In vitro lineage-specific differentiation of vascular smooth muscle cells in response to SMAD3 deficiency: Implications for SMAD3-related thoracic aortic aneurysm

Abstract

Objective:

SMAD3 pathogenic variants are associated with the development of thoracic aortic aneurysms. We sought to determine the role of SMAD3 in lineage-specific vascular smooth muscle cells (VSMCs) differentiation and function.

Approach and Results:

SMAD3 c.652delA, a frameshift mutation and nonsense-mediated decay, was introduced in human induced pluripotent stem cells (hiPSCs) using CRISPR-Cas9. The wild type and SMAD3^−/− (c.652delA) hiPSCs were differentiated into cardiovascular progenitor cells (CPCs) or neural crest stem cells (NCSCs), and then to lineage-specific VSMCs. Differentiation, contractility, extracellular matrix synthesis, and TGF-β signaling of the differentiated VSMCs were analyzed. The homozygous frameshift mutation resulted in SMAD3 deficiency and was confirmed in hiPSCs by Sanger sequencing and immunoblot analysis. In CPC-VSMCs, SMAD3 deletion significantly disrupted canonical TGF-β signaling and decreased gene expression of VSMC markers, including SM α-actin, myosin heavy chain 11, calponin-1, SM22α, and key controlling factors, SRF and myocardin, but increased collagen expression. The loss of SMAD3 significantly decreased VSMC contractility. In NCSC-VSMCs, SMAD3 deficiency did not significantly affect the VSMC differentiation but decreased elastin expression and increased phosphorylated SMAD2. Expression of mir-29 was increased in SMAD3^−/− VSMCs, and inhibition of mir-29 partially rescued elastin expression.

Conclusion:

SMAD3-dependent TGF-β signaling was essential for the differentiation of CPC-VSMCs, but not for the differentiation of NCSC-VSMCs. The lineage-specific TGF-β responses in human VSMCs may potentially contribute to the development of aortic root aneurysms in patients with SMAD3 mutations.

Graphical Abstract

Introduction

Thoracic aortic disease has a strong genetic predisposition, with 20% of affected patients having a positive family history.¹ Causative genes for heritable thoracic aortic disease (HTAD) include genes encoding proteins involved in canonical transforming growth factor-β (TGF-β) signaling, the extracellular matrix (ECM) homeostasis, and vascular smooth muscle contraction.^2–4 The involvement of the canonical TGF-β signaling pathway in the pathogenesis of thoracic aortic aneurysm (TAA) is evident by genetic studies in humans. Mutations in six genes in the signaling pathway including TGFBR1, TGFBR2, SMAD2, SMAD3, TGFB2, and TGFB3, lead to thoracic aortic disease characterized, in part, by aortic root aneurysms.^2–5 SMAD3 is a key transcription factor in the TGF-β signaling pathway, playing a critical role in vascular smooth muscle differentiation and ECM deposition.^6–8 Heterozygous SMAD3 pathogenic mutations predispose to vascular diseases, primarily aortic root aneurysms, which can progress to type A aortic dissections if the aneurysm is not surgically repaired.^{9, 10} The majority of SMAD3 mutations are loss-of-function mutations, including deletions, frameshift, and nonsense mutations.^{5, 9, 11–13}

The underlying molecular mechanism of genetic mutations, including SMAD3 mutations, in the TGF-β pathway leading to the development of TAA is unknown. It is also unclear as to why affected patients develop aortic root aneurysms that typically spare the ascending aorta.¹⁴ The aortic root is mainly comprised of VSMCs from cardiovascular progenitor cells (CPCs) in the second heart field, and the ascending aorta is comprised of VSMCs from neural crest stem cells (NCSCs) and the second heart field.^{15, 16} Although SMAD3 mutations are predicted to decrease TGF-β signaling, there is a paradoxical elevation in TGF-β signaling observed in end-stage aneurysmal tissue from patients with these mutations^17–20. In this study, we used CRISPR/Cas9 to create SMAD3 deficiency in human induced pluripotent stem cells (hiPSCs). We aimed to determine the response of VSMCs from different lineages (CPCs and NCSCs) to TGF-β during differentiation from iPSCs to VSMCs. We hypothesize that VSMCs from separate lineages (CPC or NCSC) will respond differently to TGF-β when the SMAD3-dependent pathway is obliterated. Consequently, the SMAD3 deficiency will result in defective differentiation of CPC-VSMCs.

Methods

The authors declare that all supporting data are available within the article and its online-only Data Supplement.

hiPSCs generation and SMAD3 gene editing:

The process of hiPSCs generation and teratoma formation has been previously described²¹. This study was approved by the institutional review board at the University of Michigan. Two independent single-guided RNA (sgRNA) were designed to target SMAD3 with a previously developed tool by Zhang Feng’s group. The sequence of SMAD3 sgRNA was: 5’-GTCCCCAGCACATAATAACT −3’ and 5’- TGCTGGGGACATCGGATTCG −3’. The targeted site was chr15: +67462919 and chr15: −67462906. Using a previously described method, sgRNA was cloned into PX458, which contains SpCas9–2A-EGFP²². One million iPSCs were electroporated with 5μg of PX458/ SMAD3 sgRNA using Lonza Human Stem Cell Nucleofector® Kit 2 with program U-023 on Nuclefector 2 device (Lonza Ltd.). Cell culture dishes were first coated with Corning® Matrigel® Growth Factor Reduced (GFR) Basement Membrane Matrix (Cat#354230) for 60 minutes in cell culture incubator (37°C, 5%CO₂). Transfected hiPSCs were seeded on Matrigel-coated 60mm dishes in TesRE8 medium (STEMCELL Technologies) at concentration of 200/per dish and cultured until clones were formed. The targeted site of SMAD3 was sequenced in each clone. Homozygous clones with SMAD3 frameshift mutations were selected for further experiments. Isogenic WT clones were used as controls.

Induction of hiPSCs into CPC-VSMCs

The CPC-VSMC differentiation was performed per the Cowan’s group²³ with a modification. The hiPSCs were separated with Accutase (STEMCELL Technologies) and transferred to Matrigel-coated culture dishes at a density of 2× 10⁴ cells/cm² in TesRE8 medium with 10 μM ROCK inhibitor Y27632 (Stemgent). Twenty-four hours later, the medium was converted to CPC differentiation medium containing DMEM/F12(Gibco® ThermoFisher), 1× B27 without vitamin A (Life Technologies), 25ng/ mL BMP4 (PepproTech)and 8μM CHIR99021(Biogen’s), 50μg/mL ascorbic acid (Sigma), 400μM 1-thioglycerol (Sigma), and 1%pen/strep(Gibco® ThermoFisher). Next, cells were further differentiated to CPCs for 3 days.

For VSMC differentiation, CPCs were detached and individualized with Accutase and transferred to Matrigel-coated culture dishes at a density of 1.6× 10⁴/cm² in CPC-VSMC differentiation basic medium (DMEM/F12, 1× B27 without vitamin A, 1%pen/strep, 400μM 1-thioglycerol) with 2ng/mL TGF-β1 (PeproTech) and 10ng/mL PDGFBB(PeproTech). 10 μM ROCK inhibitor Y27632 was added on day 1 to facilitate cell survival. Next, cells were cultured for another 6 days.

Induce hiPSCs into NCSC-VSMCs

We modified the method of NCSC-VSMC differentiation described previous^{24, 25}. In brief, hiPSCs were separated into single cells with Accutase. The separated cells were then reseeded on Matrigel-coated culture dishes at a density of 2× 10⁴ cells/cm² in TesRE8 medium. 10 μM Y27632 was added on day 1. Cells were cultured for 2–3 days to approach 40–60% confluence, and the medium was switched to NCSC differentiation for another 6 days. On day 1 of NCSC differentiation, cells were cultured in NCSC basic medium (DMEM/F12, 1× N2 supplement (Life Technologies), 0.1%BSA (sigma), 1% pen/strep) with 10μM SB4315421(Stemgent) and 1μM LDN193189 (Stemgent). From day 2 to day 6, 3μM CHIR99021 was added to the above medium.

For VSMC differentiation, NCSCs were separated into single cells with Accutase. The separated cells were then transferred to Matrigel-coated culture dishes at a density of 1× 10⁵/cm² in NCSC basic medium with10μM Y27632. 24 hours later, cells were induced with NCSC-VSMC differentiation basic medium (DMEM/F12, 20% knockout serum replacement (Thermo Fisher Scientific), 1%pen/strep) with 2ng/mL TGF-β1) for 8 days.

Real-time quantitative reverse transcription PCR (qRT-PCR)

Total cellular RNA was extracted with RNeasy Mini Kit (Qiagen). 0.3 μg of RNA was used to synthesize cDNA using Superscript III first-strand synthesis system (Invitrogen) according to kit manual. The qRT-PCR was performed with iQ SYBR Green Supermix kit (Bio-Rad) on a StepOne Plus Real-Time System (Life Technologies). A comparative cycle threshold method was used to calculate the relative mRNA levels. GAPDH was used as an internal control.

Western Blot Analysis

RIPA buffer supplemented with protease inhibitor and phosphatase inhibitor (Roche Applied Science) was used to extract protein from the cells. Cell lysate samples were then resolved on SDS-PAGE gel and transferred to nitrocellulose membranes. Primary antibodies used in the Western blot were listed in the Major Resources Table in the Supplemental Material. Immuno-activity and density of the protein band were visualized by the Odyssey system (LI-COR Biosciences). The protein samples were harvested from hiPSCs for SMAD3 detection and CPC/NCSC-VSMCs at the end of differentiation for contractile genes detection.

To examine the response of VSMC to TGF-β, we cultured differentiated CPC-VSMCs and NCSC-VSMCs respectively in CPC-VSMC and NCSC-VSMC differentiation basic medium without growth factors supplements for 24 hours. We then changed the culture medium to the basic medium with or without 2 ng/ml TGF-β1 for one hour in the incubator. The protein samples were harvested and analyzed in Western blotting.

To measure the secretion of tropoelastin in culture medium, differentiated NCSC-VSMCs (Day 8) were grown in DMEM media with 2ng/ml TGF-β1 for 24 hours. Proteins in the medium were precipitated by 10% trichloroacetic acid. The protein loading was normalized by the DNA concentration of the NCSC-VSMCs.

Carbachol contraction study

Differentiated CPC/NCSC-VSMCs were stimulated with 100 μM carbachol (Sigma). The change of the cell surface was recorded for 30 minutes by Olympus IX73P2F Imaging system. The percentage of surface reduction was analyzed by ImageJ software (NIH).

Gel contraction study

1 × 10⁶ CPC-VSMCs were mixed with 1.85 mg/mL type I collagen solution (Cell Biolabs Inc.) following the manufacturer’s instruction. The mixture was added into 24-well tissue culture plates (500 μL/well) and allowed to gel in the incubator for 30 min. Then 1 mL CPC-VSMC differentiation medium was added to the gel. After 24 hours, the collagen gel was mechanically lifted, and pictures were taken with Olympus IX73P2F Imaging system 12 hours later. The reduction of the gel surface was measured using ImageJ software.

miR-29b expression by RT-qPCR

For microRNA detection, total RNA was extracted by a miRNeasy mini kit (Qiagen). 0.3 μg RNA was used to synthesis cDNA following the instruction of TaqMan™ MicroRNA Reverse Transcription Kit (Thermo fisher). RT-qPCR was performed using TaqMan™ MicroRNA assay kit. RNU6 was used as an internal control. The primers of miR-29b and RNU6 were designed by Thermo fisher.

Inhibit miR-29b in differentiated NCSC-VSMCs

Transfection of miR-29b inhibitor was performed with the Lipofectamine RNAiMAX reagent (Thermo Fisher Scientific) according to manufacturer’s instruction. The differentiated WT and SMAD3^−/− NCSC-VSMCs (Day 4) were treated with 60nM mirVana™ miRNA inhibitor negative control (Thermo Fisher Scientific) or anti-hsa-miR-29b (Thermo Fisher Scientific) for 8 hours. Next, the medium was replaced with the NCSC-VSMC differentiation medium as described above. Following 16 additional hours, cells were harvested for RT-qPCR study.

Engineered tissue ring

An engineered tissue ring was generated following Dr. Rolle’s previously described protocol²⁶. Differentiated CPC-VSMCs were harvested and seeded at 1 million/well in DMEM/F12 with 10% FBS. 24 hours later, tissue rings were cultured in Ring-forming medium (DMEM (high glucose), 20% FBS, 1% pen/strep and 1% MEM Non-Essential Amino acid solution (MEM NEAA, Gibco® ThermoFisher) 10 ng/ml PDGF-BB and 1 ng/ml TGF-β1, 3ng/ml CuSO4 (Sigma) and 50ug/ml ascorbic acid) for another 7 days.

Tensile strength

Tensile strength was performed using a uniaxial testing machine. In brief, the rings were measured at three different locations around the ring circumference and were averaged to find the mean ring thickness. The tissue rings were mounted on custom grips and submerged in PBS. Tissue rings were pulled to failure at a speed of 10mm/min. Force (F) and displacement (l) were recorded continuously throughout the test. The measured thickness value was used to calculate the initial cross-sectional area. Data were analyzed to calculate ultimate tensile strength (UTS) and maximum tangent modulus (MTM).

Histology

An engineered tissue ring was fixed in 4% Paraformaldehyde and embedded in paraffin. Five micrometer sections were adhered to Superfrost Plus Microscope Slides (VWR). The sections were stained by hematoxylin and eosin stain (Thermo Fisher Scientific). To detect collagen, Masson trichrome staining was performed with Bouin’s solution and Weigert’s Iron Hematoxylin solution (Sigma).

Statistical Analyses

The data are expressed as mean ± SD, and Prism 7.0 software was used for the analysis (GraphPad Software, CA), Shapiro-Wilk normality test, and F test for the equal variance were performed. If Shapiro-Wilk normality test failed, the significance of the results was analyzed by Mann Whitney test. If Shapiro-Wilk normality test succeeded but the F test failed, the significance of the results was analyzed using a t-test with Welch’s correction. If both of the tests succeeded, the significance of the results was analyzed by two-sided Student’s t-test. P<0.05 was considered statistically significant.

Result

Homozygous frame-shift mutations in SMAD3 were created in hiPSCs with CRISPR Cas 9

The hiPSCs were generated from peripheral blood mononuclear cells of a healthy donor with normal sized aorta and aortic valve²¹.

CRISPR/Cas9 system was used to alter SMAD3 in hiPSCs. By Cas9-mediated ‘Indel’ reaction through the guidance of sgRNA (Figure 1A), the homozygous clones with SMAD3 frameshift mutations were selected, and the genotypes were confirmed by Sanger sequencing. The clones with one base pair deletion at SMAD3 designated site, named SMAD3^−/− clones, had been taken for further study (Figure 1B), which harbor a frameshift mutation at 213th amino acid and premature stop codons at 235th amino acid. This frameshift mutation was identical to a pathogenic heterozygous SMAD3 mutation called c.652delA identified in AOS patients. The SMAD3 protein was undetectable in the mutated hiPSC clones using anti-SMAD3 N-terminal antibody in Western blotting, indicating the SMAD3 deficiency in SMAD3^−/− hiPSCs (Figure 1C).

Figure 1: Frame-shift homozygous mutations in SMAD3were created in hiPSCs with CRISPR Cas 9:

(A, D): Two independent guide RNAs were designed to target SMAD3; Sanger sequencing showed the genotypes of SMAD3 homozygous truncated clones, which harbored frameshift mutations of 1 base-pair (SMAD3^−/−)and 7 base-pair deletion (SMAD3^KO) respectively. (B, E): The red markers indicated the deleted base pairs. The black arrows indicated the cutting sites. (C, F): The SMAD3 protein was undetectable using the N-terminal antibody in the SMAD3^−/− /SMAD3^KO hiPSCs.

To avoid the potential influence of off-target on the phenotype, an independent sgRNA (Figure 1D) was designed to introduce 7 base pairs deletion in a designated site of SMAD3 gene (Figure 1E). This homozygous clone designated as SMAD3^KO clone harbored a frameshift mutation at 207th amino acid and premature stop codons at 235th amino acid. SMAD3 protein was also undetected in SMAD3^KO hiPSCs (Figure 1F). Experiments of characterization of SMAD3 deficiency were repeated with SMAD3^KO clone.

Defective differentiation and decreased contractility were detected in SMAD3^−/− CPC-VSMCs.

Both wild type (WT) and SMAD3^−/− hiPSCs were differentiated into CPC-VSMCs. We assessed whether induced SMAD3 deficiency could affect CPC-VSMC differentiation and contractility. Compared to WT CPC-VSMCs SMAD3^−/− CPC-VSMCs exhibit oval and immature morphology (Figure 2A). RT-qPCR showed decreased mRNA expression of VSMC markers including ACTA2, CNN, TAGLIN in SMAD3^−/− CPC-VSMCs (Figure 2B). The key controlling factors in VSMC differentiation, myocardin and serum response factor (SRF), were also downregulated in SMAD3^−/− CPC-VSMCs (Figure 2D). The protein levels of VSMC markers (SM α-actin, calponin-1, SM22-α, and myosin heavy chain11) in the SMAD3^−/− CPC-VSMCs were also decreased when assessed by immunoblot analysis (Figure 2C, FigureⅠin online-only Data Supplement). The contractility of SMAD3^−/− CPC-VSMC was significantly decreased as demonstrated by carbachol assay (Figure 3A, FigureⅡ in online-only Data Supplement) and gel contraction assay (Figure 3B, FigureⅢ in online-only Data Supplement). Our results indicated that SMAD3 deficiency could impair CPC-VSMC differentiation and contractility.

Figure 2: Defective differentiation was detected in SMAD3^−/− CPC-VSMCs.

(A): Morphology of WT (wild type) and SMAD3^−/− CPC-VSMCs. Scale bars: 200 μm. (B): Relative mRNA expression of VSMC markers in WT and SMAD3^−/− CPC-VSMCs, including ACTA2, CNN, TAGLN, MYH11, SMTNB. (C): Western blotting of VSMCs markers in WT and SMAD3^−/− CPC-VSMCs including SM α-actin, calponin-1, SM22α, and SMMHC. Right panel: relative quantification of western blot data. (D): RT-qPCR result of critical VSMC differentiation regulators including SRF and MYOCD in WT and SMAD3^−/− CPC-VSMCs; The results are presented as mean ± s.d. of 3 independent experiments; *P<0.05; **P<0.01; ***P<0.001; NS, not significant; two-sided Student’s t-test.

Figure 3: Decreased contractility was dectected SMAD3^−/− CPC-VSMCs.

(A): Quantification result of VSMC surface area reduction after carbachol stimulation. WT (wild type) and SMAD3^−/− CPC-VSMCs were treated with 100 μM carbachol for 30 minutes. (B): Representative collagen gel assay. WT and SMAD3^−/− CPC-VSMCs were cultured in collagen gel for 24 hours, and collagen gels were then lifted to initiate the contraction. Right panel: quantification of gel surface reduction; The results are presented as mean ± s.d. of 3 independent experiments; * P<0.05; ** P<0.01; *** P<0.001; NS, not significant; two-sided Student’s t-test.

Vascular tissue ring with SMAD3 deficiency demonstrated enhanced collagen fiber deposition and structural stiffness.

The haematoxylin and eosin (HE) staining of engineered Vascular Tissue ring (Figure 4A) showed a similar histological structure in WT and SMAD3^−/− tissue ring. Masson trichrome staining (Figure 4B) showed that SMAD3^−/− tissue rings had more collagen fiber deposition. Consistently, SMAD3^−/− tissue rings showed higher failure force and a 2.5 increase in structure stiffness compared to WT groups (Figure 4C, 4D).

Figure 4: Self-assemble tissue ring of SMAD3^−/− CPC-VSMCs demostrated higher stiffness compared to wild type (WT) CPC-VSMCs.

(A): Morphology of the self-assembled tissue rings of CPC-VSMCs by HE staining. (B): Masson trichrome staining of collagen fiber deposition in the tissue rings. Scale bars: 200 μm. (C): The tensile strength represented by the stress-strain curve. (D): The relative quantification of structural stiffness represented by maximum tangent modulus (MTM) of the stress-strain curve; The results are presented as mean ± s.d. of 3 independent experiments. ** P<0.01; two-sided Student’s t-test.

Effect of SMAD3 deficiency on NCSC-VSMC differentiation and contractility

The ascending aorta was populated by VSMCs derived from NCSCs and CPCs .¹⁶ After differentiation, there was no significant difference between WT and SMAD3^−/− NCSC-VSMCs according to the morphology (Figure 5A). Interestingly, expression of SM α-actin was significantly decreased but calponin-1, SM22α, myosin heavy chain 11, or smoothelin B levels were not significantly changed in the SMAD3^−/− NCSC-VSMCs (Figure 5B and C, Figure Ⅳ in online-only Data Supplement). Furthermore, carbachol assay showed that contractility function was not affected by SMAD3 deficiency in NCSC-VSMCs (Figure 5D, FigureⅤ in online-only Data Supplement).

Figure 5: Similar differentiation and contractility were detected in SMAD3^−/− NCSC-VSMCs compared to wild type(WT) NCSC-SMCs.

(A): Morphology of WT and SMAD3^−/− NCSC-VSMC after differentiation. Scale bars: 500 μm. (B): Relative mRNA expression of VSMC markers in WT and SMAD3^−/− NCSC-VSMCs, including ACTA2, CNN, TAGLN, MYH11, and SMTNB. (C): Western blotting of VSMC markers in WT and SMAD3^−/− NCSC-VSMCs including SM α-actin, calponin-1, SM22α, and SMMHC. Right panel: relative quantification of western blotting. (D): Quantification of NCSC-VSMC surface area reduction after carbachol stimulation. WT and SMAD3^−/− NCSC-VSMCs were treated with 100μM carbachol for 30 minutes; The results are presented as mean ± s.d. of 3 independent experiments. * P<0.05; ** P<0.01; *** P<0.001; NS, not significant; two-sided Student’s t-test.

SMAD3 deficiency decreased elastin (ELN) synthesis in CPC- and NCSC-VSMCs.

Elastic fiber plays an important role in maintaining elasticity and structure integrity of the aortic wall. Disruption of elastic fiber in aortic media was reported to be a major pathogenic change in AOS patients with aneurysm.¹² We found that disruption of SMAD3 could attenuate elastin expression in both NCSC- and CPC-VSMCs in mRNA level (Figure 6A, Figure ⅥA,B,C in online-only Data Supplement). The secretion of tropoelastin, the precursor of elastin, was also examined in the conditional medium of NCSC-VSMCs. The western blotting showed dramatically decreased tropoelastin secretion from SMAD3^−/− NCSC-VSMCs (Figure 6B, Figure ⅥD in online-only Data Supplement). During the NCSC-VSMC differentiation, the expression of miR-29b and elastin were inversely correlated. SMAD3^−/− NCSC-VSMCs had a two folds higher mir-29 expression and significantly lower elastin expression from day 4 to day 8 (Figure 6C). Subsequently, WT and SMAD3^−/− NCSC-VSMCs were treated with negative control (NC) or miR-29b inhibitor on Day 4. Inhibiting miR-29 could partially rescue the synthesis of ELN (Figure 6D).

Figure 6: Decreased elastin synthesis and increasing miR-29b expression were detected in SMAD3^−/− NCSC-VSMCs compared to wild type (WT) NCSC-VSMCs.

(A): The expression of elastin (ELN) in WT and SMAD3^−/− NCSC-VSMC by RT-qPCR. (B): The secretion of tropoelastin in WT and SMAD3^−/− NCSC-VSMCs in conditional medium detected by western blotting. Lower panel: quantification of western blot data. (C): Expression of miR-29b and ELN during the differentiation from NCSCs to VSMCs on Day 0,2,4,6,8 detected by RT-qPCR in WT and SMAD3^−/− NCSC-VSMCs. (D): Relative mRNA level of ELN in WT and SMAD3^−/− NCSC-VSMCs treated with 60nM negative control (NC) and 60nM miR-29b inhibitor; The results are presented as mean ± s.d. of 3 independent experiments. ** P<0.01, *** P<0.001; two-sided Student’s t-test.

SMAD3 deficiency affected TGF-β signaling pathway in CPC- and NCSC- VSMCs

Change of TGF-β signaling caused by induced SMAD3^−/− mutation was measured at the end of hiPSC-derived VSMC differentiation. In CPC-VSMCs, the downstream genes of the canonicalTGF-β pathway, such as plasminogen activator inhibitor type 1 (PAI-1) and connective tissue growth factor (CTGF), were significantly down-regulated in SMAD3^−/− cells compared to WT cells, indicating decreased responsiveness to TGF-β due to SMAD3 deficiency (Figure 7A, Figure ⅦA in online-only Data Supplement). After TGF-β1 treatment, in CPC-VSMCs, the level of phosphorylated SMAD2 was not significantly different between WT and SMAD3^−/− cells, while SMAD3^−/− cells exhibited decreased phosphorylation of AKT, indicating down-regulated signaling in non-canonical TGF-β pathway in the absence of SMAD3 (Figure 7B). Interestingly, NCSC-VSMCs exhibited different responsiveness in TGF-β signaling in the absence of SMAD3. Expression of PAI-1 was up-regulated in the SMAD3^−/− NCSC-VSMCs (Figure 7C, Figure ⅦB in online-only Data Supplement) and phosphorylated SMAD2 levels increased after TGF-β1 treatment, indicating over-activated TGF-β canonical signaling in the SMAD3^−/− NCSC-VSMCs (Figure 7D). However, there was no change of signaling of non-canonical TGF-β pathway as indicated by p-ERK, p-AKT, and p-P38 (Figure 7D).

Figure 7: Effect of SMAD3 deficiency on TGF-β signaling in CPC-VSMCs and NCSC-VSMCs:

The CPC-VSMCs and NCSC-VSMCs were cultured in media without TGF-β and PDGF-bb 24 hours after differentiation, then treated with or without TGF-β 2 ng/ml for 1 hour. (A): The relative mRNA expression of PAI-1 and CTGF in wild type (WT) and SMAD3^−/− CPC-VSMCs. (B): Up panel: Western blotting of phosphorylated protein and total protein of intracellular mediators in TGF-β signaling pathways in WT and SMAD3^−/− CPC-VSMCs, including SMAD2, ERK, AKT, and P38. Lower panel: The relative quantification of western blotting. (C): The relative mRNA expression of PAI-1 and CTGF in wild type (WT) and SMAD3^−/− NCSC-VSMCs. (D): Up panel: Western blotting of phosphorylated protein and total protein of intracellular mediators in TGF-β signaling pathways in WT and SMAD3^−/− NCSC-VSMCs, including SMAD2, ERK, AKT, and P38. Lower panel: relative quantification of western blotting. The results are presented as mean ± s.d. of 3 independent experiments. * P<0.05; ** P<0.01; *** P<0.001; NS, not significant; two-sided Student’s t-test.

Discussion:

To determine the role of SMAD3-dependent TGF-β signaling in VSMC differentiation in different lineages, namely CPC-VSMCs and NCSC-VSMCs, we blocked the SMAD3-dependent pathway (canonical pathway) by creating homozygous genotype of SMAD3 ^c.652delA with CRISPR Cas9. Our cellular model successfully recapitulated the established phenotype of SMAD3 mutated patients and Smad3 knockout mice, such as increased collagen deposition (Figure 4A), decreased elastin in the aortic aneurysm (Figure 6A, B), and increased pSMAD2 in the ascending aortic aneurysm, which is comprised mainly by the NCSC-VSMCs (Figure 7D). Our results indicated that the abnormal findings in the aneurysmal tissue in patients with SMAD3 mutations are not due to aneurysmal dilation of the aorta, but rather the intrinsic pathology of aortic VSMCs. In addition, we found CPC-VSMC differentiation was SMAD3 dependent whereas NCSC-VSMCs differentiation was SMAD3 independent. The TGF-β-Akt pathway decreased in CPC-VSMCs but not in NCSC-VSMCs when the SMAD3-dependent pathway was blocked by knocking out SMAD3.

SMAD3, a key player in the TGF-β canonical pathway, is critical for VSMC differentiation.^{7, 27} Early embryo-originated cell line studies have demonstrated that SMAD3, but not SMAD2, is a stronger potent activator for VSMC specific gene transcription in mesodermal-originated 10T1/2 cells.²⁸ SMAD3 has been shown to directly interact with myocardin (MYOCD), a key VSMC transcription co-activator, and docked TGF-β mediated transcriptional regulatory complex on the SMADs-binding elements (SBE) even without SRF binding support.²⁹ In contrast, SMAD2 is more important than SMAD3 in controlling neural crest-originated VSMC specific gene transcription through interacting with myocardin-related transcription factor B (MRTFB).³⁰ Accordingly, our findings are consistent with studies showing that VSMC differentiation from CPC is SMAD3 dependent, whereas from NCSC is SMAD3 independent.

Our data were consistent with recent observations in an Loeys-Dietz Syndrome (LDS) mouse model (Tgfbr1^M318R/+) when two lineage-specific vascular events associated with aortic root aneurysm are comparably analyzed, namely secondary heart field and cardiac neural crest (CNC) lineages.¹⁴ Macfarlane et al. reported that the mouse vascular predisposition to aortic root dilation was primarily caused by the defective effects of TGF-β signaling in Tgfbr1^M318R/+ SHF-derived VSMCs as well as excessive Smad2/3 activation in Tgfbr1^M318R/+ CNC-derived VSMCs. The excessive pSmad2/3 were found in the root/proximal ascending aorta but not in the distal ascending aorta, and targeting of Smad2 specifically in the CNC essentially prevented aneurysm.¹⁴ In our hiPSC model of homozygous mutation of SMAD3^c.652delA, we also found increased TGF-β signaling in NCSC-VSMCs without hemodynamic stimulation. This finding indicated a potential mechanism of excessive TGF-β signaling in proximal aorta comprised of NCSC-VSMCs could contribute to the aortic aneurysm. This finding was also consistent with the conventional wisdom that increased TGF-β signaling in the proximal aorta in Marfan or Loeys-Dietz Syndrome mouse models^{14, 18} and patients^{17, 31}. Results from both labs were presented at the GenTAC meeting (Oregon Health Science Center, 2018). It was fascinating to observe that two labs independently had similar findings in different models with gene editing: in vivo mouse model and in vitro human iPSC model of mutations of genes in TGF-β pathway.

The second potential mechanism of aortic root aneurysm in patients with SMAD3 mutations could be defective differentiation of SMCs in the aortic root. The loss-of-function mutations of SMAD3 lead to impaired signaling of TGF-β canonical pathway during CPC-VSMC differentiation. As our study showed, the VSMCs differentiated from CPCs were SMAD3 dependent, which contribute to the majority VSMCs of the aortic root.^{15, 17, 32, 33} The expression of contractile protein and generation of contractile force could be compromised in CPC-VSMCs with loss-of-function mutations of SMAD3. Mutations in the VSMC contractile apparatus predispose to the formation of TAA.^34–36 Different groups have demonstrated that the down-regulation of expression for VSMC contractile protein genes widely occurred in TAA/TAAD patients and the expression of SM22α was inversely correlated with patients’ aneurysm size,^37–40 including LDS.⁴¹ In vitro cell culture of VSMC that either explanted from LDS patients who carried TGFBR2 mutation and enforced expression of TGFBR2^G357W, or knock-down SMAD3 protein are all presented by decreased VSMCs contractile protein production.^{38, 40} Mechanistically, the defective differentiation of CPC-VSMCs could be due to decreased expression of myocardin and SRF in SMAD3^−/− CPC-VSMCs, as myocardin is the master regulator of differentiation of VSMCs in mesoderm-originated VSMCs.⁴² Taken together, our study suggested that loss-of-function mutations of SMAD3 could lead to defective contractile function in CPC-VSMCs at the aortic root, and result in an aortic root aneurysm in response to blood pressure in the aorta in patients with SMAD3 mutations.

On the other hand, some other studies also show that SMC contractile proteins do not decrease in the LDS mouse model. The Smad3^−/− mice demonstrated normal expression levels of smooth muscle α-actin (SMA) protein in aneurysmal aortic tissues compare to wild-type mice.^{43, 44} Pathologically, the Smad3^−/− mouse had extensive inflammation and thickening of the aortic wall, which we do not see in the aneurysm of patients with SMAD3 mutations. The SMA is not a specific protein for VSMCs. The overstaining of the SMA in the Smad3^−/− mouse could be due to inflammation and thickening of the aortic wall since fibroblasts can be stained positive with SMA Antibody. MacFarlane¹⁴ also showed preservation of VSMC representation and marker gene expression in the root and ascending aorta of LDS mice, as marked by MYH11, SMMHC, and αSMA. This difference could be explained by the fact that our hiPSC model had a complete disruption of SMAD3, and their LDS mouse model only had a point mutation of Tgfbr1.

The third potential mechanism could be that defects in the extracellular matrix in the aortic root of patients with SMAD3 mutations. Loss and fragmentation of elastic fibers in aorta media are typical pathogenic changes in inherent TAA diseases, including SMAD3 mutations.^{2, 41, 45} The expression of elastin is mainly controlled at post-transcriptional levels, by which the elastin mRNA stability is modulated by TGF-β signaling,^{46, 47} specifically SMAD3 mediated regulation for miR-29b metabolism.⁴⁸ miR-29b is a strong inhibitor of elastin synthesis⁴⁹, and associated with TAA development.^50–53 SMAD3 was predicted to suppress miR-29b expression.⁴⁹ As expected, we found impaired elastogenesis in SMAD3^−/− VSMCs from both lineages, (Figure 6A–C, Figure VIA in online-only Data Supplement) especially in NCSC-VSMCs. Suppression of miR-29b expression with miR-29 inhibitor partially rescued elastin expression in SMAD3^−/− NCSC-VSMCs, (Figure 6D) indicating miR-29 involvement in inhibiting expression of elastin in SMAD3^−/− NCSC-VSMCs. Combined with the decrease of SM α-actin in SMAD3^−/− VSMCs from both lineages, SMAD3 mutations could cause reduced formation of elastic fiber and disrupt the elastin-contractile unit, causing structural aortic defects, leading to aortopathy of the aortic root, ascending, and arch in patients with SMAD3 mutations.

Collagen fiber is also an important ECM component and contributes to the structural stiffness of the aortic wall.⁵⁴ A histology study of aneurysmal samples from LDS patients showed enhanced accumulation of collagen fiber in the media^{12, 31}. Our finding showed that the loss of function of SMAD3 could increase collagen fiber deposition and cause enhanced structure stiffness in CPC-VSMCs tissue rings. (Figure 4) This result is consistent with clinical findings, indicating increasing collagen and stiffness in the aortic root in patients with SMAD3 mutation could be a cause of root aneurysm.

Our study was limited by the in vitro system of human stem cell differentiation. The SMAD3 deficiency model in our study did not completely reflect the pathogenic effect of SMAD3 heterozygous mutation, as we discovered in patients. However, this cellular model provides significant knowledge of how TGF-β signaling affect SMC differentiation in different lineages.

In summary, using hiPSCs and SMAD3 disruption with CRISPR/Cas9, we demonstrated that the SMAD3-dependent TGF-β signaling was essential to the differentiation of CPC-VSMCs, which populate the aortic root and part of the ascending aorta, but not NCSC-VSMCs, which mainly populate the ascending aorta, aortic arch and root. Our findings provide insights into lineage-specific TGF-β responses in human VSMCs, which may potentially contribute to the development of monogenetic syndromic aortic root aneurysms in patients with mutations in the TGF-β pathway, such as mutations in SMAD3.

Highlight:

1. Pathogenic effect of SMAD3 deficiency was studied separately in two different VSMC lineages: CPC-VSMCs and NCSC-VSMCs.

2. CPC-VSMC differentiation is TGF-β canonical pathway (SMAD3 pathway) dependent but NCSC-VSMC differentiation is independent of TGF-β canonical pathway.

3. SMAD3 deficiency decreased elastin synthesis in both CPC-VSMCs and NCSCs-VSMCs, but increased collagen in the CPC-VSMCs.

4. After canonical pathway (SMAD3 pathway) was obliterated due to SMAD3 deficiency, only NCSC-VSMCs had over-activated TGF-β signaling, but CPC-VSMCs had impaired TGF-β signaling.

Abbreviations:

VSMC

vascular smooth muscle cell

hiPSC

human induced pluripotent stem cell

TAA

thoracic aortic aneurysm

NCSC

neural crest stem cell

CPC

cardiovascular progenitor cell

TGF-β

transforming growth factor-β

ECM

extracellular matrix

LDS

Loeys-Dietz syndrome