Inhibition of protein glycosylation is a novel pro-angiogenic strategy that acts via activation of stress pathways

Abstract

Endothelial cell (EC) metabolism is thought to be one of the driving forces for angiogenesis. Here we report the identification of the hexosamine D-mannosamine (ManN) as an EC mitogen and survival factor for bovine and human microvascular EC, with an additivity with VEGF. ManN inhibits glycosylation in ECs and induces significant changes in N-glycan and O-glycan profiles. We further demonstrate that ManN and two N-glycosylation inhibitors stimulate EC proliferation via both JNK activation and the unfolded protein response caused by ER stress. ManN results in enhanced angiogenesis in a mouse skin injury model. ManN also promotes angiogenesis in a mouse hindlimb ischemia model, with accelerated limb blood flow recovery compared to controls. In addition, intraocular injection of ManN induces retinal neovascularization. Therefore, activation of stress pathways following inhibition of protein glycosylation can promote EC proliferation and angiogenesis and may represent a therapeutic strategy for treatment of ischemic disorders.

Fig. 1. ManN is a mitogen/survival factor for eye-derived microvascular endothelial cells.

a Bell-shaped effects of Mannosamine (ManN) on bovine choroidal microvascular endothelial cells (BCEC) proliferation. BCECs were treated with various concentrations of ManN ranging from 0.5 μM to 1 mM for 5–6 days, with or without 5 ng/ml VEGF. At the end of the experiment, cells were fixed and stained with crystal violet. Cell-covered areas in various treatment groups were quantified by ImageJ software. b Cell numbers were quantified by AlamarBlue® assay and fluorescence was measured at 530 nm/590 nm. n = 3 independent samples. c Effect of ManN on bovine retinal microvascular endothelial cells (BREC) proliferation. n = 3 biologically independent samples. d Effects of hexosamines other than ManN on BCEC proliferation. Each treatment group was tested in duplicate. e BCEC confluent monolayers were scratched with 1 ml pipet tip, washed and then incubated for 40 h in low-glucose DMEM containing 1% FBS. n = 3 independent samples. Scale bar = 400 µm. Images were taken and gaps between leading wound front were quantified using AxioVision LE Rel.4.4 software. Representative images from crystal violet staining are shown. f Effects of ManN in BCEC transwell migration assay. n = 4 independent samples. Asterisks indicate a significant difference compared with control. When statistical analysis was done using a different control, a line was used between specific groups. A representative experiment is shown in two independent studies. Data are means +/− SD, Statistical analysis was done by two-tailed, two-sample unequal variance t test. *p < 0.05, **p < 0.01. Data are provided as a Source data file.

Fig. 2. Activation of ERK, AKT, mTOR, AMPKα, CREB, ACC, and eNOS is not unique to ManN.

Enhanced activation of ERK (Thr 202/Tyr 204), AKT (Ser 473), and CREB (Ser 133) in BCECs following treatment with ManN together with VEGF for various times (a) or following pre-treatment with ManN for 8 h, followed by VEGF stimulation for 15 min (b). c BCECs were treated with 40 μM ManN, ManNAc, or mannose for various times. Total mTOR, ACC, eNOS, AMPKα, ERK, AKT, CREB as well as phosphorylation of mTOR (Ser 2448), ACC (Ser 79), eNOS (Ser1177), AMPKα (Thr 172), ERK (Thr 202/Tyr 204), AKT (Ser 473), and CREB (Ser 133) were examined by western blot analysis. β-actin served as the loading control. Molecular weight (kDa) was labeled at the right. A representative experiment is shown from two independent studies. Data are provided as a Source data file.

Fig. 3. ManN specifically activates the JNK pathway in BCECs.

a BCECs grown in growth media (GM: low-glucose DMEM containing 10% bovine calf serum (BCS), 10 ng/ml VEGF and 5 ng/ml bFGF) were switched to growth factor-free media, followed by treatment with ManN or Mannose at 4 μM–4 mM. Four hours later, cell lysates were collected and subjected to western blot analysis for phosphorylated JNK (Thr 183/Tyr 185), p38 (Thr 180/Tyr 182), and ERK (Thr 202/Tyr 204), as well as total JNK, p38, and ERK. b ManN, but not mannose, could activate JNK and its downstream c-Jun. β-actin served as the loading control. For each study, a representative experiment is shown from two to three independent studies. c BCECs plated in 96-well plates were attached, pre-treated with the specific JNK inhibitor SP600125 (5 μM) for 2 h, followed by ManN at either 40 μM or 2 mM, with or without 5 ng/ml VEGF. Six days later, cell proliferation was quantified after addition of AlamarBlue®. n = 3 independent samples. d Screening of siRNAs against JNK1 and JNK2. Twenty-four hours after siRNA transfection, BCECs were lysed and proteins were subjected to western blot analysis. β-actin served as the loading control. Quantification of target knockdown is shown. e A representative experiment shows that ~80% knockdown of JNK1 and/or JNK2 by two independent siRNAs was associated with a significant reduction in the stimulatory effects of ManN on BCEC proliferation. n = 3 independent samples. Data are means +/− SD, Asterisks indicated a significant difference compared with the control. When statistical analysis was done using a different control, a line was used between specific groups. Statistical analysis was done by two-tailed, two-sample unequal variance t test. *p < 0.05, **p < 0.01. Data are provided as a Source data file.

Fig. 4. ManN affects protein glycosylation.

a Reduction of VEGFR2 molecular mass following ManN treatment. BCECs were treated with various hexosamines, their derivatives, and monosaccharides at 40 μM or with VEGF at 5 ng/ml for 24 h. VEGFR2 western blot analysis was performed. b Dose-dependent effects of ManN on VEGFR2 molecular mass in BCECs. c Mannose could dose-dependently reverse the effect of 2 mM ManN on VEGFR2 molecular mass change, whereas mannose alone had no effect even at 10 mM. d 5 mM mannose could completely reverse the bell-shaped effects of ManN on BCEC proliferation with or without 5 ng/ml VEGF. BCECs plated in 96 wells were allowed to attach, followed by ManN addition. Two hours later, cells were treated with different concentrations of Mannose, with or without VEGF. Six days later, cell proliferation was quantified using AlamarBlue®. n = 3 independent samples. e Effects of ManN are reversible. BCECs, after treatment with 40 μM ManN for 24 h, were washed three times with low-glucose DMEM. Cells were kept in low-glucose DMEM for additional 8 or 24 h. VEGFR2 western blot analysis was performed. f Reduction of molecular mass of VEGFR2, Neuropilin-1, CD31, and c-met in HUVEC following ManN treatment at various concentrations. g Reduction of molecular mass of VEGFR2, β1 integrin, and bFGFR1 in hDMVECs by ManN at various concentrations. β-actin served as the loading control. Data are means +/− SD, asterisks denote a significant difference compared with the control. For each study, a representative experiment is shown from two to five independent studies. Statistical analysis was done by two-tailed, two-sample unequal variance t test. *p < 0.05, **p < 0.01. Data are provided as a Source data file.

Fig. 5. ManN specifically induces expression of unfolded protein response (UPR) responsive proteins.

a BCECs were cultured in growth media (GM) until ~80% confluency. Media were changed to growth factor-free media containing 10% BCS in the presence or absence of 40 or 400 μM of ManN or mannose for various times. At the end of each incubation, cell lysates were collected, proteins were separated on 4–12% Bis-Tris gel for western blot analysis. b Cells were treated with various concentrations of ManN, mannose, 5 ng/ml VEGF, or a combination of ManN and VEGF for 24 h. Cell lysates were separated on NuPAGE 3–8% Tris-Acetate gel for western blot analysis. c 4-PBA, but not TUDCA, could effectively block the induction of CHOP in BCECs, accompanied by a restoration of expression of transcription factor ATF-6 upon 400 μM ManN treatment. BCECs were pre-treated with 2 mM 4-PBA or 500 μM TUDCA, two chemical chaperons. Sixteen hours later, cells were switched to growth factor-free media for 4 h in the presence of ManN. GM: growth media. d 4-PBA significantly blocked the bell-shaped effects of ManN on BCEC proliferation. Pre-treatment of cells with 1 mM 4-PBA for 8 h abrogated additive effects of 40 μM ManN and 5 ng/ml VEGF and protected cells from toxic effect induced by 2 mM ManN. n = 3 independent samples. For each study, a representative experiment is shown from two to three independent studies. Data are means +/− SD, asterisks indicate a significant difference compared with control. Statistical analysis was done by two-tailed, two-sample unequal variance t test. *p < 0.05, **p < 0.01. Data are provided as a Source data file.

Fig. 6. Effects of ManN on non-ECs.

Effects of ManN on non-endothelial cells of bovine, mouse, or human origin. ManN did not promote growth of Calu6 (a), A673 (b), U87MG (c), and 4T1 (d) tumor cells. Ten-percent FBS was used as positive control for Calu6 and A673, whereas 10 ng/ml bFGF and 1 μg/ml human apo-transferrin were used as positive controls for U87MG and 4T1, respectively. Similarly, no increases in proliferation were induced by ManN on AML12 (e), bovine pituitary cells (f), NIH3T3 cells (g), human RPEs (h), human dermal fibroblasts (i), and human keratinocytes (j), alone or in combination with growth factors. Proliferation quantification was performed using AlamarBlue® or MTS (for 4T1 cells). n = 3 independent samples. Inserted are representative western blot analyses showing dose-dependent effects of ManN and mannose at 400 μM (2,4) and 2 mM (3,5) on bFGFR1 or β1 integrin (for 4T1, AML12, NIH3T3 cells, human skeletal muscle cells, human dermal fibroblasts, and human keratinocytes) molecular mass compared to the untreated control (1). β-actin served as loading control. GM: growth media. Proteins were separated on NuPAGE 3–8% Tris-Acetate gel for western blot analysis. For each study, a representative experiment is shown from two independent studies. Asterisks indicate a significant difference compared with control. When statistical analysis was done using a different control, bracket was used between specified groups. Data were means +/− SD of the mean or an average when n = 2. Statistical analysis was done by two-tailed, two-sample unequal variance t test. *p < 0.05, **p < 0.01. Data are provided as a Source data file.

Fig. 7. Effects of protein glycosylation inhibitors on BCEC proliferation.

a Dose-dependent stimulation of BCEC proliferation by various inhibitors of glycosylation. Inhibitors were added at concentrations ranging from 0.01 to 100 μM for 3 days, with or without 5 ng/ml VEGF. At the end of the experiment, cells were fixed and stained with crystal violet. A representative experiment is shown. Kifunesine (Kif), an ERα-1,2-mannosidase I and Golgi α-mannosidase I inhibitor; Castanospermine (Cas), an a-glucosidase inhibitor. Cell-covered areas in various treatment groups were quantified by ImageJ software. b Dose-dependent effects of Kif and Cas in promoting BCEC proliferation with or without 5 ng/ml VEGF. n = 3 independent samples. c Both inhibitors reduced VEGFR2 molecular mass and induced Bip expression in a dose-dependent fashion as assessed by western blot analysis. Proteins from total cell lysates were separated using 3–8% Tris-Acetate gel. BCECs were treated with various inhibitors for 24 h. Quantification of western blots was done by densitometry. β-actin was the loading control. d Acceleration of closure of monolayer gaps by Kif and Cas in BCEC scratch assay, with controls for Kif (H₂O) and Cas (DMSO). Gaps were quantified using AxioVision LE Rel.4.4 software. n = 3 independent samples. Scale bar = 400 µm. e Activation of AKT and JNK in BCECs by glycosylation inhibitors at 40 μM and VEGF at 10 ng/ml. However, Cas did not activate ERK. Quantification of phosphorylated AKT, JNK, and ERK was done by densitometry analysis relative to total protein. f Pre-treatment of BCECs with 5 μM SP600125 for 2 h significantly blocked the effects of both glycosylation inhibitors on BCEC proliferation. n = 3 independent samples. A representative experiment is shown from two to four independent studies. Data shown are means +/− SD. Statistical analysis was done by two-tailed, two-sample unequal variance t test. *p < 0.05, **p < 0.01. Data are provided as a Source data file.

Fig. 8. Topical application of ManN and VEGF stimulated angiogenesis and accelerates wound healing.

a Wounds were made on the dorsal skin of mice by 6 mm punch. VEGF and ManN each was administered daily at 20 μg per wound in 25 μl PBS for the first 4 days, with PBS as control. A 10-day wound healing study with 5 mice in each group. Wound closure rate (%) was quantified by ImageJ software in two independent studies. Asterisks indicated a significant difference compared with the control at each time point. b A 4-day wound healing study with images of the wound healing process at day 1, day 2, and day 4. n = 5 animals/treatment group. c Representative images of immunohistochemical staining for CD31 in PBS control group and in VEGF and ManN combination groups (scale bar = 200 μm). d Quantification of CD31-positive blood vessel (red dotted circles) density around the wound areas was counted by eyes under microscopy (×20 magnification). Data are means +/− SD. Statistical significance was further confirmed using Wilcoxon rank-sum test between treatment groups of interest. Asterisks indicated a significant difference compared with the PBS control. For each study, a representative experiment is shown. n = 3 animals/treatment group. When statistical analysis was done using a different control, a line was used between specific groups. Statistical analysis was done by two-tailed, two-sample unequal variance t test. *p < 0.05, **p < 0.01. Data are provided as a Source data file.

Fig. 9. ManN accelerates blood flow recovery in a mouse ischemic hindlimb model.

a Serial laser Doppler analysis of blood perfusion in hindlimbs of ManN-treated, Kif-treated, and control mice. Different colors were used to indicate blood perfusion in the ischemic limb (ligated; left side) to nonischemic limb (sham; right side). Representative images at week 0 and week 1 were shown. b Quantification of blood perfusion ratio between region 2 (ischemic; left limb) and region 1 (nonischemic; right limb), n = 8 animals/treatment group. c Three weeks after surgery, skeletal muscle tissues were harvested and fixed. CD31 immunostaining on these tissue sections was performed to label the vasculature. H&E staining was also performed. Representative CD31 staining and H&E histological image of ischemic hindlimbs 21 days after surgery were shown. Scale bar = 50 µm. d Quantification of vascular density by CD31 immunostaining was performed using ImageJ software, n = 8 animals/treatment group, three independent experiments; data are means +/− SEM. Statistical analysis was done by two-tailed, two-sample unequal variance t test. *p < 0.05, **p < 0.01. Data are provided as a Source data file.

Fig. 10. ManN promotes retinal neovascularization in mice.

a Intravitreal injection of ManN increases blood vessel density. Adult mice were intravitreally injected once 500 ng of ManN, Kif or 200 ng of bFGF. PBS was used as vehicle control. Seven days after injection, PFA-fixed retinas were subjected to CD31 immunofluorescence. Representative images of CD31-positive vessels are shown. n = 10 animals/treatment group, 3 independent experiments, scale bar = 50 µm. b Vascular density was determined with ImageJ software, n = 10 animals/treatment group, 3 independent experiments. Data were means +/− SEM. Statistical analysis was done by two-tailed, two-sample unequal variance t test. *p < 0.05, **p < 0.01, ***p < 0.001. Data are provided as a Source data file.