure intracellular ROS for HUVECs treated with VEGF165a or glucosamine (5mM) where DAPI is blue and CM-H2DCFDA (marker for intracellular ROS) is green. Scale bar is 100 μM. B. Quantification of fluorescent intensity of CM-H2DCFDA from A Statistical comparison between the two groups was done using unpaired two tail t-test. C. Schematic showing amino acids contributing to redox balance via synthesis of glutathione and NADPH. D. Total glutathione levels in HUVECs treated with L-Glucose (5 mM), glucosamine (5 mM) or VEGF165a (100 ng/mL) for 24 hours under HSS conditions. Statistical comparison between the groups was done using One-way ANOVA. E. Intracellular NAPDH concentrations for HUVECs treated with control (L-glucose), glucosamine, or VEGF165a for 6 hours under HSS conditions Statistical comparison between the groups was done using Mann-Whitney test. F. Western blot for HUVECs treated with control (L-glucose), Glucosamine (5 mM), or VEGF165a (100 ng/mL) 24 hours under HSS conditions checking the expression of MTHFD2. Quantification of MTHFD2 expression normalized to HSP90. Statistical comparison between the groups was done using One-way ANOVA. G. qPCR for the serine biosynthesis pathway normalized to RPLP0. H. NADP/NADPH ratio for HUVECs treated with VEGF165a(100 ng/mL), Glucosamine (5 mM), or glucosamine (5 mM) in the presence of MTHFD2 inhibitor for 8 hours under HSS conditions. Statistical comparison between the groups was done using One-way ANOVA. I. FITC dextran permeability for HUVECs treated with vehicle, glucosamine (5 mM) or VEGF165a (100 ng/mL). Statistical comparison between the groups was done using Mann-Whitney test . J. Transendothelial electrical resistance (TEER) measured by ECIS for HUVECs treated with vehicle, glucosamine (5 mM) or VEGF165a. K. Quantification of resistance measurements at 24 hours. Statistical comparison between the groups was done using Mann-Whitney test. *, p-value <0.05, **, p-value <0.01, ***, p-value <0.005, ****, p-value <0.001

Figure 5.. Induction of O-GlcNAcylation by glucosamine upregulates ATF4 to induce amino acids uptake and metabolism and hypoxia dependent angiogenesis.

A. Transcriptional regulatory relationships unraveled by sentence-based text-mining (TRRUST) for the top 1000 differential genes in glucosamine vs. L-glucose treated HUVECs showing the transcriptional factors upstream of these genes. B. Western blot for HUVECs treated with control (L-glucose), Glucosamine (5 mM), or VEGF165a (100 ng/mL) for 24 hours under HSS conditions checking the expression of ATF4. C. Quantification of ATF4 expression normalized to HSP90. Statistical comparison between the groups was done using Kruskal-Wallis test. D. qPCR for ischemic CD31+ cells isolated from GA muscles of Balb/c mice treated with glucosamine vs. PBS, 3 days post HLI checking the expression of ATF4. Statistical comparison between the groups was done using unpaired two tail t-test. E. qPCR for HUVECs after knocking down ATF4 checking the expression of ATF4. Statistical comparison between the groups was done using unpaired two tail t-test. . F. Tube formation for HUVECs after ATF4 knocking down treated with control (L-glucose 5 mM), glucosamine (5 mM) or VEGF165a (100 ng/mL) under HSS conditions. Scale bare is 100 μM. G. Quantification of the number of tubes in F. Statistical comparison between the groups was done using Kruskal-Wallis test. *, p-value <0.05, **, p-value <0.01, ***, p-value <0.005, ****, p-value <0.001

Figure 6.. Exercise-induced angiogenesis in PAD patients is associated with increased O-GlcNAcylation in endothelial cells.

A. Muscle sections from patients with PAD before and after 12 weeks of supervised exercise stained with DAPI (blue), O-GlcNAc antibody (green) and CD31 (purple) for ECs. Representative images were taken with 40X magnification. B. Quantification of capillary density (CD31 numbers) per magnification field (per 20X). C. Quantification of colocalized O-GlcNAc and CD31 to total CD31. Statistical comparisons between the groups in B and C were done using multiple comparisons Two-way ANOVA. *, p-value <0.05, **, p-value <0.01, ***, p-value <0.005, ****, p-value <0.001.