Comparative therapeutic strategies for preventing aortic rupture in a mouse model of vascular Ehlers-Danlos syndrome

Abstract

Vascular Ehlers-Danlos syndrome is a rare inherited disorder caused by genetic variants in type III collagen. Its prognosis is especially hampered by unpredictable arterial ruptures and there is no therapeutic consensus. We created a knock-in Col3a1+/G182R mouse model and performed a complete genetic, molecular and biochemical characterization. Several therapeutic strategies were also tested. Col3a1+/G182R mice showed a spontaneous mortality caused by thoracic aortic rupture that recapitulates the vascular Ehlers-Danlos syndrome with a lower survival rate in males, thin non-inflammatory arteries and an altered arterial collagen. Transcriptomic analysis of aortas showed upregulation of genes related to inflammation and cell stress response. Compared to water, survival rate of Col3a1+/G182R mice was not affected by beta-blockers (propranolol or celiprolol). Two other vasodilating anti-hypertensive agents (hydralazine, amlodipine) gave opposite results on aortic rupture and mortality rate. There was a spectacular beneficial effect of losartan, reversed by the cessation of its administration, and a marked deleterious effect of exogenous angiotensin II. These results suggest that blockade of the renin angiotensin system should be tested as a first-line medical therapy in patients with vascular Ehlers-Danlos syndrome.

Fig 1. Survival rate and basal hemodynamic parameters of Col3a1^+/G182R mice.

A Kaplan-Meier Survival curve for comparing col3a1^+/+ (n = 62) to Col3a1^+/G182R (n = 76) and to Col3a1^G182R/G182R (n = 10), which die from vascular rupture or dissection. Significant differences are calculated using Log-Rank (Mantel-Cox) analysis. B Kaplan-Meier Survival curve for comparing Col3a1^+/G182R males (n = 41) and Col3a1^+/G182R females (n = 35). Significant difference is calculated using Log-Rank (Mantel-Cox) analysis. C Graph represents basal systolic blood pressure (SBP; lower bars: left scale) and heart rate (HR; upper lines: right scale), measured between 8 and 24 weeks in the same Col3a1^+/+ and Col3a1^+/G182R mice represented in A using a tail-cuff method. Numbers within the bars indicate the number of living mice studied at each time of measurement. SBP was comparable between groups of mice (p>0.05 at each time of measurement: 8, 12, 16 and 20, except for 24 weeks, student t-test) and HR was lower in Col3a1^+/G182R (p<0.05 at 8 and 20 weeks, student t-test). D Graph represents weight measured between 5 and 24 weeks in the same Col3a1^+/+ and Col3a1^+/G182R mice represented in A. Weight was significantly different in Col3a1^+/+ and Col3a1^+/G182R mice (p = 0.020 at 5 weeks and p<0.005 between 8 and 24 weeks, student t-test). Data are expressed as the mean ± SEM.

Fig 2. Assessment of collagen content in Col3a1^+/G182R untreated mice (Picrosirius red staining).

A Hemothorax in Col3a1^+/G182R mice, white arrow shows the massive hemorrhage in the chest. B Histological staining (Hematoxylin) of the descending thoracic aorta of Col3a1^+/G182R dead mice. White arrow shows the spontaneous aortic rupture and hemorrhage. Scale bar = 50 μm. C Localization of the aortic dissection/rupture determined through dissection in Col3a1^G182R/+ mice. D Histological staining (Picrosirius Red) of a descending thoracic aortic section in Col3a1^+/G182R mice. Scale bar = 250 μm. E details of a descending thoracic aortic section to show the two considered layers in the aortic wall (1): the intima-media (2) and the adventitia (3). The area and the collagen density in the adventitia were calculated by subtracting the intima-media pixels from the total aorta pixels. F The graph represents the area of the total aorta, the intima-media and the adventitia in Col3a1^+/+ (n = 9) and Col3a1^+/G182R (n = 7) untreated male mice. The area of descending thoracic aorta and corresponding adventitia are lower in Col3a1^+/G182R mice compared to Col3a1^+/+ mice (Student-t test, p<0.05). G The graph represents the collagen density in the total aorta, the intima-media and the adventitia in Col3a1^+/+ (n = 9) and Col3a1^+/G182R (n = 7) untreated male mice. The collagen density in the total aorta and the intima-media are lower in Col3a1^+/G182R mice compared to Col3a1^+/+ mice (Student-t test, p<0.05) whereas it remains constant in the adventitia (Student-t test, p>0.05). Data are expressed as the mean ± SEM.

Fig 3. Col3a1^+/G182R aortas have abnormal extracellular matrix architecture.

A-F electron microscopy images of the thoracic aorta in 24-week-old Col3a1^+/+ and Col3a1^+/G182R mice (n = 3 per group). N = nucleus, ECM = extracellular matrix, RER = rough endoplasmic reticulum. Scale bar = 500 nm. A-B electron microscopy images of vascular smooth muscle cell in Col3a1^+/+ and Col3a1^+/G182R mice: normal morphology of vascular smooth muscle cell in Col3a1^+/G182R mice. C-D electron microscopy images of adventitial fibroblasts in Col3a1^+/+ and Col3a1^+/G182R mice: black arrows indicate dilated RER in adventitial fibroblasts with intense ribosomal activity in Col3a1^+/G182R mice. E-F electron microscopy images of collagen cross fiber diameter in Col3a1^+/+ and Col3a1^+/G182R mice: important heterogeneity in diameter and decreased density of collagen fibers in Col3a1^+/G182R mice. Black arrows indicate heterogeneous diameter of collagen fibers.

Fig 4. Expression of Col3a1 gene and collagen III in aortas of Col3a1^+/G182R mice.

A Relative allele specific mRNA expression of type III collagen in thoracic aorta of Col3a1^+/+ (n = 21) and Col3a1^+/G182R (n = 18) 24-week-old mice of both sexes determined by ddPCR. The total expression of the Col3a1 gene was evaluated (a: unpaired student t-test). The expression of the mutated allele, harboring the Glycine substitution, was not significantly different than the expression of the WT allele in Col3a1^+/G182R mice, (b: comparative test between theoretical and observed proportions). Furthermore the absence of expression of the mutated allele in Col3a1^+/+ mice confirmed the specificity of the probe of the mutated allele. B Quantification of procollagen III, p-ERK, p-PKC, and TGFβ1R levels with western blot analysis, using total protein normalization in Col3a1^+/G182R (n = 6) and Col3a1^+/+ (n = 6) mice. C Corresponding representative Western blot analysis of proteins of TGF-β/Smad and MAPK/ERK signaling pathways: procollagen III, p-ERK, p-PKC and TGFβ1R to compare their expression in Col3a1^+/+ and Col3a1^+/G182R mice.

Fig 5. The p.G182R Col31a variation results in inflammation and cell stress response and normal PLC/IP3/PKC/ERK signaling pathway in TA transcriptome.

A Unsupervised hierarchical clustering using the 210 differentially expressed genes (DEG, adjusted p<0.05) from RNAseq in Col3a1^+/G182R (n = 3) and Col3a1^+/+ (n = 4). B Volcano plot of the complete gene dataset in descending TA. Grey represents the genes that were not modified; light blue, the genes with insignificant expression decreased more than 1.5-fold; dark blue, the DEGs with expression decreased more than 1.5-fold; light red, the genes with insignificant expression increased more than 1.5-fold, dark red, the DEGs with expression increased more than 1.5-fold; black, the 223 genes in the dataset and belonging to the MAPK signaling pathway from the KEGG database. The 5 DEG of the MAPK signaling pathway were labelled indicated their name. The sixth indicated gene Dnajb4 does not belong to this pathway. C Normalized enrichment score (NES) of the 14 enriched gene sets from the “hallmarks” database in GSEA. Positive and negative values indicate the upregulated and downregulated genes respectively in the Col3a1^+/G182R mice in comparison to Col3a1^+/+ mice (controls).

Fig 6. Therapeutic challenges: hemodynamic parameters and survival curves of β-blockers.

The upper panels represent the survival curves on either active treatment or water. The lower panels represent SBP (lower bars: left scale) and HR (upper lines: right scale), measured between 8 and 24 weeks. Data are expressed as mean ±SEM. Numbers within the bars indicate the number of living mice studied at each time of measurement. A-B Comparison between propranolol and water. A Kaplan-Meier Survival curve for comparing Col3a1^+/G182R treated with propranolol (n = 20) to Col3a1^+/G182R treated with water (n = 15). Insignificant difference is calculated using Log-Rank (Mantel-Cox) analysis (p = 0.6862). B Despite a significant decrease in HR (student t-test, p<0.05 at each time except 24 weeks), there was no decrease in SBP on propranolol that was even higher than in the untreated group (student t-test, p<0.05 at 8 and 20 weeks). C-D Comparison between celiprolol and water. C Kaplan-Meier Survival curve for comparing Col3a1^+/G182R treated with celiprolol (n = 16) to Col3a1^+/G182R treated with water (n = 19). Insignificant difference is calculated using Log-Rank (Mantel-Cox) analysis (p = 0.5411). D SBP and HR were not changed by celiprolol (student t-test, p>0.05 throughout the follow-up period).

Fig 7. Consequences of treatment by amlodipine and the association amlodipine-propranolol.

The upper panel represents the survival curves on either active treatment or water. The lower panels represent SBP (lower bars: left scale) and HR (upper lines: right scale), measured between 8 and 24 weeks. Data are expressed as mean ±SEM. Numbers within the bars indicate the number of living mice studied at each time of measurement. A Survival comparison between Amlodipine and ethanol 1%, and between the association amlodipine-propranolol and amlodipine (monotherapy). Kaplan-Meier Survival curve (red curve) for comparing Col3a1^+/G182R treated with amlodipine (n = 20) to Col3a1^+/G182R treated with ethanol 1% (n = 23). Insignificant difference is calculated using Log-Rank (Mantel-Cox) analysis (p = 0.1261). Kaplan-Meier Survival curve (brown curve) for comparing Col3a1^+/G182R treated with amlodipine-propranolol (n = 10) to Col3a1^+/G182R treated with amlodipine (n = 20). Insignificant difference is calculated using Log-Rank (Mantel-Cox) analysis (p = 0.3402). B SBP and HR comparison between amlodipine and ethanol 1%, and between the association amlodipine-propranolol and amlodipine (monotherapy). SBP was slightly lower in the treated group (student t-test, p>0.05 throughout the follow-up period except at 20 weeks) and HR was slightly increased (student t-test, p<0.05 at 16 and 20 weeks) when comparing amlodipine to ethanol 1% (red bars and curve). SBP remained unchanged throughout the follow-up period (Student-t test, except at 8 weeks, p = 0.0086) and HR decreased significantly (student t-test, p<0.05 throughout the follow-up period, except at 12 weeks) when comparing amlodipine-propranolol to amlodipine (monotherapy, brown bars and curve).

Fig 8. Consequences of treatment by hydralazine and the association hydralazine-celiprolol.

The upper panels represent the survival curves on either active treatment or water. The lower panels represent SBP (lower bars: left scale) and HR (upper lines: right scale), measured between 8 and 24 weeks. Data are expressed as mean ±SEM. Numbers within the bars indicate the number of living mice studied at each time of measurement. A Survival comparison between hydralazine and water, and between the association hydralazine-celiprolol and water. Kaplan-Meier Survival curve (orange curve) for comparing Col3a1^+/G182R treated with hydralazine (n = 17) to Col3a1^+/G182R treated with water (n = 36). Insignificant difference is calculated using Log-Rank (Mantel-Cox) analysis (p = 0.1358). Kaplan-Meier Survival curve (green curve) for comparing Col3a1^+/G182R treated with hydralazine-celiprolol (n = 16) to Col3a1^+/G182R treated with water (n = 36). Insignificant difference is calculated using Log-Rank (Mantel-Cox) analysis (p = 0.8477). B SBP and HR comparison between hydralazine and water, and between the association hydralazine-celiprolol and water. No significant decrease of SBP was observed and significant decrease of HR (student t-test, p<0.05 from 12 weeks) was observed between hydralazine treated and water groups. SBP and HR were not changed by hydralazine-celiprolol (student t-test, p>0.05 throughout the follow-up period).

Fig 9. Consequences of on- and off-treatment by losartan.

The upper panels represent the survival curves on either active treatment or water. The lower panels represent SBP (lower bars: left scale) and HR (upper lines: right scale), measured between 8 and 24 weeks. Data are expressed as mean ±SEM. Numbers within the bars indicate the number of living mice studied at each time of measurement. A Comparison between losartan started in pregnant mothers and water. Kaplan-Meier Survival curves for comparing Col3a1^+/G182R mice (n = 20) with a 24-week period of losartan followed by a 24-week washout period and Col3a1^+/G182R (n = 8) with no treatment. Significant difference in mortality in treated mice (18% mortality at 24 weeks) compared to the untreated group (63% mortality at 24 weeks) is calculated using Log-Rank (Mantel-Cox) analysis (p = 0.0207) at the end of the 24-week treatment. When the administration of losartan was stopped, an important increased mortality was observed and survival rates at 48 weeks of age were comparable between the 2 groups of mice (41% vs 38%, p = 0.4150). B-C Comparison between losartan started at weaning and water. B SBP was significantly changed by losartan (student t-test, p<0.05 at 24 weeks of the follow-up period) and HR was not changed by losartan (student t-test, p>0.05 throughout the follow-up period). C Kaplan-Meier Survival curve for comparing Col3a1^+/G182R treated with Losartan started at age 4 weeks (n = 13) to Col3a1^+/G182R treated with water (n = 8). Losartan significantly improve the survival using Log-Rank (Mantel-Cox) analysis (p = 0.0031). D Kaplan-Meier Survival curve for comparing Col3a1^+/+ (n = 9, doted curve) to Col3a1^+/G182R (n = solid curve) male mice with Angiotensin II infusion (0.5 μg/kg/min). Significant difference is calculated using Log-Rank (Mantel-Cox) analysis (p<0.0001).