Thresholds of Endoglin Expression in Endothelial Cells Explains Vascular Etiology in Hereditary Hemorrhagic Telangiectasia Type 1

Abstract

Hereditary Hemorrhagic Telangiectasia type 1 (HHT1) is an autosomal dominant inherited disease characterized by arteriovenous malformations and hemorrhage. HHT1 is caused by mutations in ENDOGLIN, which encodes an ancillary receptor for Transforming Growth Factor-β/Bone Morphogenetic Protein-9 expressed in all vascular endothelial cells. Haploinsufficiency is widely accepted as the underlying mechanism for HHT1. However, it remains intriguing that only some, but not all, vascular beds are affected, as these causal gene mutations are present in vasculature throughout the body. Here, we have examined the endoglin expression levels in the blood vessels of multiple organs in mice and in humans. We found a positive correlation between low basal levels of endoglin and the general prevalence of clinical manifestations in selected organs. Endoglin was found to be particularly low in the skin, the earliest site of vascular lesions in HHT1, and even undetectable in the arteries and capillaries of heterozygous endoglin mice. Endoglin levels did not appear to be associated with organ-specific vascular functions. Instead, our data revealed a critical endoglin threshold compatible with the haploinsufficiency model, below which endothelial cells independent of their tissue of origin exhibited abnormal responses to Vascular Endothelial Growth Factor. Our results support the development of drugs promoting endoglin expression as potentially protective.

Keywords: cell signaling; endoglin; endothelial cells; hereditary hemorrhagic telangiectasia.

Figure 1

Endoglin mRNA expression levels in multiple mouse organs. (A) Ct values for all candidate genes in skin, lung, intestine, brain, heart, kidney and liver isolated from 9 weeks old 129/Svj (n = 6) and C57Bl/6J (n = 7) mice. (B) Average expression stability values (M) of the candidate reference. M is represented from the least stable (left) to the most stable (right), analyzed by GeNorm software. (C) Pairwise variation analysis between normalization factors to determine the optimal number of control genes for normalization. (D,E) Real time PCR for endoglin normalized to a reference pool of three endothelial reference genes (Tie1, Tie2 and Icam2) determined by GeNorm comparing endoglin levels in multiple tissues isolated from 9 weeks old 129/Svj (n = 6) and C57Bl/6J (n = 7) mice. (D) Non significant (ns) results from Mann–Whitney U test that compares the median of two groups. (E) Error bars represent SEM. * p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001 result from one-way ANOVA and Tukey’s multiple comparison tests comparing the mean of each group with the mean of the other groups. ns: Not significant.

Figure 2

Marked heterogeneity in endoglin levels across different mouse tissues. (A,B) Sections of multiple organs isolated from 9-week-old C57Bl/6J mice stained for Pecam1 (A) or endoglin (B) (scale bar: 50 μm). (C) Fluorescence intensity was measured over the full range of PMT voltages for liver and the skin sections to determine the linear fluorescence intensities for each tissue. We identified the PMT gain of 850 as optimal to ensure an accurate quantification of endoglin levels across the various tissues. (D) Endoglin levels were quantified using ImageJ software defined as the mean grey intensity of the blood capillaries selected in each image. A minimum of 30 blood vessels was quantified per organ isolated from 3 wild-type mice. Error bars represent SEM. * p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001 result from 1-way ANOVA and Tukey’s multiple comparison tests comparing the mean of each group with the mean of the other groups. ns: Not significant.

Figure 3

Heterogeneity in endoglin expression along the arteriovenous axis. (A) Fluorescence intensity was measured over the full range of PMT voltages for liver, lung and skin sections from 9-week-old C57Bl/6J wild type (n = 3) and Eng^+/− mice (n = 3) and confirmed that endoglin levels were reduced by half in all organs. (B) Upper panels. Confocal images of whole mount ears isolated from 8-week-old wild-type and Eng^+/− mice and stained for endoglin or PECAM1 and α-SMA (Scale bar: 500 μm). Lower panels are higher magnifications showing that endoglin is predominantly expressed in veins and becomes almost undetectable in arteries and capillaries in Eng^+/− mice (Scale bar: 100 μm). Error bars represent SD. * p < 0.05, ** p < 0.01 and **** p < 0.0001 result from 2-way ANOVA and Šidàk’s multiple comparisons test.

Figure 4

Endoglin expression levels in human tissues. (A,B) Sections of multiple tissues isolated from a 57 year-old patient with HHT1 stained for PECAM1 (A) or endoglin (B) (Scale bar: 50 μm). Pecam1 and endoglin levels were quantified using ImageJ software defined as the mean grey intensity of the blood capillaries selected in each image. Blood vessels were quantified per organ isolated from at least three independent tissue sections. Error bars represent SEM. ** p < 0.01, *** p < 0.001 and **** p < 0.0001 result from 1-way ANOVA and Tukey’s multiple comparison tests comparing the mean of each group with the mean of the other groups. ns: Not significant.

Figure 5

Different sensitivity of organ-specific ECs to reduced endoglin levels. (A) Schematic illustration of the strategy used to study how endoglin levels influence EC signaling pathways. It includes the microbeads-based protocol for the isolation of ECs [46] and the Eng gene deletion strategy. (B) Confocal images of cultured ECs isolated from lungs and liver and stained for PECAM1, VE-Cadherin and Endoglin (Scale bar: 10 μm). (C) Confocal images of lung ECs stained for endoglin at different time points after Eng gene deletion and fluorescence intensity quantification of its expression at the endothelial cell surface (a minimum of 10 cells were quantified per time point) (Scale bar: 10 μm). (D) Confocal images of lung and liver ECs stained for PECAM1, VE-Cadherin and Endoglin at 0, 20 and 60 h after viral infection, showing the efficacy of the Eng gene deletion (Scale bar: 10 μm) and fluorescence intensity quantification of endoglin expression at the endothelial cell surface (a minimum of 35 cells were quantified per time point). (E) ECs isolated from lungs and liver of P7 pups were exposed to VEGF (25 ng.ml⁻¹), TGF-β1 (1 ng.ml⁻¹) and BMP9 (1 ng.ml⁻¹) for 30 min at 37 °C before lysis. Whole cell extracts were fractionated by SDS-page and blotted. The filters were incubated with Phospho-Akt, Phospho-Smad1, Akt, Smad1, endoglin and β-actin. Representative results from at least 4 independent experiments are shown. Graphs represent quantification of the Western blotting. Error bars represent SEM. * p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001 result from 1-way ANOVA and Dunnett’s post hoc tests comparing the mean of each group to the unstimulated condition. ns: Not significant. (F) Western blot analysis of Akt, p42/p44 MAPK and Smad1 phosphorylation at different time points after Eng gene deletion.

Figure 6

Schematic representation of how heterogeneity in endoglin levels in ECs predict the tissue-specific manifestations of HHT1.