https://orcid.org/0000-0002-9885-9686
Abstract
Pathogenic variants in canonical transforming growth factor β (TGFβ) signaling genes predispose patients to thoracic aortic aneurysm and dissection (TAAD), predominantly in aortic root. Although TAAD pathogenesis associated with TGFβ receptor defects is well characterized, distinct and redundant mechanisms of TGFβ isoforms in TAAD incidence and severity remain elusive.
Here we examined the biological role of TGFB2 in smooth muscle cell (SMC) differentiation and investigated how TGFB2 defects can lead to regional TAAD manifestations.
To characterize the role of TGFB2 in SMC differentiation and function, we employed human-induced pluripotent stem cell (hiPSC)-derived SMC differentiation, CRISPR/Cas9 gene editing, three-dimensional SMC constructs, and human aortic tissue samples.
Despite the similar effects of different TGFβ isoforms on hiPSC-derived SMC differentiation, siRNA experiments revealed that TGFB2 distinctively displays TGFBR3 dependence for signal transduction, an understudied TGFβ receptor in TAAD. Molecular evaluation of different thoracic aorta regions suggested TGFB2 and TGFBR3 enrichment in the aortic root tunica media. TGFB2 haploinsufficiency (TGFB2KO/+) and TGFB2 neutralization impaired the differentiation of second heart field-derived SMCs. TGFBR3KO/KO prevented the molecular rescue of TGFB2KO/+ by TGFB2 supplementation indicating the involvement of TGFBR3 in TGFB2-mediated SMC differentiation. Lastly, a missense TGFB2 variant (TGFB2G276R/+) caused mechanical defects in SMC tissue ring constructs that were rescued by TGFB2 supplementation or genetic correction.
Our data suggests the distinct regulation and action of TGFB2 in SMCs populating the aortic root, while redundant activities of TGFβ isoforms provide implications about the milder TAAD aggressiveness of pathogenic TGFB2 variants.
Genetic defects in transforming growth factor β (TGFβ) signaling genes predispose patients to thoracic aortic aneurysm and dissection (TAAD), predominantly in aortic root. Although canonical TGFβ receptors have been extensively studied in the TAAD field due to their central role in TGFβ signaling, the molecular link between different TGFβ isoforms and TAAD is not fully understood. Using hiPSC modeling, our data suggest TGFBR3-dependent mechanism of TGFB2 in smooth muscle cell differentiation implicating the nonredundant role of TGFB2 in aortic SMC physiology while compensatory activities of TGFβ isoforms provide implications about the milder TAAD aggressiveness associated with their pathogenic variants.
Introduction
Genetic predisposition is a risk factor for thoracic aortic aneurysm and dissection (TAAD), which causes significant morbidity and mortality.1-3 Inactivating pathogenic variants in canonical transforming growth factor β (TGFβ) signaling genes (TGFBR1, TGFBR2, SMAD3, SMAD2, TGFB2, and TGFB3) predispose patients to TAAD in the aortic root, also known as Loeys-Dietz Syndrome (LDS).2,4-8 In fact, TGFβ signaling dysregulation is observed in other syndromic TAAD forms including Marfan Syndrome.9,10 TGFβ signaling is essential for vascular smooth muscle cell (SMC) differentiation and aortic wall integrity.6,7,11-14 An LDS mouse model and our human induced pluripotent stem cell (hiPSC) disease modeling studies demonstrated that mutations in TGFBR1 disrupt SMCs derived from second heart field cardiovascular progenitor cells (CPC-SMCs) in a lineage-specific manner, sparing neural crest stem cell lineage SMCs (NCSC-SMCs).15,16 Likewise, Tgfbr2 deletion in postnatal SMCs causes aneurysms both in aortic root and ascending aorta in mouse models.17 As aortic root is mostly populated with CPC-SMCs, these findings informed the etiology of aortic root aneurysms in LDS patients.
TGFβ signaling is initiated when a TGFβ ligand engages type II and type I TGFβ receptors (TGFBR2 and TGFBR1). In the canonical signaling, the TGFβ receptor complex phosphorylates the regulatory SMADs; SMAD3 (Mothers against decapentaplegic homolog 3) and SMAD2 (Mothers against decapentaplegic homolog 2), to alter gene transcription. Non-canonical TGFβ signaling refers to the activation of other downstream effectors such as ERK (extracellular signal-regulated kinase), p38 mitogen-activated protein kinase, and AKT (Protein kinase B) by the receptor complex.18 There are 3 TGFβ isoforms (TGFB1, TGFB2, and TGFB3) regulating TGFβ receptor signaling and their knockout mice phenotypes show both common and non-overlapping features.19Tgfb2 knockout mice display perinatal mortality with congenital heart, lung, and outflow tract defects, skeletal malformations, and cleft palates while Tgfb1 null mice have widespread immune system defects with an inflammatory phenotype.20-23 Similar to Tgfb2 knockout mice, Tgfb3 null mice also have cleft palates and lung defects; however, aortic defects observed in Tgfb2 knockout mice are minimal.24-28Tgfb2 and Tgfb3 expression patterns in the developing cardiovascular system show significant overlap with enrichment in arterial tunica media and adventitia, while Tgfb1 expression appears more restricted to intima.29 These observations suggest nonredundant activity of TGFB2 in cardiovascular development that invites further exploration within TAAD context.
Despite the association of several TGFβ signaling genes in human TAAD, there are also differences between their phenotypic manifestations. The pathogenic variants in TGFBR1, TGFBR2, and SMAD3 have similar aortic phenotypes with early disease onsets. Pathogenic TGFB2 or TGFB3 variants are often less penetrant in families and have milder aortic phenotypes,6,7,30,31 while TGFB1 variants are not strongly associated with TAAD formation.32 In addition, TGFB2 variants can cause Kawasaki disease often characterized with coronary artery aneurysm, ischemic heart disease, and sudden cardiac death implying pervasive effects of TGFB2 in the human cardiovascular system.33 Here we investigated the distinct mode of action of TGFB2 on SMC differentiation using both monolayer and 3D tissue ring cultures. Through the utilization of hiPSC-derived SMCs, SMC tissue constructs, human aorta samples, and CRISPR/Cas9 gene editing, our data have implications about the unique contribution of TGFB2 defects to aortic root aneurysm formation.
Materials and methods
Additional materials and methods, and the detailed list of reagents, primers, and primary and secondary antibodies are available in Tables S1 and S2.
Study approval
All experiments were performed according to the protocols approved by the Institutional Review Board at the University of Michigan (HUM00054585 and HUM00052866).
Human iPSCs generation and CRISPR/Cas9 gene editing
Human induced pluripotent stem cells were generated as described previously.16,34 For CRISPR/Cas9 gene editing, single guide RNA (sgRNA, IDT) with the sequence 5’-CCACTAGGAAAAAAAACAGT-3’ was used to target the upstream of TGFB2 c.826G. To form the ribonucleoprotein (RNP) complex, 10 μg S.p. HiFi Cas9 Nuclease V3 (IDT) and 7.5 μg of sgRNA were first incubated for 10 minutes at room temperature. Then, 10 μg of designed single-strand oligo donor DNAs (ssODNs, IDT) were added to the mixture. About 2 × 106 hiPSCs were electroporated with the RNP mixture using Celetrix electroporation system (Celetrix, Manassas, VA, USA) with the program: V set (630V); T set (30 ms); P num (1N); P int (1 ms). Transfected hiPSCs were seeded on Corning GFR Matrigel-coated 35mm dishes in TesRE8 medium (Catalog No. 05990, STEMCELL Technologies) and cultured until the formation of visible clones. A heterozygous knockout clone with a 7 base-pair deletion causing a premature stop codon in TGFB2 gene was selected for further experiments. Two ssODNs were designed to induce the TGFB2 c.826G>A (p. Gly276Arg) mutation in healthy male hiPSCs or to correct TGFB2 c.826G>A (p. Gly276Arg) variant in hiPSCs derived from a male patient. A heterozygous clone with TGFB2 c.826G>A mutation, and 2 independent patient correction clones were selected for subsequent experiments. The TGFB2 heterozygous knockout clone was also used to perform TGFBR3 gene editing. SgRNA with the sequence 5’-ACACTATTCCTCCTGAGCTA-3’ was used to target TGFBR3 gene. A homozygous clone with 14 base-pair deletion and a homozygous clone with 1 base-pair insertion, which caused premature stop codons in TGFBR3 gene, were selected for further experiments. The Sanger sequencing primers used in the study are listed in Table S1.
Generation of SMCs from human iPSCs
Differentiation of hiPSC to CPC-SMCs and NCSC-SMCs was performed as described previously.16,34-38 For CPC-SMC differentiation, hiPSCs were seeded on Matrigel-coated plates at a density of 3 × 104 cells/cm2 in TesRE8 medium with 10 μmol/L Y27632 (Biogems), and then incubated in CPC differentiation medium (DMEM/F12 (Gibco), B27 (without vitamin A, Gibco), 25 ng/mL BMP4 (PeproTech), 8 μmol/L CHIR99021 (Biogems), 50 μg/mL ascorbic acid (Sigma), 400 μmol/L 1-thioglycerol (Sigma), and 1% penicillin-streptomycin (Gibco)) for 3 days. The resulting CPCs were seeded at a density of 1.5 × 104 cells/cm2 and incubated for 7 days in CPC-SMC basic medium (DMEM/F12, B27, 1% penicillin-streptomycin, 400 μmol/L 1-thioglycerol, 10 ng/mL PDGFBB (PeproTech)) with either vehicle (0.1% BSA with no TGFβ ligand), 2 ng/mL active TGFB1 (PeproTech), 2 ng/mL active TGFB2 (PeproTech) or 2 ng/mL full-length TGFB2 (Acro). For NCSC-SMC differentiation, hiPSCs were seeded on Matrigel-coated dishes at a density of 2 × 104 cells/cm2 in TesRE8 medium with 10 μmol/L Y27632. After reaching nearly 50% confluency, they were incubated in NCSC differentiation medium (DMEM/F12, 1 × N2 supplement [Gibco], 0.1% BSA [Sigma], 1% penicillin-streptomycin plus 10 μmol/L SB4315421 [Biogems], and 1 μmol/L LDN193189 [Biogems]) for 6 days. From day 2 to day 6, 3 μmol/L CHIR99021 was added to the medium. The resulting NCSCs were seeded at a density of 8 × 104 cells/cm2 in NCSC medium with 10 μmol/L Y27632. One day later, the cells were differentiated using NCSC-SMC basic medium (DMEM/F12, 20% knockout serum replacement (Gibco), 1% penicillin-streptomycin) with either vehicle (0.1%BSA with no TGFβ ligand), 2 ng/mL active TGFB1 or 2 ng/mL active TGFB2 for 8 days.
Tissue ring assays and analysis
Tissue rings with a diameter of 2 mm were prepared as described previously.16,34,39 Briefly, on day 5 of the CPC-SMC differentiation, SMCs were harvested using Accutase (Gibco) and seeded on Matrigel-coated 100 mm dishes with a density of 2.5 × 104 cells/cm2. The cells were cultured in DMEM/F12 with 10% FBS. The cells were harvested when they reached 80%-90% confluence (3-4 days) and seeded on agarose molds at a density of 1 × 106 cells/well in DMEM/F12 with 10% FBS. The tissue ring culture medium (DMEM (high glucose), 20% FBS, 1% penicillin-streptomycin, 1 ng/mL TGFB1, and supplemented with proline (50 μg/mL), glycine (50 μg/mL), alanine (20 μg/mL), CuSO4 (3 ng/mL), and ascorbic acid (50 μg/mL) was applied on the second day of the tissue ring culture. Medium was replaced every 2 days. The rings were harvested on day 14 for subsequent analysis. For uniaxial tensile tests, both sides of each ring were measured to calculate its thickness. The length and width of the rings were also measured and recorded. Mechanical tests were performed using a uniaxial test machine (TA Instruments RSA-G2). The tissue rings were stretched with custom rectangular loops threaded through each ring and then clamped into standard thin film fixtures (TA Instruments). The samples were then pulled to failure at a speed of 10 mm/minute. For each ring, the force (N) and strain (%) were recorded throughout the test. Two mechanical parameters were calculated from the acquired data using MATLAB software: ultimate tensile strain (maximal tension divided by a cross-sectional area of tissue during the recording [MPa = N/mm2]), ultimate strain (elongated length at the maximal tension divided by the initial length [%]). Active human TGFB2 protein levels were assessed in tissue ring lysates using a TGFB2 enzyme-linked immunosorbent assay (ELISA) kit (Catalog No. DB250, R&D Systems). Active TGFB2 levels were measured without sample acidification as described previously.40,41 Active TGFB2 concentration in each tissue ring was normalized to total protein concentration. The immunofluorescence stainings were performed on CPC-SMC tissue rings generated in independent differentiation batches. Paraffin-embedded tissue rings were sectioned into 5 μm slices. After deparaffinization, rehydration, permeabilization, and heat antigen retrieval, the sections were blocked with 5% BSA for 1 hour at room temperature and then incubated with Myosin heavy chain 11 (Catalog No. ab133567, 1:100, Abcam) primary antibody in 1% BSA overnight at 4°C. Alexa Fluor 594 Donkey Anti-Rabbit IgG (Catalog No. 711585152, 1:500, Jackson Immuno Research) was used as the secondary antibody. The sections were counterstained with DAPI. Images were taken using BZ-X800 Keyence microscope. MYH11 staining intensity was quantified using Image J.42 The cell density of each ring was determined by DAPI staining and quantified using Image J.
Statistics
All quantitative data were presented as mean ± standard deviation with at least 3 biological replicates. We conducted a Shapiro-Wilk normality test prior to all analysis. When analyzing >2 groups that were not normally distributed or when analyzing >2 groups with <6 biological replicates, we performed Kruskal-Wallis test with Dunn’s multiple comparisons test. When analyzing >2 groups that were normally distributed (n = 6), we tested for equal variance using the Brown-Forsythe test. If the standard deviations were not significantly different, we performed 1-way analysis of variance (ANOVA) analysis with Dunnett’s multiple comparisons test to compare the mean of each column with the mean of a control column. If the standard deviations were significantly different, we performed 1-way ANOVA analysis with Dunnett’s T3 multiple comparisons test to compare the mean of each column with the mean of a control column. When analyzing only 2 datasets that were not normally distributed (n ≥ 3), we performed Mann-Whitney U test. When analyzing only 2 datasets that were normally distributed (n ≥ 3), we used the unpaired t-test. If the variances were significantly different, we conducted the unpaired t-test with Welch’s correction. The statistical analyses were performed using GraphPad Prism Software.
Results
TGFB1 and TGFB2 have similar effects on hiPSC-derived SMC differentiation
Human aortic root is populated with SMCs derived from second heart field cardiovascular progenitor cells (CPC-SMCs) and TGFB2 defects predispose patients to TAAs predominantly in the aortic root.43-45 Among TGFβ isoforms, TGFB1 is typically used in monolayer SMC differentiation protocols. To compare the activity of TGFB1 and TGFB2, we treated CPC progenitor cells either with vehicle (no TGFβ ligand), 2 ng/mL biologically active recombinant TGFB1, or with 2 ng/mL biologically active recombinant TGFB2 during CPC-SMC differentiation. RT-qPCRs revealed a robust and comparable increase in SMC markers; ACTA2 (actin alpha 2), SM22α (smooth muscle protein 22-alpha), CNN1 (calponin 1), MYOCD (myocardin), SMTN (smoothelin), and MYH11 (myosin heavy chain 11) in TGFB1 and TGFB2 treated CPC-SMCs compared to the vehicle (Figure 1A). We also measured the effects of full-length TGFB2 (FL-TGFB2) on CPC-SMC differentiation as a previous report suggested a lower activation threshold for FL-TGFB2.46 However, FL-TGFB2 treatment failed to induce adequate expression of SMC markers in our culture conditions indicating the requirement of active TGFB2 for hiPSC-derived SMC marker expression (Figure 1B).
TGFB1 and TGFB2 have similar effects on hiPSC-derived SMC differentiation. (A) Relative expression of SMC markers in CPC-SMCs after differentiation with either 2 ng/mL active TGFB1, 2 ng/mL active TGFB2, or control (vehicle/no ligand). The average expression in the TGFB1-treated samples was set to 1. n = 6 biological replicates. (B) Relative expression of smooth muscle cell (SMC) markers in cardiovascular progenitor cells (CPC)-SMCs after differentiation with either 2 ng/mL active TGFB2 or 2 ng/mL full-length TGFB2 (FL-TGFB2). The average expression in the active TGFB2-treated samples was set to 1. n = 6 biological replicates. (C) Top: western blots of CPC-SMC protein extracts after differentiation with either active TGFB1 or TGFB2. Bottom: quantification of western blot data showing relative protein SMC marker levels in CPC-SMCs. n = 6 biological replicates. (D) Top: western blots of canonical and non-canonical TGFβ signaling mediators in CPC-SMCs after 1-hour stimulation with either active TGFB1, TGFB2, or vehicle. Bottom: quantification of western blot data for pSMAD2, pSMAD3, pERK, and pAKT (T308) after the stimulation. The average protein levels in the TGFB1 treated samples were set to 1. n = 6 biological replicates. 1-way analysis of variance (ANOVA) with multiple comparisons test.
We also assayed the protein levels after CPC-SMC differentiation in active TGFB1 or active TGFB2 by immunoblots. Western blots revealed that both TGFB1 and TGFB2 treatments significantly increased SMC marker levels in CPC-SMCs (Figure 1C). The marker levels were similar between TGFB1 and TGFB2 treated CPC-SMCs further suggesting their redundant activity in CPC-SMC differentiation (Figure 1C). We also measured the activation of both canonical and non-canonical TGFβ signaling mediators in CPC-SMCs treated with either TGFB1 or TGFB2. To induce the phosphorylation of TGFβ downstream effectors, the differentiated CPC-SMCs were cultured in CPC-SMC basic medium for 24 hours, and then incubated with basic medium with or without 2 ng/mL active TGFB1 or TGFB2 for 1 hour prior to cell lysis. Consistent with the mRNA and protein quantifications, we observed similar levels of phosphorylated SMAD3 (pSMAD3), pSMAD2, pERK, and pAKT in CPC-SMCs treated with either TGFB1 or TGFB2 (Figure 1D).
Next, we examined whether the redundant activities of TGFB1 and TGFB2 in SMC contractile gene expression were independent of SMC developmental origins. To do this, we differentiated hiPSCs to NCSC-SMCs through neural crest stem cell (NCSC) lineage. MYH11 appeared to be the most sensitive gene in NCSC-SMCs to both TGFB1 and TGFB2 treatments (Figure S1A and B). Importantly, TGFB2-treated NCSC-SMCs had similar SMC marker expression compared to TGFB1-treated NCSC-SMCs suggesting their comparable activity in NCSC-SMC differentiation. Overall, these results suggest that TGFB1 and TGFB2 have indistinguishable effects on SMC marker expression in hiPSC-derived SMC differentiation.
TGFBR3 facilitates TGFB2 signaling transduction during SMC differentiation
Previous biophysical studies demonstrated that TGFB2 binds to TGFBR2 weakly and has the highest affinity to TGFBR3.47 TGFBR3 does not have kinase activity and plays a part in TGFβ ligand presentation to TGFBR1-TGFBR2 complex with the kinase activity.48 To investigate the contribution of TGFBR3 to SMC differentiation, we transfected CPC-SMCs with TGFBR3-siRNA and performed RT-qPCRs on CPC-SMCs differentiated with either active TGFB1 or TGFB2. TGFBR3-siRNA transfection reduced TGFBR3 levels by nearly 50% in both conditions (Figure 2A and B). TGFB1-treated CPC-SMCs were insensitive to TGFBR3 knockdown consistent with the low affinity of TGFB1 for TGFBR3 (Figure 2A). SMC markers including MYOCD and SMTN were reduced in TGFB2 treated CPC-SMCs upon TGFBR3 knockdown (Figure 2B), suggesting that TGFB2 additionally depends on TGFBR3 to exert its effects on SMC differentiation.
TGFBR3 facilitates TGFB2 signaling transduction during SMC differentiation. (A and B), Relative expression of SMC markers after TGFBR3-siRNA treatment. CPC-SMCs were transfected with 40nM TGFBR3-siRNA or control siRNA for 4 days during differentiation with either active TGFB1 or TGFB2. n = 6 biological replicates. ns: not significant; unpaired t-test or Mann-Whitney U test. (C) pSMAD3 and pSMAD2 western blots of CPC-SMC protein extracts stimulated with either active TGFB1, TGFB2 or vehicle for 1 hour. CPC-SMCs were transfected with 40 nM target siRNA (TGFBR1-siRNA or TGFBR3-siRNA) or control siRNA for 2 days. (D) Quantification of pSMAD3 western blot data. Each quantification was normalized to its vehicle control. The average levels in control siRNA-treated samples were set to 1. n = 5 biological replicates. ns: not significant; Kruskal-Wallis test with multiple comparisons test.
To understand the signaling differences between TGFB1 and TGFB2, we transfected CPC-SMCs with TGFBR1-siRNA, TGFBR3-siRNA, or control siRNA for 2 days including the starvation period and treated the cells with either active TGFB1 or active TGFB2 for 1 hour prior to cell lysis (Figure S2A). As expected, control siRNA did not alter the induction profile for the regulatory SMADs; SMAD3 and SMAD2, and pSMAD3 level was significantly reduced in CPC-SMCs treated with either TGFB1 or TGFB2 upon TGFBR1 knockdown (Figure 2C and D). In addition, TGFBR3 knockdown did not alter pSMAD3 and pSMAD2 levels in CPC-SMCs treated with TGFB1. We observed a sharp decrease in pSMAD3 levels in CPC-SMCs treated with TGFB2 upon TGFBR3 knockdown further suggesting that TGFB2 depends on TGFBR3 to activate SMAD3 (Figure 2C and D). pSMAD2 level was not sensitive to TGFBR3 knockdown in CPC-SMCs consistent with a previous report implicating the role of SMAD2 in NCSC-SMC differentiation.49
TGFB2 and TGFBR3 expression show enrichment in the human aortic root
To understand the expression differences between different TGFβ signaling ligands and receptors, we next collected matching aortic root, ascending aorta, and aortic arch samples from non-aneurysmal patients, who were eventually disqualified for the heart transplant. To measure the gene expression changes, we used different thoracic aorta regions from both female and male patients, denoted as ♀ Donor-1, ♀ Donor-2, ♂ Donor-1, and ♂ Donor-2 (Figure 3A, Figure S2B). RT-qPCRs of tunica media showed TGFB2 enrichment in the aortic root (Figure 3A). TGFB1 expression also appeared higher in the aortic root, while TGFB3 did not show a regional difference (Figure S2B). Among the TGFβ signaling receptors, the expression of core TGFβ receptors; TGFBR2 and TGFBR1; did not show a regional expression bias while TGFBR3 expression was higher in the aortic root (Figure 3B, Figure S2B). Next, we performed immunoblots on different thoracic aorta segments to confirm the enrichment of TGFB2 and TGFBR3 expression in the aortic root at the protein level (Figure 3C). The quantification of western blots validated higher TGFB2 and TGFBR3 levels in the aortic root tunica media (Figure 3D). These data suggest that there is a TGFB2-TGFBR3 expression gradient in the human aorta peaking closer to the heart.
TGFB2 and TGFBR3 expression show enrichment in the human aortic root. (A) Relative expression of TGFB2 in aortic media of root, ascending, and arch samples. TGFB2 expression level in each aortic root sample was set to 1. (B) Relative expression of TGFBR3 in aortic media of root, ascending, and arch samples. TGFBR3 expression level in each aortic root sample was set to 1. (C) TGFB2 and TGFBR3 western blots of tunica media protein extracts from different thoracic aorta regions. (D) Quantification of western blot data showing TGFB2 and TGFBR3 enrichment in the aortic root media. n = 4 biological replicates. Kruskal-Wallis test with multiple comparisons test.
TGFB2 haploinsufficiency causes SMC differentiation defects
To characterize the molecular changes caused by TGFB2 defects in the hiPSC model, we introduced a monoallelic 7 base-pair deletion in human TGFB2 gene (denoted as TGFB2KO/+) using CRISPR/Cas9 gene editing. The deletion causes a premature stop codon prior to the active TGFB2 (cytokine) domain (Figure 4A). RT-qPCRs and western blots revealed a significant reduction in TGFB2 mRNA and protein levels in heterozygous TGFB2KO/+ CPC-SMCs (Figure 4B). To examine the molecular consequences of TGFB2KO, we performed transwell co-culture assays. We seeded TGFB2KO/+ CPCs in the top layer and co-cultured them with either TGFB2KO/+ or TGFB2+/+ CPCs for 6 days in CPC-SMC differentiation medium (Figure 4C). In addition, TGFB2+/+ CPCs were co-cultured with TGFB2+/+ CPCs as a control. The 0.4 µm pores allowed the passage of secreted factors between the hanging insert and the dish while preventing the migration of the cells (Figure 4C). On day 7 of differentiation, TGFB2+/+ CPC-SMCs had higher expression of several SMC markers compared with TGFB2KO/+ CPC-SMCs (Figure 4D). Interestingly, TGFB2KO/+ CPC-SMCs co-cultured with TGFB2+/+ CPC-SMCs had significantly higher expression of SMC markers including MYH11 and CNN1 suggesting that secreted factors from TGFB2+/+ CPC-SMCs can improve SMC marker expression in TGFB2KO/+ CPC-SMCs (Figure 4D).
TGFB2 haploinsufficiency causes SMC differentiation defects. (A) The Sanger sequencing results for TGFB2+/+ and TGFB2KO/+ showing the position of 7 base-pair deletion resulting in a premature stop codon in TGFB2 gene. (B) Left: relative TGFB2 expression in TGFB2+/+ and TGFB2KO/+ SMCs. n = 6 biological replicates. unpaired t-test. Right: relative TGFB2 protein levels in TGFB2+/+ and TGFB2KO/+ SMCs. n = 3 biological replicates; unpaired t-test. (C) Diagram showing the co-culture assay design with TGFB2+/+ and TGFB2KO/+ CPC-SMCs. (D) Relative SMC marker levels in the following conditions: TGFB2+/+ CPC-SMCs co-cultured with TGFB2+/+ CPC-SMCs (black columns); TGFB2KO/+ CPC-SMCs co-cultured with TGFB2KO/+ CPC-SMCs (red columns) and TGFB2KO/+ CPC-SMCs co-cultured with TGFB2+/+ CPC-SMCs (gray columns). n = 6 biological replicates. ns: not significant; 1-way ANOVA with multiple comparisons test. (E) Relative SMC marker levels in TGFB2KO/+ CPC-SMCs supplemented with either active TGFB2 or vehicle for 7 days during CPC-SMC differentiation. The average expression in vehicle samples was set to 1. n = 6 biological replicates. ns: not significant; unpaired t-test or Mann-Whitney U test. (F) Relative SMC marker levels in TGFB2+/+ CPC-SMCs treated with 1ug/mL TGFB2 neutralization antibody. The average expression in the control samples was set to 1. n = 6 biological replicates. ns: not significant; unpaired t-test.
To formally test whether TGFB2 is among the secreted factors that promote SMC marker expression, we supplementedTGFB2KO/+ CPC-SMCs with either vehicle or 2 ng/mL TGFB2 for 7 days during the standard CPC-SMC differentiation with TGFB1. Consistently, SMC marker expression was significantly improved in TGFB2KO/+ CPC-SMCs after TGFB2 supplementation (Figure 4E). Lastly, we blocked TGFB2 activity in our culture system using a neutralizing TGFB2 antibody.50 To accomplish this, we supplemented the culture medium with the neutralizing TGFB2 antibody during the differentiation of TGFB2+/+ CPC-SMCs (Figure 4F). TGFB2 neutralization significantly reduced the expression of several SMC markers suggesting that SMC-produced TGFB2 contributes to CPC-SMC differentiation (Figure 4F).
TGFBR3KO/KO causes reduced TGFB2 sensitivity in TGFB2KO/+ SMCs
To further investigate the interplay between TGFB2 and TGFBR3, we used TGFB2KO/+ hiPSCs, and generated 2 independent clones of TGFB2KO/+TGFBR3KO/KO using CRISPR/Cas9 gene editing. The Clone 1 had a 14 base-pair bi-allelic deletion and the Clone 2 had 1 base-pair bi-allelic insertion in TGFBR3 gene resulting in premature stop codons in exon 9 of TGFBR3 gene (Figure 5A). Both clones had more than 1000-folds reduction in TGFBR3 expression (Figure 5B). Next, we differentiated TGFB2KO/+, TGFB2KO/+TGFBR3KO/KO, and their isogenic control hiPSCs (TGFB2+/+TGFBR3+/+) to CPC-SMCs using active TGFB2 (Figure 5C). As expected, SMC differentiation with active TGFB2 effectively rescued SMC marker expression in TGFB2KO/+ CPC-SMCs (Figure 5C). Strikingly, the marker expression in TGFB2KO/+TGFBR3KO/KO CPC-SMCs did not recover during TGFB2-mediated differentiation further implicating the TGFBR3 dependence in TGFB2 mediated signal transduction (Figure 5C).
TGFBR3KO/KO causes reduced TGFB2 sensitivity in TGFB2KO/+ SMCs. (A) The Sanger sequencing results showing 2 independent TGFBR3KO/KO hiPSC clones with premature TGFBR3 stop codons generated by CRISPR/Cas9 gene editing. (B) Relative TGFBR3 expression in different conditions. n = 6 biological replicates; 1-way ANOVA with multiple comparisons test. (C) Relative SMC marker expression in TGFB2KO/+, 2 TGFB2KO/+TGFBR3KO/KO clones and isogenic control (TGFB2+/+TGFBR3+/+) CPC-SMCs differentiated with TGFB2. n = 6 biological replicates; ns: not significant; 1-way ANOVA with multiple comparisons test. (D-E) MYH11 staining intensity and cell density in TGFB2KO/+, TGFB2KO/+TGFBR3KO/KO, and isogenic control (TGFB2+/+TGFBR3+/+) tissue rings generated using TGFB2. The average values in TGFB2+/+TGFBR3+/+ rings were set to 1. n = 6 biological replicates; ns: not significant; 1-way ANOVA with multiple comparisons test.
We also assessed the functional defects caused by TGFB2 and TGFBR3 deficiencies using 3D tissue ring constructs.51,52 To generate the tissue rings, TGFB2KO/+, TGFB2KO/+TGFBR3KO/KO, and their isogenic control CPC-SMCs were differentiated with active TGFB2 and seeded around 2 mm agarose molds. The rings were supplemented for 2 weeks with TGFB2 instead of TGFB1. To understand the molecular and cellular defects, we performed immunofluorescence stainings to assess SMC organization and contractile protein levels in the tissue ring constructs. MYH11 staining intensity and SMC distribution were improved in the TGFB2KO/+ rings generated with TGFB2, similar to the isogenic control rings (Figure 5D). MYH11 stainings of TGFB2KO/+TGFBR3KO/KO tissue rings from both clones revealed significantly weaker MYH11 staining intensity (Figure 5D and E). MYH11+ cells in the control and the TGFB2KO/+ rings had elongated morphology while they had uncharacteristic round shape for SMCs in the TGFB2KO/+TGFBR3KO/KO rings (Figure 5D). The cell density was also unaffected by TGFBR3KO/KO (Figure 5E) suggesting defective SMC differentiation in the presence of TGFB2 as the underlying cause for the ring defects.
TGFB2G276R variant disrupts the mechanical properties of vascular tissue rings
Lastly, we identified a family with heritable cardiovascular disorders for 3 generations of aortic aneurysm and dissection at Michigan Medicine. The Clinical Genetic testing revealed that the 54-year old male family member with aortic root aneurysm (root diameter: 5.2 cm) carries a variant in TGFB2 gene (c.826G > A; Chr1: 218436041 [GRCh38]; rs1169804851; ClinVar accession: VCV001450986; allele frequency = 0.000011 [3/264690, TOPMED]) and did not carry any variants in the remaining aortopathy panel genes. This TGFB2 variant causes a single amino acid substitution nearby the furin cleavage site in the latency-associated peptide domain of TGFB2 protein (p.Gly276Arg). To examine G276R substitution in the same isogenic background, we generated hiPSCs using the peripheral blood mononuclear cells of the male patient carrying the variant (PatientG276R/+), and corrected the mutation in the hiPSCs (Patient+/+) using CRISPR/Cas9 gene editing (Figure 6A). TGFB2 expression in PatientG276R/+ and Patient+/+ CPC-SMCs were similar while TGFB2 ELISA on cell lysates indicated lower TGFB2 protein levels in the PatientG276R/+ (Figure S3A). Cycloheximide chase assay showed a higher TGFB2 degradation rate in PatientG276R/+ cells compared to Patient+/+ cells suggesting reduced stability of TGFB2 p.Gly276Arg (Figure S3B). The transwell co-culture assays revealed that Patient+/+ CPC-SMCs have higher expression of SMC markers including MYH11, CNN1, and SMTN (Figure S3C). Similarly, PatientG276R/+ CPC-SMCs co-cultured with Patient+/+ CPC-SMCs had higher SMC marker expression (Figure S3C). We also treated PatientG276R/+ and Patient+/+ CPC-SMCs with vehicle or 2 ng/mL TGFB2 for 7 days during the standard CPC-SMC differentiation with TGFB1. As expected, SMC markers including MYH11, CNN1, and SMTN were upregulated in PatientG276R/+ CPC-SMCs after TGFB2 treatment (Figure S3D), while Patient+/+ CPC-SMCs did not appear to respond to the TGFB2 treatment (Figure S3E).
TGFB2G276R variant disrupts the mechanical properties of vascular tissue rings. (A) The Sanger sequencing results showing the position of c.826G > A variant and its CRISPR/Cas9-based genetic correction (black rectangle). (B) Representative stress-strain curves and maximum tensile stress values from PatientG276R/+, Patient+/+ tissue rings as well as PatientG276R/+ rings treated with TGFB2. n = 6 biological replicates; 1-way ANOVA with multiple comparisons test. (C-D) MYH11 staining intensity and cell density in PatientG276R/+, Patient+/+ tissue rings as well as PatientG276R/+ rings supplemented with TGFB2. The average values in PatientG276R/+ samples were set to 1. n = 6 biological replicates; ns: not significant; 1-way ANOVA with multiple comparisons test. (E) The Sanger sequencing results for TGFB2+/+ and TGFB2G276R/+ hiPSCs. (F) Representative stress-strain curves and maximum tensile stress values from TGFB2+/+, TGFB2G276R/+ as well as TGFB2G276R/+ tissue rings treated with TGFB2. n = 6 biological replicates; 1-way ANOVA with multiple comparisons test. (G-H) MYH11 staining intensity and cell density in TGFB2+/+, TGFB2G276R/+ as well as TGFB2G276R/+ tissue rings supplemented with TGFB2. The average values in PatientG276R/+ samples were set to 1. n = 6 biological replicates; ns: not significant; 1-way ANOVA with multiple comparisons test.
Next, we assessed the functional defects caused by the p.Gly276Arg substitution using the tissue ring model. TGFB2 ELISA without the acidification step confirmed the elevated active TGFB2 levels in the Patient+/+ tissue rings (Figure S3F). The tensile strength of the rings was improved in the Patient+/+ constructs (Figure 6B). TGFB2 supplementation during the tissue ring formation also improved the mechanical properties of PatientG276R/+ tissue rings (Figure 6B). Immunostainings revealed increased MYH11 intensity in the Patient+/+ tissue rings and the TGFB2-supplemented PatientG276R/+ rings, while the cell density was unchanged across different conditions (Figure 6C and D, Figure S4). To confirm this finding in an independent genetic background, we used hiPSCs (TGFB2+/+) from a male donor with no aortic abnormalities and generated hiPSCs carrying c.826G > A mutation (TGFB2G276R/+) by CRISPR/Cas9 gene editing (Figure 6E). The mutant TGFB2G276R/+ tissue rings had significantly lower tensile strength compared with the isogenic TGFB2+/+ tissue rings (Figure 6F). TGFB2 supplementation of the TGFB2G276R/+ rings improved the tensile strength (Figure 6F). Consistent with the PatientG276R/+ tissue ring data, immunostainings showed reduced MYH11 staining intensity and uncharacteristic round cell morphology in the TGFB2G276R/+ rings compared to the isogenic control TGFB2+/+ and TGFB2 supplemented TGFB2G276R/+ rings (Figure 6G and H, Figure S5). The cell density was unaffected by the p.Gly276Arg substitution suggesting defective SMC differentiation as the primary culprit for the mechanical and molecular changes (Figure 6H).
Discussion
Here we used hiPSC-derived SMC differentiation, CRISPR/Cas9 gene editing, SMC tissue constructs, and human tissue specimens to gain mechanistic insights into TGFB2-related aortopathy. Despite the similarities between the biological activities of different TGFβ isoforms in hiPSC-derived SMC differentiation, we found that TGFB2 additionally requires TGFBR3 to exert its effects on SMC gene expression. The molecular evaluation of different human thoracic aorta segments suggested TGFB2 and TGFBR3 enrichment in the aortic root. We found that TGFB2 haploinsufficiency causes second heart field-derived SMC differentiation defects that can be rescued by TGFB2 supplementation. This is in line with the findings from the Tgfb2 haploinsufficiency mouse model, which has aortic root dilation.7TGFBR3KO prevented the molecular rescue of TGFB2 haploinsufficiency with TGFB2 supplementation suggesting the TGFBR3 dependence in TGFB2 mediated SMC differentiation. SiRNA experiments were also in support of this finding. Our data also suggests that missense variant (c.826G > A, p.Gly276Arg) in TGFB2 gene causes mechanical and cellular deficits in the 3D tissue ring constructs, which can be rescued by CRISPR/Cas9-based genetic correction and TGFB2 supplementation. As this missense variant is located near the furin cleavage site, we speculate that it may impair the proteolytical activation of TGFB2, reducing activated TGFB2 levels in our culture conditions. In sum, this study reveals the regional TGFB2 expression patterns in the human thoracic aorta and its reliance on TGFBR3 in CPC-SMC differentiation, defects which could contribute to aortic root aneurysm formation.
There are 3 TGFβ isoforms identified in mammalian tissues; TGFB1, TGFB2, and TGFB3.53 Although pathogenic variants in TGFB2 and TGFB3 genes have been associated with LDS, there are no known pathogenic TGFB1 variants linked to the disease. Genetic disruption of different TGFβ isoforms in mice has non-overlapping features with Tgfb2 deficiency causing congenital heart defects and distinct cardiovascular abnormalities such as aortic wall thinning.20,26,54Tgfb2 is highly expressed in the aortic valves and aortic annulus in the developing mouse heart and MacFarlane et al. demonstrated TGFB2 enrichment in the mouse aortic root compared with ascending aorta.15,55 Our analysis of tunica media from different thoracic aorta regions revealed both TGFB2 and TGFBR3 enrichment in the human aortic root. The distinct TGFB2 expression pattern likely contributes to its nonredundant activity in cardiovascular development and the association between pathogenic TGFB2 variants and aortic root aneurysms. Additionally, we found that CPC-SMC differentiation is sensitive to TGFB2 supplementation and speculate that lineage-driven SMC sensitivity to TGFβ isoforms can also contribute to regional aortic aneurysm presentations in patients carrying pathogenic TGFβ signaling variants.
TGFBR3 is a critical but understudied member of TGFβ signaling. It has a high affinity for TGFB2 and promotes signaling via TGFBR1-TGFBR2 receptor complex by facilitating TGFβ ligand presentation.47 Pathogenic TGFBR3 variants have been associated with intracranial aneurysms.56 Similar to the unique phenotypic characteristics in Tgfb2 knockout mice, Tgfbr3 deficient embryos exhibit defects in outflow tract, heart valves, and coronary arteries, and comparative analyses of Tgfbr3 and Tgfb2 knockout embryos point to overlapping roles of these genes in heart development.57,58 Our data indicate that TGFB2 requires TGFBR3 to activate downstream TGFβ signaling during SMC differentiation. This distinct mode of action suggests that the interplay between TGFBR3 and TGFB2 should be better characterized in the context of aortic root aneurysms.
Patients with pathogenic TGFB2 variants exhibit milder TAAD and systemic phenotypes distinguishable from those seen in patients carrying TGFBR1, TGFBR2, or SMAD3 variants.7,30 Our comparative analysis indicates that different TGFβ isoforms display similar biological effects on hiPSC-derived SMC differentiation. Although TGFB2 shows enrichment in the aortic root, the other TGFβ isoforms are also expressed in this part of the aorta. We predict that the redundant activities of different TGFβ isoforms can compensate for TGFB2 defects resulting in relatively later onset and milder aggressiveness of pathogenic TGFB2 variants. In addition, a compensatory increase in the expression of TGFβ isoforms was reported in human aortas from patients with Marfan Syndrome or pathogenic TGFB2 variants implying additional transcriptional regulation to offset the defects.7,31,59
Summary
This study posits TGFB2 as a physiologically relevant cue for SMC differentiation in human aortic root, its reliance on TGFBR3 for signal transduction, and provides mechanistic implications about the TAAD manifestations in patients with TGFB2 defects.
Author contributions
Y.T.: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing—original draft, review and editing. J.C.: Data curation, Investigation, Validation. C.H.: Data curation, Investigation, Validation. P.Q.: Conceptualization, Methodology. J.L.: Data curation, Investigation, Validation. Y.E.C.: Conceptualization, Funding acquisition. D.M.: Conceptualization, Methodology, Writing—original draft, review and editing, Funding acquisition, Supervision, Visualization. B.Y.: Conceptualization, Resources, Writing—review and editing, Funding acquisition, Supervision, Visualization.
Funding
This study was supported by National Institutes of Health grants HL130614, HL141891, and HL151776 (B.Y.), and HL109946, HL134569, and HL159871 (Y.E.C.). D.M. is supported by MI-AORTA Rapid Research and Frankel Cardiovascular Center Aortic Research Grants.
Conflicts of interest
The authors have declared no potential conflicts of interest.
Data availability
The data that support the findings of this study are available from the corresponding authors upon reasonable request.
