Excessive glycosylation drives thoracic aortic aneurysm formation through integrated stress response

Abstract

Background and aims: Thoracic aortic aneurysms and dissections (TAADs) are depicted by aortic medial degeneration characterized by glycan-rich matrix accumulation. Marfan syndrome (MFS) is the most common inherited connective tissue disorder associated with TAAD. Although vascular smooth muscle cell metabolic dysfunction has emerged as a pathogenic driver of TAAD, surgical repair remains the mainstay of treatment. This study aimed to investigate the role of the hexosamine biosynthetic pathway (HBP) in sporadic and genetic TAAD pathophysiology.

Methods: Hexosamine biosynthetic pathway activation was analysed in aortas from an MFS mouse model, a β-aminopropionitrile-induced non-genetic TAAD model, and patients with sporadic TAAD using transcriptomic and metabolomic approaches. Aortic dilatation was monitored by ultrasound imaging. Pharmacological inhibition of HBP and integrated stress response (ISR) was performed to assess their therapeutic potential.

Results: Hexosamine biosynthetic pathway was up-regulated in both an MFS mouse model and β-aminopropionitrile-induced TAAD, as well as in aortic samples from MFS and sporadic TAAD patients. Enhanced HBP activity contributed to aortic dilatation and medial degeneration via vascular smooth muscle cell dysfunction and ISR activation. Inhibition of HBP or ISR reversed these effects in the MFS model.

Conclusions: The HBP-ISR axis drives medial degeneration in TAAD. These findings identify HBP and ISR as a potential target in TAAD of both genetic and non-genetic origin.

Keywords: Aortic medial degeneration; Hexosamine Biosynthetic pathway; Integrated stress response; Marfan Syndrome; Thoracic aortic aneurysm.

Structured Graphical Abstract

In thoracic aortic aneurysms and dissections, increased activity of the hexosamine biosynthetic pathway (HBP) promotes glycan-rich matrix accumulation via up-regulation of the rate-limiting enzyme glutamine-fructose-6-phosphate transaminase 2 (GFPT2). Exogenous glucosamine fuels the HBP, inducing aortic medial degeneration in wild-type mice. This glycan-rich matrix is characterized by a high negative charge, leading to increased accumulation of cations such as Na⁺ raising osmotic pressure and contributing to medial thickening, elastin fragmentation, and mechanical stress. Additionally, excessive glycosylation driven by HBP activates the integrated stress response (ISR), reducing the canonical protein translation rate and promoting activating transcription factor 4 (ATF4) translation. Pharmacological inhibition of HBP via DON/FR054 or ISR via ISR inhibitor (ISRIB) reverses aortic dilatation and medial degeneration in Marfan mice. Circulating HBP metabolites and glycosaminoglycans emerge as potential biomarkers in Marfan syndrome (MFS). The metabolic HBP–ISR axis emerges as a promising therapeutic target in thoracic aortic aneurysms and dissections.

Figure 1

The hexosamine biosynthetic pathway genes are up-regulated in the aortas of a murine model Marfan syndrome. (A) Schematic of hexosamine biosynthetic pathway involvement in O-/N-linked glycosylation and glycosaminoglycan biosynthesis. (B) Heat map of hexosamine biosynthetic pathway, O-glycosylation, N-linked glycosylation, and GAG-related genes from RNA-sequencing analysis of aortic medial tissue from 24-week-old Fbn1^+/+ and Fbn1^C1041G/+ mice (n = 4). (C) Quantitative reverse transcription polymerase chain reaction analysis of Gfpt2, Uap1, Gnpat1, Mgat2, Ogt, Oga mRNA relative expression in aortic extracts from 20/24-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice. (D) Quantitative reverse transcription polymerase chain reaction analysis of Gfpt2 mRNA relative expression in aortic extracts from 4- or 8-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice. (E) Representative confocal imaging of Gfpt2, Uap1, O-GlcNac (red); elastin (autofluorescence), and DAPI (4′,6-diamidino-2-phenylindole)-stained nuclei in the ascending aorta from 20-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice (n = 6). (F) Representative immunoblot analysis of O-GlcNac proteins, Gfpt2 in aortic extracts from 20-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice (n = 6). Actin was used as a loading control. (G) Representative confocal imaging of wheat germ agglutinin (grey), smooth muscle actin (Sma, red), elastin (green, autofluorescence), and DAPI-stained nuclei (blue) in the ascending aorta from 20-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice (n = 6). (H) Representative immunoblot analysis of Gfpt2, Uap1, O-GlcNac proteins in extracts from primary murine vascular smooth muscle cells transduced with shFbn1 or shScr (ShControl) for 5 days (n = 3). Actin was used as a loading control. Data are mean ± standard error of the mean. Student’s t-test. *P < .05, **P < .01, ***P < .001 vs Fbn1^+/+ mice

Figure 2

The hexosamine biosynthetic pathway is increased in the aortas of a murine model of sporadic thoracic aortic aneurysms and dissections and patients. (A) Experimental design, A–F, 4-week-old male C57/BL6 wild-type mice were treated with β-aminopropionitrile (0.5%) in drinking water for 42 days. (B) Evolution of maximal AsAo diameter (mm). (C) Percentage of increase of AsAo compared with baseline. (D) Representative histological staining of the ascending aorta with Elastin van Gieson and Alcian Blue. (E) Representative immunoblot analysis and quantification of O-GlcNac proteins, Gfpt2 in aortic extracts from 4-week-old control and β-aminopropionitrile-treated mice. Tubulin was used as a loading control. (F) Representative confocal imaging of Gfpt2, O-GlcNac; wheat germ agglutinin, elastin (autofluorescence), and DAPI-stained nuclei and quantification (H), in the ascending aorta from control and β-aminopropionitrile-treated mice. (G) Representative confocal immunostainings and quantification of O-GlcNac, GFPT2; and wheat germ agglutinin elastin (autofluorescence), and DAPI-stained nuclei in medial layer of aortic sections from control and sporadic non-genetic thoracic aortic aneurysms patients. Data are mean ± standard error of the mean. Student’s t-test. **P < .01, ***P < .001, ****P < .001 vs control

Figure 3

Boosting hexosamine biosynthetic pathway induces aortic dilatation and medial degeneration. (A) Experimental design, A–E, 20-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice were with or without glucosamine in drinking water for 28 days. (B) Representative aortic ultrasound images after 28 days of control or glucosamine treatment. Discontinuous red lines mark the lumen boundary, scale bar 1 mm. (C) Evolution of maximal AsAo and AbAo diameter. (D) Representative histologic staining with Elastin van Gieson and Alcian blue in the AsAo and quantification of elastin breaks (E). (F) Representative confocal imaging of O-GlcNac proteins (red), elastin (green, autofluorescence), and DAPI-stained nuclei (blue) in the ascending aorta from Fbn1^+/+ mice treated with or without glucosamine for 28 days (n = 6). (G) Representative confocal imaging of wheat germ agglutinin, Smooth muscle actin (Sma), elastin (autofluorescence), and DAPI-stained nuclei in the ascending aorta from Fbn1^+/+ mice treated with/without glucosamine for 28 days (n = 6). (H) Representative immunoblot analysis of Gfpt2 and O-GlcNac proteins in LV-Mock (control) or Lv-GFPT2 transduced primary murine vascular smooth muscle cells for 5 days (n = 3). (I) Experimental design, J–M, 12-week-old C57BL/6 wild-type mice were injected with LV-Mock or Lv-GFPT2 lentivectors for 28 days. (J) Evolution of maximal AsAo diameter after inoculation of LV-Mock (control) or Lv-GFPT2 lentiviral vectors. (K) Representative confocal imaging GFP or GFPT2, elastin (autofluorescence), and DAPI-stained nuclei in ascending aortas from mice after 30 days of inoculation of LV-Mock (control) or Lv-GFPT2 lentivectors (n = 6). (L) Representative confocal imaging of wheat germ agglutinin, O-GlcNac proteins (red), elastin (autofluorescence), and DAPI-stained nuclei in ascending aortas from mice after 30 days of inoculation of LV-Mock (control) or Lv-GFPT2 lentivectors (n = 6). (M) Representative histologic staining with Elastin van Gieson and Alcian blue in the AsAo and quantification of elastin breaks in ascending aortas from mice after 30 days of inoculation of LV-Mock (control) or Lv-GFPT2 lentivectors. Data are mean ± standard error of the mean. Two-way repeated measurements analysis of variance (C, J), by one-way analysis of variance (E) or t-test (M). *P < .05, **P < .01, ***P < .001, ****P < .0001 vs control

Figure 4

Hexosamine biosynthetic pathway inhibition by 6-diazo-5-oxo-L-norleucine restores aortic dilatation and medial degeneration in Marfan syndrome mice. (A) Representative flow cytometry histograms and statistical analysis of O-GlcNAc and wheat germ agglutinin staining of primary vascular smooth muscle cells from Fbn1^C1041G/+ and Fbn1^+/+ mice treated with or without 6-diazo-5-oxo-L-norleucine for 24 h. (B) Experimental design, B–J, 20/22-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice were infused with minipumps with saline (control) or 6-diazo-5-oxo-L-norleucine for 28 days. (C) Representative aortic ultrasound images after 28 days of control or 6-diazo-5-oxo-L-norleucine treatment. Discontinuous red lines mark the lumen boundary, scale bar 1 mm. (D) Evolution of maximal AsAo and AbAo diameter. (E) Representative histological staining with Elastin van Gieson and Alcian blue in the AsAo, quantification of elastin breaks and aortic medial thickness (F) (n = 6). (G) Representative confocal imaging of O-GlcNac proteins or SMA, wheat germ agglutinin, elastin (autofluorescence), DAPI-stained nuclei in the ascending aorta from Fbn1^+/+ and Fbn1^C1041G/+ mice treated with/without 6-diazo-5-oxo-L-norleucine for 28 days (n = 6). (H) Representative immunoblot analysis and quantification of O-GlcNac proteins in aortic extracts from Fbn1^C1041G/+ and Fbn1^+/+ mice with saline or 6-diazo-5-oxo-L-norleucine for 28 days (n = 4). (I) Glycosaminoglycans serum levels, N-acetylglucosamine/N-acetylglucosamine and glucosamine in sera from 20-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice with saline or 6-diazo-5-oxo-L-norleucine for 28 days. Data are mean ± standard error of the mean. One-way analysis of variance (A, F, H, I) or two-way repeated measurements analysis of variance (D). *P < .05, **P < .01, ***P < .001, for Fbn1^C1041G/+ vs Fbn1^+/+; #P < .05, ##P < .01, ####P < .0001 for Fbn1^C1041G/+ DON vs Fbn1^C1041G/+

Figure 5

Phosphoacetylglucosamine mutase 3 (PGM3) inhibition by FR054 decreases aortic dilatation and glycans in Marfan syndrome aortas. (A) Representative flow cytometry histograms and statistical analysis of O-GlcNAc and wheat germ agglutinin staining of primary vascular smooth muscle cells from Fbn1^C1041G/+ and Fbn1^+/+ mice treated with or without FR054 100 μM for 48 h. (B) Quantitative reverse transcription polymerase chain reaction analysis of Spp1 and Fgf2 mRNA relative expression in extracts from vascular smooth muscle cells of Fbn1^C1041G/+ and Fbn1^+/+ mice treated with or without FR054 100 μM for 48 h. (C) Experimental design, D–H, 18/20-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice were infused with minipumps with saline (control) or FR054 for 28 days. (D) Evolution of maximal AsAo and AbAo diameter. (E) Representative histologic staining with Elastin van Gieson and Alcian blue in the AsAo and quantification of elastin breaks. (F) Representative confocal imaging of O-GlcNac proteins or SMA, wheat germ agglutinin, elastin (autofluorescence), DAPI-stained nuclei in the ascending aorta from Fbn1^+/+ and Fbn1^C1041G/+ mice treated with/without 6-diazo-5-oxo-L-norleucine for 28 days (n = 6). (G) Glycosaminoglycans serum from Fbn1^C1041G/+ and Fbn1^+/+ mice with saline or FR054 for 28 days. (H) Quantitative reverse transcription polymerase chain reaction analysis of Col1a1 and KLF4 mRNA relative expression in aortic extracts from Fbn1^C1041G/+ and Fbn1^+/+ mice treated with or without FR054 28 days. Data are mean ± standard error of the mean. One-way analysis of variance (A, B, E, G, H) or two-way repeated measurements analysis of variance (D). *P < .05, **P < .01, ***P < .001, for Fbn1^C1041G/+ vs Fbn1^+/+; #P < .05, ##P < .01, ####P < .0001 for Fbn1^C1041G/+ DON vs Fbn1^C1041G/+

Figure 6

Integrated stress response is activated by hexosamine biosynthetic pathway in aortic tissue. (A) Heat map of integrated stress response–related genes from RNA-Sequencing analysis of aortic medial tissue from 24-week-old Fbn1^+/+ mice (n = 4) and from Fbn1^C1041G/+ mice (n = 4). (B) Representative confocal imaging p-Perk, p-eIF2α, or atf4 (red); elastin (autofluorescence), and DAPI-stained nuclei in ascending aortas from 20-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice infused with saline or 6-diazo-5-oxo-L-norleucine for 28 days (n = 5). (C) Representative immunoblot analysis and quantification of O-GlcNac proteins, p-Perk p-eIF2a, or Atf4 in aortic extracts from 20-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice infused with saline or 6-diazo-5-oxo-L-norleucine for 28 days. (D) Quantitative reverse transcription polymerase chain reaction analysis of Atf4, Atf6, and Tgfb2 mRNA relative expression in aortic extracts from 20/24-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice treated for 28 days with 6-diazo-5-oxo-L-norleucine . (E) Heat map of integrated stress response–related and extracellular matrix–related genes from RNA-sequencing analysis of aortic medial tissue from 20-week-old Fbn1^+/+ mice (n = 4) and from Fbn1^C1041G/+ mice infused with saline or 6-diazo-5-oxo-L-norleucine for 28 days (n = 4). (F) Representative confocal imaging p-Perk, p-eIF2α, or Atf4; elastin (autofluorescence), and DAPI-stained nuclei in ascending aortas from 20/22-week-old Fbn1^+/+ mice treated with or without glucosamine for 28 days (n = 6). (G) Representative confocal imaging p-Perk, p-eIF2α, or Atf4 elastin (green, autofluorescence), and DAPI-stained nuclei in ascending aortas from 20-week-old Fbn1^+/+ mice transduced with LV-Mock or LV-GPFT2 for 30 days (n = 5). Data are mean ± standard error of the mean. One-way analysis of variance *P < .05, **P < .01 or ***P < .001 for Fbn1^C1041G/+ vs Fbn1^+/+; #P < .05 or ###, ##P < .01, P < .001 for Fbn1^C1041G/+ FR054 vs Fbn1^C1039G/+

Figure 7

Integrated stress response contributes to vascular smooth muscle cells dysfunction and aortic pathology in Marfan syndrome mice. (A) Representative confocal imaging of Scarlet fluorescence in primary vascular smooth muscle cells from Fbn1^C1041G/+ and Fbn1^+/+ mice transduced with LV-ATF4Scarlet after 3 days treated with vehicle, 6-diazo-5-oxo-L-norleucine or integrated stress response inhibitor for 48 h (n = 3). (B) Representative confocal imaging of puromycinylated proteins and smooth muscle actin Sma, in primary vascular smooth muscle cells from Fbn1^C1041G/+ and Fbn1^+/+ mice incubated with puromycin for 5 min and treated with vehicle, 6-diazo-5-oxo-L-norleucine or integrated stress response inhibitor for 48 h (n = 4). (C) Experimental design, A–E, 20/22-week-old Fbn1^C1041G/+ and Fbn1^+/+ mice were infused with vehicle or integrated stress response inhibitor in minipumps for 28 days. (D) Evolution of maximal ascending (AsAo) and abdominal (AbAo) diameter. (E) Representative aortic ultrasound images after 28 days of control or integrated stress response inhibitor treatment. Discontinuous red lines mark the lumen boundary, scale bar 1 mm. (F) Representative histologic staining with Elastin van Gieson and Alcian blue in the AsAo and quantification of elastin breaks after 28 days of vehicle or integrated stress response inhibitor infusion (G). (H) Representative confocal imaging of Atf4, elastin (autofluorescence), and DAPI-stained nuclei in the ascending aorta from Fbn1^+/+ and Fbn1^C1039G/+ mice treated with/without integrated stress response inhibitor for 28 days (n = 4). (I) Representative immunoblot analysis and quantification of Atf4 in aortic extracts from Fbn1^C1041G/+ and Fbn1^+/+ mice with vehicle or integrated stress response inhibitor for 28 days. (J) Quantitative reverse transcription polymerase chain reaction analysis of Atf4, Atf6, and Tgfb2 mRNA relative expression in aortic extracts from Fbn1^C1041G/+ and Fbn1^+/+ mice infused with vehicle or integrated stress response inhibitor for 28 days. Data are mean ± standard error of the mean. Two-way repeated measurements analysis of variance (B) or one-way analysis of variance (G, J, I) or **P < .01, ***P < .001, for Fbn1^C1041G/+ vehicle vs Fbn1^+/+ vehicle; #P < .05, ##P < .01, ####P < .0001 for Fbn1^C1041G/+ ISRIB vs Fbn1^C1041G/+ vehicle

Figure 8

The HBP–ISR axis is up-regulated in Marfan syndrome patients. (A) Representative confocal immunostainings and quantification (B) of O-GlcNac (red) and wheat germ agglutinin (grey) elastin (green, autofluorescence), and DAPI-stained nuclei (blue) in medial layer of aortic sections from control donors or patients with Marfan syndrome (n = 8; n = 4 females, n = 4 males; per group). (C) Glycosaminoglycans serum and plasmatic levels, in two different cohorts of patients. (D) N-acetylglucosamine/N-acetylglucosamine and glucosamine levels in plasma from Cohort-2 of control or Marfan syndrome patients. (E) Representative confocal immunostainings and (F) quantification of GFPT2, p-PERK, p-EIF2α, ATF4, elastin (autofluorescence), and DAPI-stained nuclei in medial layer of aortic sections from control donors or patients with Marfan syndrome (n = 8; n = 4 females, n = 4 males; per group). (G) Proposed model depicting the differences between control and aortic pathology in Marfan syndrome and subsequent crosstalk of the HBP–ISR axis. Data are mean ± standard error of the mean, Student’s t-test. *P < .05, **P < .01, ***P < .01 ****P < .0001 vs control