uAUG creating variants in the 5’UTR of ENG causing Hereditary Hemorrhagic Telangiectasia

Hereditary Hemorrhagic Telangiectasia (HHT) is a rare, autosomal dominant, vascular disorder. About 80% of cases are caused by pathogenic variants in ACVRL1 (also known as ALK1) and ENG, with the remaining cases being unexplained. We identified two variants, c.-79C>T and c.-68G>A, in the 5’UTR of ENG in two unrelated patients. They create upstream AUGs at the origin of upstream overlapping open reading frames (uoORFs) ending at the same stop codon. To assess the pathogenicity of these variants, we performed functional assays based on the expression of wild-type and mutant constructs in human cells and evaluated their effect on ALK1 activity in a BMP-response element assay. This assay is mandatory for molecular diagnosis and has been so far only applied to coding ENG variants. These variants were associated with a decrease of protein levels in HeLa and HUVEC cells and a decreased ability to activate ALK1. We applied the same experiments on three additional uoORF-creating variants (c.-142A>T, c.-127C>T and c.-10C>T) located in the 5’UTR of ENG and previously reported in HHT patients. We found that all the analyzed variants alter protein levels and function. Additional experiments relying on an artificial deletion in our mutated constructs show that identified uAUGs could initiate the translation indicating that the associated effect is translation-dependent. Overall, we have identified two 5’UTR ENG variations in HHT patients and shed new light on the role of upstream ORFs on ENG regulation. Our findings contribute to the amelioration of molecular diagnosis in HHT.

To assess the pathogenicity of these variants, we performed functional assays based on the expression of wild-type and mutant constructs in human cells and evaluated their effect on ALK1 activity in a BMP-response element assay.This assay is mandatory for molecular diagnosis and has been so far only applied to coding ENG variants.These variants were associated with a decrease of protein levels in HeLa and HUVEC cells and a decreased ability to activate ALK1.We applied the same experiments on three additional uoORFcreating variants (c.-142A>T, c.-127C>T and c.-10C>T) located in the 5'UTR of ENG and previously reported in HHT patients.We found that all the analyzed variants alter protein levels and function.Additional experiments relying on an artificial deletion in our mutated constructs show that identified uAUGs could initiate the translation indicating that the associated effect is translationdependent.
Overall, we have identified two 5'UTR ENG variations in HHT patients and shed new light on the role of upstream ORFs on ENG regulation.Our findings contribute to the amelioration of molecular diagnosis in HHT.Supplemental Table 3

figure 1 .
Quantification of ENG levels in HeLa (a-b) and HUVEC cells (c-d).(a) Apart for c.-9G>A variant, ENG steady-state levels in HeLa cells is significantly decreased with variants in comparison to the wild-type (WT) construct.For quantification, the average of each duplicate has been calculated from the quantified values and ENG levels for each sample have been normalized to the corresponding β-actin levels then to the WT (%).The two bands obtained for the Endoglin, corresponding to the more glycosylated (upper band) and less/non glycosylated (lower band) ENG monomers 1 , were taken together for the quantification.(b) RNA levels of Endoglin do not variate between wild-type and variants.Normalized 2-ΔΔCT to the WT are shown.(c) Decrease of ENG protein levels in HUVECs cells in comparison to the wild-type (WT).ENG levels for each sample have been normalized to the corresponding β-actin levels then to the WT (%).C-, Negative control corresponding to the empty vector.(d) ENG levels associated to variants are similar or slightly higher of endogenous endoglin (C-on the Figure).ENG levels for each sample have been normalized to the corresponding β-actin levels then to the negative control (C-) (%).Graphs with standard error of the mean are representative of 5 independent experiments.***, p-value < 10 -3 , *, p < 5.10 -2 , ns, nonsignificant (two-factor ANOVA followed by Tukey's multiple comparison test of variants versus WT (a, b, c) or negative control (d)).Supplemental figure 2. Western blot results on total proteins extracted from transfected HeLa cells with pcDNA3.1-L-ENGconstructs (a, b and c) or from transduced HUVEC cells with lentiviruses containing ENG (d).The blots shown in a, b and c correspond to the uncropped upper, middle and lower blots in main Figure1c, respectively and the blot shown in d corresponds to the uncropped blot in Figure1d.Each blot was cut in 2 parts and each part was incubated with the corresponding antibody: monoclonal anti-(c-Myc Tag) (HeLa extracts), anti-ENG (HUVECs extracts) or anti-β-actin (HeLa and HUVECs extracts) as a loading control.Specific fluorescent secondary antibodies were used.Finally, Odyssey Infrared Imaging System (Li-Cor Biosciences) in 700 and 750 channels was used to scan, reveal, and quantify the blots.HeLa cells extracts were analyzed in duplicates.For HeLa cells, variants were analyzed in duplicates.M, protein ladder; WT, wild-type, C-, negative control corresponding to pcDNA3.1-emptyvector.

Supplemental figure 3 .Supplemental figure 4 .Supplemental figure 5 .
Optimization of cellular-based BRE activity assay to be applicable on ENG 5'UTR variants.(a) pcDNA3.1-L-ENG-WTvector gives similar BRE activity to ENG-WT plasmid only containing the CDS of ENG (WT(CDS)) used in Mallet et al., 2015 1 .C-corresponds to pcDNA3.1 empty vector used as negative control.WT, wild-type; 5'UTR, 5' UnTranslated Region; CDS, CoDing Sequence.The graph with standard error of the mean is representative of 4 independent experiments.ns, non-significant (two-factor ANOVA followed by Tukey's multiple comparison test of WT(5'UTR+CDS) versus WT(CDS)).(b) Titration of ENG vectors with increasing quantities in order to determine optimal conditions to study 5'UTR variants.Fold induction corresponds to the detected activity normalized to control without ENG (C-).Transfected quantities of pcDNA3.1-L-ENG-WTor variant plasmids, and empty vector (C-) in 96 well plates are indicated.Graphs with standard error of the mean is representative of 3 independent experiments.ng, nanograms; ns, non-significant; *, p < 5.10 -2 (two-factor ANOVA followed by Tukey's multiple comparison test of c.-142A>T variant versus WT).Western blot results on total proteins extracted from transfected HeLa cells with pcDNA3.1-L-ENG-c.1_2delconstructs.The blots shown in a and b correspond to the uncropped right and left blots in main Figure 3b, respectively.Each blot was cut in 2 parts and each part was incubated with the corresponding antibody: monoclonal anti-(c-Myc Tag) or anti-β-actin as a loading control.Specific fluorescent secondary antibodies were used.Finally, Odyssey Infrared Imaging System (Li-Cor Biosciences) in 700 and 750 channels was used to scan, reveal, and quantify the blots.HeLa cells extracts were analyzed in duplicates.Each variant was analyzed in duplicate.M, protein ladder; WT, wild-type, C-, negative control corresponding to pcDNA3.1-emptyvector.Created uAUGs in the 5'UTR of ENG seem to be able to initiate the translation.(a) Western blot results on total proteins extracted from transfected HeLa cells with 1 µg of pcDNA3.1-L-ENGconstructs.Two bands of different molecular weights are observed for the Endoglin likely corresponding to more glycosylated (upper band) and less/non glycosylated (lower band) ENG monomers 1 .Anti-Myc and anti-βactin correspond to the used antibodies for the target and the reference proteins, respectively.kDa, kilodalton; M, protein ladder; WT, wild-type, C-, negative control corresponding to pcDNA3.1-emptyvector.Uncropped blots are shown in Supp.figure 06.All blots were processed in parallel and derive from the same experiments.(b) Quantification of protein steady-state levels obtained in (a) and probably resulting from translation initiation at the created uAUG in presence of the c.-10C>T variant.For quantification, the average of each duplicate has been calculated from the quantified values and ENG levels have been normalized to the corresponding β-actin levels then to the WT (%).Graphs with standard error of the mean are representative of 5 independent experiments.*, p-value < 5.10 -2 , ns, non-significant (two-factor ANOVA followed by Tukey's multiple comparison test of variants versus WT).(c) Quantification of ENG RNA levels in HeLa cells transfected with pcDNA3.1-L-ENGconstructs containing the c.1_2del deletion in presence or in absence (na) of uAUG-creating variants.Normalized 2-ΔΔCT values to the WT are shown.Graphs with standard error of the mean are representative of 5 independent experiments.*, p-value < 5.10 -2 , ns, non-significant (two-factor ANOVA followed by Tukey's multiple comparison test of variants versus na corresponding to the construct only containing the c.1_2del).Supplemental figure 6.Western blot results on total proteins extracted from transfected HeLa cells with pcDNA3.1-L-ENG-c.1_2delconstructs.The shown blot corresponds to the uncropped blot in Supp.figure 04a.The blot was cut in 2 parts and each part was incubated with the corresponding antibody: monoclonal anti-(c-Myc Tag) or anti-β-actin as a loading control.Specific fluorescent secondary antibodies were used.Finally, Odyssey Infrared Imaging System (Li-Cor Biosciences) in 700 and 750 channels was used to scan, reveal, and quantify the blots.HeLa cells extracts were analyzed in duplicates.Each variant was analyzed in duplicate.M, protein ladder; WT, wild-type, C-, negative control corresponding to pcDNA3.1-emptyvector.

. Gene content of the used custom Next Generation Sequencing panel.
Specific HHT genes are bolded.