![Steady state RNA sequencing in control and HHT BOECs. (A) Unsupervised Euclidean distance analyses for the 16 BOEC RNA sequencing data sets. When unblinded, donor replicates were seen to have clustered together. (B) Unsupervised principal component analysis of the 16 BOEC RNA sequencing data sets based on the distance matrix in which samples were projected to a 2-dimensional plane spanned by their first 2 principal components. Colored dots distinguish an experimental covariate (10 ng/mL BMP9 for 1 hour), which was discernible in control BOECs (top left) but not in the HHT clusters (ENG+/PTC and ACVRL1+/PTC/SMAD4+/PTC BOECs). (C-F) Genotype-specific data colored as for other figures (ACVRL1+/− pink/purple; ENG+/− blue; SMAD4+/− orange). (C) Normalized alignments to HHT gene Ensembl identifiers (i) ACVRL1 (ENSG00000139567), (ii) ENG (ENSG00000106991), and (iii) SMAD4 (ENSG00000141646) in control and HHT BOECs. Data from all samples per donor, mean and standard deviation shown, ∗∗P < .005 and ∗P < .05, by Dunn's test after Kruskal-Wallis test. (D) Alignments to nonsense alleles in ACVRL1, ENG, and SMAD4 nucleotides in GRCh38/hg3870 corresponding to donor heterozygous pathogenic variants. Left: wild-type allele at (i) chr12:51,916,158, (ii) chr9:127,829,769, and (iii) chr18:51065563. Right: alternate alleles at genomic position at 10× scale. Color-code key uses recommended75 nomenclature to describe both alleles per HHT donor: the relevant nonsense donor allele is indicated in bold. Mean and standard deviation displayed, pairwise P values (∗∗P < .005) calculated by Dunn's test after Kruskal-Wallis test; for overall P values P < .0001 [ACVRL1], P = .0002 [ENG], and P = .003 [SMAD4]. (E) Alignments to PTC allele in the HHT BOECs by genotype, color coded as in panels C and D. (i) Percentage of total expected if equal to wild-type allele alignment expected for heterozygotes. (ii) Percentage “loss” of total expected allele alignments. (F) Alignments to common single-nucleotide variants (SNVs) in ACVRL1 3′ UTR quantified by VarScan2.53 (i) GRCh38/hg3870 genomic positions of 6 common SNVs (V1-V6) in BOECs from donors heterozygous for ACVRL1 c.1171G>T, (p.Glu391X) or ENG c.277C>G, (p.Arg93X). Positions are illustrated by custom tracks uploaded to the UCSC Genome Browser.71 (ii) Variant: wild-type ratios in ACVRL1+/PTC and wild-type ACVRL1 (ACVRL1+/+; ENG+/PTC) BOECs. The higher ratios for V1 and V2 in ACVRL1+/PTC BOECs can be attributed to presence in cis with wild-type c.1171G, loss of the c.1171T in-frame transcripts generating the PTC, and presence of the SNV in all (V1), or only longer (V2) ACVRL1 3′ UTR transcripts (see also supplemental Figure 7).](https://ash.silverchair-cdn.com/ash/content_public/journal/blood/143/22/10.1182_blood.2023021777/2/blood_bld-2023-021777-gr2ef.jpeg?Expires=1768920091&Signature=Kf8QZhgm6QH6a0C~o0uoM9gGKhlaAIhfLVXHsnvF6roSU7q2AIWrJ4Etc4ZxeMT091Y9gVCYikDJEtEEcYPgJ4ZE5CYwyulyinCkRgEwHLs6XNbcr4EYR9Y0y3K6EoGnYkxHYeMTq5hnbc3TylzLg3cuWpl~Cc2wl8S5fnTwj4TDJjahR2w8dLS3pJs-umAJXT9H~7ELAwwSh1NZqyURTv65Q2573Ii2rLmEkykEB1UIPV8k9yBhaDyx3YY~lztpaTySBlO67DhOE4VjcrSTGv1fehcwiwzWwqzmmDGGllp0lkChaNNv5SIsHvtL9ZmECY5U514TwUgzqpyo~MCh6g__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Steady state RNA sequencing in control and HHT BOECs. (A) Unsupervised Euclidean distance analyses for the 16 BOEC RNA sequencing data sets. When unblinded, donor replicates were seen to have clustered together. (B) Unsupervised principal component analysis of the 16 BOEC RNA sequencing data sets based on the distance matrix in which samples were projected to a 2-dimensional plane spanned by their first 2 principal components. Colored dots distinguish an experimental covariate (10 ng/mL BMP9 for 1 hour), which was discernible in control BOECs (top left) but not in the HHT clusters (ENG+/PTC and ACVRL1+/PTC/SMAD4+/PTC BOECs). (C-F) Genotype-specific data colored as for other figures (ACVRL1+/− pink/purple; ENG+/− blue; SMAD4+/− orange). (C) Normalized alignments to HHT gene Ensembl identifiers (i) ACVRL1 (ENSG00000139567), (ii) ENG (ENSG00000106991), and (iii) SMAD4 (ENSG00000141646) in control and HHT BOECs. Data from all samples per donor, mean and standard deviation shown, ∗∗P < .005 and ∗P < .05, by Dunn's test after Kruskal-Wallis test. (D) Alignments to nonsense alleles in ACVRL1, ENG, and SMAD4 nucleotides in GRCh38/hg3870 corresponding to donor heterozygous pathogenic variants. Left: wild-type allele at (i) chr12:51,916,158, (ii) chr9:127,829,769, and (iii) chr18:51065563. Right: alternate alleles at genomic position at 10× scale. Color-code key uses recommended75 nomenclature to describe both alleles per HHT donor: the relevant nonsense donor allele is indicated in bold. Mean and standard deviation displayed, pairwise P values (∗∗P < .005) calculated by Dunn's test after Kruskal-Wallis test; for overall P values P < .0001 [ACVRL1], P = .0002 [ENG], and P = .003 [SMAD4]. (E) Alignments to PTC allele in the HHT BOECs by genotype, color coded as in panels C and D. (i) Percentage of total expected if equal to wild-type allele alignment expected for heterozygotes. (ii) Percentage “loss” of total expected allele alignments. (F) Alignments to common single-nucleotide variants (SNVs) in ACVRL1 3′ UTR quantified by VarScan2.53 (i) GRCh38/hg3870 genomic positions of 6 common SNVs (V1-V6) in BOECs from donors heterozygous for ACVRL1 c.1171G>T, (p.Glu391X) or ENG c.277C>G, (p.Arg93X). Positions are illustrated by custom tracks uploaded to the UCSC Genome Browser.71 (ii) Variant: wild-type ratios in ACVRL1+/PTC and wild-type ACVRL1 (ACVRL1+/+; ENG+/PTC) BOECs. The higher ratios for V1 and V2 in ACVRL1+/PTC BOECs can be attributed to presence in cis with wild-type c.1171G, loss of the c.1171T in-frame transcripts generating the PTC, and presence of the SNV in all (V1), or only longer (V2) ACVRL1 3′ UTR transcripts (see also supplemental Figure 7).
