Ubiquitination of Smad2 by Smurf1 inhibits endothelial-to-mesenchymal transition in human coronary artery endothelial cells Authors: Panpan Liu, Mingyang Zhang, Shuhui Wang, Zhiheng Liu, Nana Wang, Zhiyuan Liu, Yiyi Shen, Jiaying Zhang, Jing Li, Ying Liu, Haitao Lv, Guanghui Qian Published: 2026 ABSTRACT Smad2 is a well-established regulator involved in tissue development and in the pathogenesis of endothelial-to-mesenchymal transition (EndMT) mediated by TGF-β signaling. However, the mechanism underlying the regulation of SMAD2 in human coronary artery endothelial cells (HCAECs), particularly the identity of the responsible E3 ubiquitin ligase and its role during EndMT remain unclear. In this study, we identified Smad ubiquitination regulatory factor 1 (Smurf1) as a negative regulator of Smad2 protein levels in HCAECs and demonstrated that the E3 ligase activity of Smurf1 is essential for this function. Mechanistically, Smurf1 interacts with Smad2, promoting its ubiquitination, and subsequent proteasomal degradation. Specifically, Smurf1 catalyzes K48-linked polyubiquitination of Smad2 at lysine residues K156, K383 and K420. Functionally, Smad2 was found to promote EndMT in HCAECs, an effect that was partially attenuated either by co-expression of Smurf1 or by mutation of Smad2 at lysine 420 (Smad2-K420R), which replaces lysine with arginine. Taken together, our findings identify, for the first time, specific lysine residues on Smad2 targeted by Smurf1 for K48-linked ubiquitination and highlight their crucial regulatory role in modulating EndMT in HCAECs.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-55315-0. Introduction Endothelial-to-mesenchymal transition (EndMT) is a biological process involved in both normal physiological development and the progression of various diseases, and it is associated with the loss of endothelial barrier integrity1. Initially, EndMT was believed to play a critical role in embryogenesis, particularly during cardiogenesis2,3. However, subsequent studies have highlighted its pathological involvement in multiple disease states4–6. During EndMT, endothelial cells undergo significant alterations in morphology, function, and gene expression profiles7. This transition is characterized by the upregulation of mesenchymal markers such as vimentin, α-smooth muscle actin (α-SMA), Fn1, n-cadherin, and Snail, and endothelial markers such as von Willebrand factor (vWF), vascular endothelial cadherin (VE-cadherin), and CD31 are decreased7. These molecular changes lead to enhanced cellular motility and invasiveness8. Morphologically, endothelial cells transition from a tightly-adherent, cobblestone-like monolayer to an elongated, spindle-shaped mesenchymal phenotype with disrupted intercellular junctions9. EndMT is mainly triggered by transforming growth factor β (TGF-β) signaling, which operated through both Smad-dependent and Smad-independent pathways10,11. Among the Smad family proteins, Smad2 constitutes a central mechanistic node in mediating TGF-β-induced EndMT12. Given the pathophysiological relevance of EndMT, therapeutic strategies targeting this process—or its reversal, mesenchymal-to-endothelial transition (MET)—may offer promising avenues to restore endothelial function in disease contexts. Ubiquitin-mediated proteolysis is a fundamental post-translational regulatory mechanism that governs a wide array of physiological processes. The ubiquitination cascade involves three sequential enzymatic components: E1 ubiquitin-activating enzymes, E2 ubiquitin-conjugating enzymes, and E3 ubiquitin ligases13. Among the E3 ligases, the Smad ubiquitination regulatory factor (Smurf) family-comprising evolutionarily conserved HECT-domain proteins-plays a critical role in modulating TGF-β and BMP signaling pathways14. Smurf1, a representative HECT-type ubiquitin ligase, has been shown to regulate multiple signaling pathways, including TGF-β1, BMP, MAPK and TLR pathways15. Within the canonical TGF-β/BMP receptor superfamily, ligand binding leads to the phosphorylation and activation of receptor-regulated Smad proteins (R-Smads). Specifically, Smad1, Smad5, and Smad8 are mainly involved in BMP signaling, whereas Smad2 and Smad3 mediate canonical TGF-β-mediated signaling16. Accumulating evidence indicates that Smurf1 preferentially targets Smad1/5, while Smurf2 predominantly regulates Smad2/317. Although our previous study identified Smad2 as a key regulator of EndMT and found its modulation by the deubiquitinating enzyme USP718, it remains unclear whether and how Smad2 is regulated by E3 ligase during EndMT process. In this study, we investigated the expression and functional relationship between Smad2 and Smurf1 during EndMT. We found that Smurf1 inhibits EndMT by destabilization Smad2 protein. Mechanistically, Smurf1 interacts with Smad2 and promotes its degradation through enhanced K48-linked polyubiquitination at lysine 420. This post-translation modifications leads to a reduced Smad2 protein levels and subsequent suppression of EndMT in human coronary artery endothelial cells (HCAECs). These findings provide novel insights into the post-translational regulation of Smad2 by Smurf1 and highlight a previously unrecognized role of Smurf1 in modulating TGF-β-induced EndMT. Material and methods Cell culture This study utilized two distinct cell models: human embryonic kidney 293 T cells (HEK293T) and HCAECs (Shanghai Biological Cell Bank, Shanghai, China). Both cell types were maintained in Dulbecco’s Modified Eagle Medium (DMEM; HyClone) supplemented with 10% fetal bovine serum (FBS, Gibco #10270–106) and 1% penicillin–streptomycin under standard culture conditions (37 °C, 5% CO₂, humidified atmosphere). Typically, 1 mL of a 100 × liquid formulation (Beyotime, C0222) was added to 100 mL of medium, yielding a final concentration of 100 I.U./mL penicillin and 100 µg/mL streptomycin. For experimental interventions, cellular stimulation was achieved using recombinant human TGF-β2 (Proteintech #HZ-1092) at specified concentrations and durations as detailed in subsequent sections. Cell transfection and plasmids For Smurf1 knockdown in cells, we employed both small interfering RNAs (siRNAs, si-Smurf1; GenePharma) and a CRISPR/sgRNA plasmid (pLenti-Smurf1-sgRNA, L10610; Beyotime, Shanghai, China). To overexpress target genes, we constructed pcDNA3.1-based expression vectors containing full-length coding sequences of Smurf1, Smad2, ubiquitin (Ub), or K48-linked ubiquitin (K48), with pcDNA3.1 vector serving as negative control. HA-tagged ubiquitin and Myc-tagged wild-type Smurf1 (Dr. Lingqiang Zhang, State Key Laboratory of Proteomics); Flag-tagged wild-type Smurf1 (Dr. Chengjiang Gao, Shandong University); as well as HA-tagged K48 ubiquitinatin (Prof. Zheng Hui, Soochow University). Flag-tagged Smad2, and Myc-tagged Smad2 (KeyGEN, Nanjing, China). A Smurf1 mutant (C699A) plasmid with no catalytic activity, and Smad2 mutant plasmids (K63R, K19R, K46R, K420R, K13R, K156R) were respectively created by using Quick Change Site-directed Mutation Kit (Agilent Technology). By using the K156R mutant plasmid as the template, we introduced a second mutation at position 383 (lysine to arginine, K383R) using mutation-specific primers and the Quick Change Site-Directed Mutagenesis Kit (Agilent Technologies), thereby generating the SMAD2 K156/383R double mutant plasmid. All transfections were performed following manufacturers’ protocols, using jetPRIME (101000046, Polyplus-transfection SA) for siRNA delivery and LongTrans (TF07, UcallM, Wuxi, China) for plasmid transfection. RNA isolation and qRT-PCR quantification TRIzol reagent (Invitrogen, USA) was used to extract total cellular RNA. Following isolation, RNA samples underwent DNase I treatment (RNase-free) to remove potential genomic DNA contamination. According to the instructions of commercial reverse transcription kit (Takara, Shiga, Japan), DNA (cDNA) complementary to the isolated RNA was synthesized. Gene expression analysis was then conducted through quantitative real-time PCR (qRT-PCR) employing SYBR Green chemistry (Selleck Chemicals, Houston, TX, USA) on a Light Cycler 480 system (Roche Diagnostics, Basel, Switzerland). The housekeeping gene gapdh or actin was used for normalization of all qRT-PCR data. All quantitative PCR reactions were conducted in technical triplicates to ensure experimental reproducibility and data reliability. The following primer sequences were used for amplification: Smurf1(5’-CACAAGCCATTTTCTTTGCT-3’ &5’-ACTTGGCTGCATATCGAAAG-3’), Smad2 (5’-TGGGGAAGTTTTTGCCGAGT-3’ & 5’-ACCGTCTGCCTTCGGTATTC-3’). Gapdh (5’-GGAGTCCACTGGCGTCTTCA-3’ & (5’-GTCATGAGTCCTTCCACGATACC-3’), Actin (5’-CTCTTCCAGCCTTCCTTCCT- -3’ & 5’-CAGGGCAGTGATCTCCTTCT-3’), Fn1(5’-GCACCACCCCAGACATTACT-3’& 5’-CGGGACTCAGGTTATCAAAAGTG-3’) Snail (5’-AAGATGCACATCCGAAGCCA-3’& 5’-CATTCGGGAGAAGGTCCGAG-3’), VE-cadherin(5’-CAGATCTCCGCAATAGACAAGGA-3’& 5’-ACTGTGATGTTGGCCGTGTTATC-3’), CD31(5’-GTCGGACAGTGGGACGTATATC-3’& 5’-ACCTCAAACTGGGCATCATAAGA-3’). Immunoblotting (IB) and immunoprecipitation (IP) Cellular proteins were isolated using RIPA buffer (Beyotime, Shanghai, China) or NP-40 lysis buffer (Biosharp, China), both add protease inhibitors and phosphatase inhibitors (Roche, Basel, Switzerland). For ubiquitination studies, lysis buffer was additionally supplemented with 10 mM N-ethylmaleimide (NEM) to preserve ubiquitin conjugates. The biocinchoninic acid (BCA) assay kit (Beyotime, Shanghai, China) was used to determine protein concentration. Extracted proteins were immunoprecipitated with bead-conjugated Flag antibody (Selleck Chemicals, USA), anti-Myc magnetic beads (Selleck Chemicals, USA), or protein A/G beads (Merck Millipore) conjugated with Smad2 antibody in a rotating incubator for 2 h at 4 °C. Following immunoprecipitation, protein-bound beads were subjected to four consecutive washes with co-IP lysis buffer. Then add laemmli buffer containing 5% β-mercaptoethanol to the protein complex and boil at 100 °C for 10 min. The proteins were then separated by SDS-PAGE using either 7.5% or 10% resolving gels (determined by target protein molecular weight) and transferred onto PVDF membranes (Bio-Rad Laboratories, Hercules, CA) using wet transfer conditions. Then, membranes were blocked with 5% non-fat milk in TBST (Tris-buffered saline with 0.1% Tween-20) for 1 h at room temperature to prevent non-specific binding. Primary antibody incubations were carried out overnight at 4 °C with gentle agitation to ensure optimal antigen–antibody interactions. The following primary antibodies were employed for immunoprecipitation and immunoblotting experiments: anti-Smurf1 (Santa Cruz Biotechnology, 4833S; 1:500), anti-Flag (Sigma-Aldrich, F1804; 1:5000), anti-Myc (Abbkine Scientific, A02060; 1:1000), anti-Smad2 (Cell Signaling Technology, 5339; 1:1000), anti-HA (Santa Cruz Biotechnology, sc-7392; 1:2000), anti-ubiquitin (Santa Cruz Biotechnology, sc-8017; 1:1000), anti-GAPDH (Cell Signaling Technology, 188S, 1:5000), β-actin (Proteintech, #78; 1:5000), anti-FNI (Abcam, Cambridge, MA, USA, AB268020, 1:1000), anti-Snail1 (Cell Signaling Technology, 3879S; 1:1000), and anti-N-cadherin (Santa Cruz Biotechnology, sc-393933, 1:500). Membranes were incubated with secondary antibodies for 60 min at 25 °C with gentle agitation, after primary antibody incubation. Exposure of target proteins using commercially available enhanced chemiluminescence (ECL) substrate systems. Immunofluorescence (IF) assay Place sterile glass coverslips in a 12-well plate and seed the cells at 5 × 103 cells/well. After relevant transfection of the cells, the medium was discarded, and the cells were gently washed 3 times with PBS, then 1 ml of 4% paraformaldehyde was added to fix for 30 min, permeabilized with 0.5% Triton X-100 for 30 min at room temperature, and blocked with 5% BSA solution for 1 h. Subsequently, incubate cells with anti-Smurf1 and Smad2 primary antibodies overnight and then with species-matched fluorescent secondary antibodies (1:500 dilution, 1 h). Counterstain the nuclei with DAPI, and images were using an Olympus microscope (Tokyo, Japan) under standardized exposure conditions to ensure consistency across sample. Cell migration assays A scratch wound assay was used to assess the migratory capacity of HCAECs. Inoculate HCAECs using 6-well tissue culture plates and allowed to reach full confluency. After this, change the complete medium to serum-free medium. Using a sterile 200 μl pipette tip, a consistent scratch was introduced into each monolayer. Following incubation, Non-adherent cells are removed using a gentle wash with PBS, then replenished with serum-free medium. Wound closure was monitored and photographed at the same predetermined locations at 0, 6, 12, and 24 h post-wounding monitored by phase-contrast microscopy (Olympus, Tokyo, Japan). The relative wound closure rate was calculated by calculating the percentage reduction in the wound area at specified intervals relative to the baseline measurement (0 h), with ImageJ software used for analysis. Cycloheximide chase assay To assess Smad2 degradation kinetics, transfected cells were analyzed under controlled conditions. Plated cells were transfected upon reaching 60% confluency. Following a 48-h transfection, cycloheximide (CHX; 10 μg/ml, Sigma) was added to block new protein synthesis, with subsequent sample collection at predetermined intervals (0, 4, and 8 h). Cellular proteins were then extracted and subjected to immunoblotting. To validate experimental reliability, three independent biological replicates were conducted for each condition. The protein degradation rate was calculated by quantifying Smad2 band intensity using ImageJ software (National Institutes of Health, USA, https://downloads.micron.ox.ac.uk/fiji_update/mirrors/fiji-latest/fiji-latest-win64-jdk.zip). Statistical analysis To ensure reproducibility, three independent biological replicates were performed for all experiments, each containing triplicate technical samples. Semi-quantitative analysis was conducted using ImageJ software, followed by statistical processing with GraphPad Prism 8.0 (GraphPad, USA, https://app.graphpad.com/) and SPSS 26.0 software (IBM SPSS Statistics, USA, https://souurl.cn/NhIGed). Data are presented as mean ± SEM. Intergroup comparisons were analyzed by one-way ANOVA and multiple comparisons use Tukey’s post-hoc test. ns, p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001. Statement All experimental protocols were approved by institute of pediatric research, Children’s Hospital of Soochow University, Suzhou, Jiangsu Province, China. We confirm that all methods were performed in accordance with the relevant guidelines and regulations. Results Smurf1 represses the expression of Smad2 in protein level Our results demonstrate that overexpression of Smurf1 dose-dependently reduced levels of endogenous and exogenous Smad2 protein (Fig. 1A and B), while knockdown of endogenous Smurf1 by siRNA caused Smad2 protein level increased significantly in HCAECs (Fig. 1C). Notably, Smurf1 did not alter Smad2 mRNA levels in HCAECs (Fig. 1D and E). To further confirm that Smurf1 regulates the stability of Smad2, HCAECs were treated with cycloheximide, a protein synthesis inhibitor and the degradation rate was examined. The results showed that transfection of Smurf1 accelerate the degradation rate of Smad2 protein (Fig. 1F and G). Consistently, depletion of endogenous Smurf1 with siRNA attenuated the degradation rate of Smad2 protein (Fig. 1H and I). These results together demonstrate that Smurf1 mediates time- and dose-dependent proteasomal degradation of Smad2 proteins, and the expression level of Smurf1 is negatively correlated with the Smad2 protein level and degradation. Fig. 1Smurf1 decrease the turnover and stability of Smad2proteins. (A) Overexpression of Smurf1 dose-dependently resulted in reduced levels of endogenous Smad2 protein. (B) Exogenous Smad2 levels decreased as Myc-Smurf1 concentrations increased in 293 T cells. (C) Smad2 immunoblot in HCAECs following siRNA-mediated knockdown of endogenous Smurf1. (D,E) qRT-PCR quantification of Smurf1 and Smad2 mRNA levels in HCAECs expressing Myc-Smurf1. (F) Myc-tagged Smad2 was transiently expressed in 293 T cells either alone or in combination with Myc-Smurf1. Following transfection, by using CHX to assess protein degradation for varying time intervals. Protein expression patterns were subsequently evaluated through immunoblot analysis. (G) Densitometric quantification (ImageJ) of Smad2 protein levels in (F). (H) CHX chase assay of endogenous Smad2 in 293 T cells transfected with control (si-NC) or Smurf1-targeting siRNA (si-Smurf1). (I) Quantification (ImageJ) of Smad2 levels in (H) All values are expressed as mean ± SEM, n = 3, nsP > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001. Smurf1 decrease the turnover and stability of Smad2proteins. (A) Overexpression of Smurf1 dose-dependently resulted in reduced levels of endogenous Smad2 protein. (B) Exogenous Smad2 levels decreased as Myc-Smurf1 concentrations increased in 293 T cells. (C) Smad2 immunoblot in HCAECs following siRNA-mediated knockdown of endogenous Smurf1. (D,E) qRT-PCR quantification of Smurf1 and Smad2 mRNA levels in HCAECs expressing Myc-Smurf1. (F) Myc-tagged Smad2 was transiently expressed in 293 T cells either alone or in combination with Myc-Smurf1. Following transfection, by using CHX to assess protein degradation for varying time intervals. Protein expression patterns were subsequently evaluated through immunoblot analysis. (G) Densitometric quantification (ImageJ) of Smad2 protein levels in (F). (H) CHX chase assay of endogenous Smad2 in 293 T cells transfected with control (si-NC) or Smurf1-targeting siRNA (si-Smurf1). (I) Quantification (ImageJ) of Smad2 levels in (H) All values are expressed as mean ± SEM, n = 3, nsP > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001. Smurf1 interacts with Smad2 To investigate whether Smurf1 interacts with Smad2, coimmunoprecipitation assays were performed. 293 T cells were co-transfected with Flag-Smad2 and Myc-Smurf1 constructs. Subsequent co-immunoprecipitation experiments using anti-Myc antibody successfully pulled down the Smurf1 complex, demonstrating specific binding to Flag-Smad2 (Fig. 2A). Reciprocal coimmunoprecipitation experiments with anti-Flag antibody yielded consistent results (Fig. 2B). To further validate these findings,Fig. 2Smurf1 interacts with Smad2. (A) 2Myc-Smurf1 and Flag-Smad2 expression vectors were used for transient transfection of 293 T cells. Following lysis, protein was immunoprecipitated using anti-Myc magnetic beads and analyzed with Flag antibody. (B) Co-immunoprecipitation was used in 293 T cells expressing both Myc-Smurf1 and Flag-Smad2. Cell extracts were precipitated with bead-conjugated Flag antibody and immunoblotting with anti-Myc antibody. (C) HCAEC transfected with Myc-Smurf1 were treated with or without TGF-β2. Immunoprecipitation was performed with anti-Myc antibody, and the binding of Smad2 to Smurf1 was detected by IB. Input was used to confirm the protein expression in each group, with β-actin as the loading control. (D) HCAEC transfected with Myc-Smurf1 were treated with or without TGF-β2. Immunoprecipitation was performed with anti-Myc antibody, and the binding of Smad1 to Smurf1 was detected by IB. (E) HCAEC were transfected with si-Smurf1 or control si-NC. Cells were immunostained for Smad2 (green), Smurf1 (red), and nuclei (DAPI, blue). Confocal microscopy was used to observe the subcellular distribution of the proteins. Scale bars = 20 μm. Smurf1 interacts with Smad2. (A) 2Myc-Smurf1 and Flag-Smad2 expression vectors were used for transient transfection of 293 T cells. Following lysis, protein was immunoprecipitated using anti-Myc magnetic beads and analyzed with Flag antibody. (B) Co-immunoprecipitation was used in 293 T cells expressing both Myc-Smurf1 and Flag-Smad2. Cell extracts were precipitated with bead-conjugated Flag antibody and immunoblotting with anti-Myc antibody. (C) HCAEC transfected with Myc-Smurf1 were treated with or without TGF-β2. Immunoprecipitation was performed with anti-Myc antibody, and the binding of Smad2 to Smurf1 was detected by IB. Input was used to confirm the protein expression in each group, with β-actin as the loading control. (D) HCAEC transfected with Myc-Smurf1 were treated with or without TGF-β2. Immunoprecipitation was performed with anti-Myc antibody, and the binding of Smad1 to Smurf1 was detected by IB. (E) HCAEC were transfected with si-Smurf1 or control si-NC. Cells were immunostained for Smad2 (green), Smurf1 (red), and nuclei (DAPI, blue). Confocal microscopy was used to observe the subcellular distribution of the proteins. Scale bars = 20 μm. we examined potential interactions between endogenous Smad2 and Smurf1 proteins, and we found that endogenous Smad2 physiologically interacted with My-Smurf1 in HCAECs, indicating an interaction between Smad2 and Smurf1, and interestingly, this interaction is decreased under TGF-β2 treatment (Fig. 2C). As a positive control, Myc-Smurf1 also interacts with Smad1 (Fig. 2D). Furthermore, immunofluorescence staining revealed a co-localization between Smurf1 and Smad2 proteins in HCAECs (Fig. 2E), demonstrating a colocalization of the two proteins. These findings collectively establish that Smurf1 interacts with Smad2 at the cellular level. Smurf1 ubiquitylates Smad2 Given Smurf1 is a HECT-type ubiquitin ligase, we next questioned whether Smurf1 affects the ubiquitination level of Smad2. We overexpressed Flag-Smad2 and HA-ub together with or without Myc-Smurf1 in cells. The data showed that Smurf1 overexpression causes both endogenous and exogenous ubiquitination levels of Smad2 increased in 293 T cells (Fig. 3A,B). Consistently, Smurf1 overexpression causes an increased polyubiquitin level of SMAD2 (Fig. 3C,D). Furthermore, to clarify whether the E3 enzyme activity of Smurf1was required for its role in upregulating the ubiquitination level of Smad2, we generated an enzymatically inactive mutant by replacing 699 cysteine with alanine (Smurf1-C699A). The results showed that the catalytically inactive mutant Smurf1-C699A failed to increase the ubiquitination level of Smad2, demonstrating that the ubiquitin ligase activity of Smurf1 was required for down-regulating Smad2 protein level (Fig. 3E). This result was further corroborated by the sgRNA-mediated knockout of endogenous Smurf1, which caused a significantly decreased polyubiquitination level of Smad2 (Fig. 3F). Together, these results establish Smad2 as a substrate of Smurf1 and demonstrate the essential role of Smurf1’s E3 ligase activity in mediating this post-translational modification.Fig. 3Smurf1 ubiquitylates Smad2. (A,B) Myc-tagged Smurf1 and Flag-tagged Smad2 were used to co-transfect 293 T cells, either in the presence or absence of an HA-ubiquitin. Ubiquitination was detected by immunoblotting using HA and ubiquitin antibodies. (C,D) Endogenous Smad2 was immunoprecipitated after transfection with Flag-tagged Smurf1 in 293 T cells with or without HA‑ubiquitin co-expression; ubiquitination of precipitated Smad2 was detected by immunoblotting using anti-HA and anti-ubiquitin antibodies. (E) Expression of wild type smurf1 increased ubiquitination of Smad2 in 293 T cells, while the catalytically ineffective C699a mutant had no effect on ubiquitination of Smad2. (F) Depletion of endogenous Smurf1 reduced ubiquitination levels of Smad2 in HCAECs. Smurf1 ubiquitylates Smad2. (A,B) Myc-tagged Smurf1 and Flag-tagged Smad2 were used to co-transfect 293 T cells, either in the presence or absence of an HA-ubiquitin. Ubiquitination was detected by immunoblotting using HA and ubiquitin antibodies. (C,D) Endogenous Smad2 was immunoprecipitated after transfection with Flag-tagged Smurf1 in 293 T cells with or without HA‑ubiquitin co-expression; ubiquitination of precipitated Smad2 was detected by immunoblotting using anti-HA and anti-ubiquitin antibodies. (E) Expression of wild type smurf1 increased ubiquitination of Smad2 in 293 T cells, while the catalytically ineffective C699a mutant had no effect on ubiquitination of Smad2. (F) Depletion of endogenous Smurf1 reduced ubiquitination levels of Smad2 in HCAECs. Smurf1 catalyzes K48-linked ubiquitination of Smad2 Based on the results discussed above, we next questioned what type of ubiquitin signals of Smad2 is regulated by Smurf1. Because K48-linked polyubiquitination is recognized as the degradation signal, targeting substrates for proteasomal destruction through specific recognition by the 26S proteasome complex. To test this hypothesis, we co-transfected Smad2 and Smurf1 plasmids into 293 T cells to examine the endogenous K48 polyubiquitin levels of Smad2. As shown in Fig. 4A, Smurf1 overexpression significantly increases the endogenous K48 polyubiquitin levels of Smad2. Additionally, we also generated a mutant HA-K48R plasmid by replacing all the lysine residue with arginine except for the lysine 48 and transfected this mutant plasmid with or without Myc-Smurf1 and Flag-Smad2 into 293 T cells. The results showed that Smurf1 overexpression also caused exogenous K48 polyubiquitin levels of Smad2 increased significantly (Fig. 4B). Consistently, Smurf1 knockdown caused a decreased K48-linked polyubiquitination level of Smad2 proteins (Fig. 4C).Fig. 4Smurf1 catalyzes K48-linked ubiquitination of Smad2. (A) Endogenous Smad2 ubiquitination in 293 T cells overexpressing Smurf1, detected using K48-specific ubiquitin antibodies. (B) 293 T cells co-transfected with Flag-Smad2, Myc-Smurf1 and HA-K48 ubiquitin. Immunoprecipitates were immunoblotted with HA antibodies. (C) siRNA-mediated Smurf1 knockdown reduces K48-polyubiquitinated Smad2 levels in 293 T cells. (D) Transfections included wild-type Smad2 or the following lysine-to-arginine mutants: K19R, K13R, K46R, K63R, K156R, K420R, the double mutant K156R/K383R, and either Flag-tagged Smurf1 or empty vector control. (E) 293 T cells were transfected with wild-type Smad2, Smad2 K420R, Smad2 K156R, or Smad2 K156R/K383R and HA—ubiquitin, in combination with either Flag-Smurf1 or an empty vector control. The isolated proteins were analyzed using an antibody targeting HA to assess ubiquitin modification. (F) 293 T cells were transfected with wild-type Smad2, Smad2K420R, Smad2K156R, or Smad2K156R/K383R, together with Flag‑Smurf1 or empty vector control. Endogenous ubiquitination of the isolated proteins was detected by immunoblotting with an anti‑ubiquitin antibody. Smurf1 catalyzes K48-linked ubiquitination of Smad2. (A) Endogenous Smad2 ubiquitination in 293 T cells overexpressing Smurf1, detected using K48-specific ubiquitin antibodies. (B) 293 T cells co-transfected with Flag-Smad2, Myc-Smurf1 and HA-K48 ubiquitin. Immunoprecipitates were immunoblotted with HA antibodies. (C) siRNA-mediated Smurf1 knockdown reduces K48-polyubiquitinated Smad2 levels in 293 T cells. (D) Transfections included wild-type Smad2 or the following lysine-to-arginine mutants: K19R, K13R, K46R, K63R, K156R, K420R, the double mutant K156R/K383R, and either Flag-tagged Smurf1 or empty vector control. (E) 293 T cells were transfected with wild-type Smad2, Smad2 K420R, Smad2 K156R, or Smad2 K156R/K383R and HA—ubiquitin, in combination with either Flag-Smurf1 or an empty vector control. The isolated proteins were analyzed using an antibody targeting HA to assess ubiquitin modification. (F) 293 T cells were transfected with wild-type Smad2, Smad2K420R, Smad2K156R, or Smad2K156R/K383R, together with Flag‑Smurf1 or empty vector control. Endogenous ubiquitination of the isolated proteins was detected by immunoblotting with an anti‑ubiquitin antibody. To further identify which lysine residue is essential for Smad2 ubiquitination, we generated seven mutants including K13R, K19R, K46R, K56R, K63R, K156R and K383R. To screen the specific lysine residue, we respectively transfected these mutant Smad2 plasmid or wild-type Smad2 plasmid together with Flag-smurf1 plasmid, then we detected the protein level changes of Smad2 proteins in each group. The results showed that Smurf1 overexpression causes a decreased level of Smad2 except for the Smad2 K420R, Smad2 K156R, and Smad2 K156/383R plasmid (Fig. 4D), suggesting these lysine residues is required for the ubiquitin modification mediated by Smurf1. Next, cells were transfected with the wild-type Smad2 or the mutant including Smad2K420R, -Smad2K156R and -Smad2K156R/K183R, together with or without Smurf1, the results showed that all these mutants had lower levels of ubiquitination compared with the wild-type Smad2 plasmid (Fig. 4E,F), suggesting that the three lysine residues, especially Smad2K420R are the ubiquitin acceptor residues of Smad2. Taken together, our results demonstrate that Smurf1 mediates K48-linked polyubiquitination of Smad2, and lysine 420 is essential for Smad2 ubiquitination. Smurf1 antagonizes EndMT by decreasing the expression level of Smad2 protein During EndMT, cells undergo profound functional and transcriptional changes, characterized by the upregulation of mesenchymal marker proteins and a concurrent downregulation of endothelial markers. These alterations enhance the migratory and invasive capacities of the cells8. Based on this rationale, we next investigated whether Smurf1 and Smad2 regulate EndMT in HCAECs. The plasmids of WT Smad2, Smad2K420R and Flag-Smurf1 or empty vector plasmid were respectively used to transfect HCAECs. And the scratch test showed that the over-expression of WT-Smad2 or the co-expression of Smurf1 and Smad2k420R could increase the cell migration, while the co-expression of Smurf1 or Smurf1 and WT-Smad2 could reduce the cell migration (Fig. 5A and B). To examine the molecular shifts underlying the morphological changes, we extracted the proteins and conducted immunoblotting assay. The results showed that over-expression of WT-Smad2 or the co-expression of Smurf1 and Smad2K420R caused an increased expression level of FN1 N-cadherin and Snail. Interestingly, the co-expression of Smurf1 or Smurf1 and WT-Smad2 caused the expression level of FN1 N-cadherin and Snail downregulated, suggesting that Smad2-WT promoted the increase of FN1 and snail expression, and slightly up-regulated N-cadherin, however, when Smad2 and Smurf1 were co-transfected, FN1 N-cadherin and Snail were down-regulated to varying degrees, and this down-regulation trend was partially reversed by Smad2K420R mutants (Fig. 5C). Furthermore, Smad2 overexpression increased the expression levels of Fn1 and Snail while decreasing those of VE-cadherin and CD31, suggesting the promotion of EndoMT. Interestingly, co-overexpression of Smurf1 with either wild-type or mutant Smad2 partially reversed this EndoMT phenotype (Fig. 5D–G).Fig. 5Smurf1 inhibits EndMT by decreasing Smad2 protein. (A,B) The migration ability of HCAECs was measured by wound healing test. Scale bars = 200 μm. (C) The expression levels of Fn1, Snail, and N-cadherin in HCAECs were assessed via western blot following transfection with plasmids encoding WT Smad2, Smad2K420R, and either Flag-Smurfl or an empty vector. (D–G) The mRNA expression levels of Fn1, Snail, CD31 and VE-cadherin in HCAECs were detected by qRT-PCR after transfection with plasmids encoding WT Smad2, Smad2K420R, and either Flag-Smurfl or an empty vector. β-actin was used as an internal reference to normalize the mRNA expression levels. All values are expressed as mean ± SEM, n = 3 nsP > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Smurf1 inhibits EndMT by decreasing Smad2 protein. (A,B) The migration ability of HCAECs was measured by wound healing test. Scale bars = 200 μm. (C) The expression levels of Fn1, Snail, and N-cadherin in HCAECs were assessed via western blot following transfection with plasmids encoding WT Smad2, Smad2K420R, and either Flag-Smurfl or an empty vector. (D–G) The mRNA expression levels of Fn1, Snail, CD31 and VE-cadherin in HCAECs were detected by qRT-PCR after transfection with plasmids encoding WT Smad2, Smad2K420R, and either Flag-Smurfl or an empty vector. β-actin was used as an internal reference to normalize the mRNA expression levels. All values are expressed as mean ± SEM, n = 3 nsP > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Discussion Firstly, the Smurf family was identified as key regulators of TGF-β/BMP signal transduction. Early studies established their ability to bind and ubiquitinate receptor-regulated Smads and the TGF-β type I receptor (TβRI), targeting these proteins for proteasomal degradation19. Beyond their canonical role in TGF-β/BMP regulation, subsequent research has revealed that Smurfs participate in diverse cellular processes. Notably, multiple lines of evidence indicate that Smurf1/2 regulate fundamental biological functions including cell cycle progression and migratory behavior through TGF-β/BMP-independent mechanisms20. The Smurf family is a key regulator of protein homeostasis, controlling the stability of diverse substrates to modulate essential cellular processes including migration, proliferation, and programmed cell death. Abnormal cell physiology and imbalances in cell homeostasis, often linked to disruptions in these processes, contribute to the development of various diseases21. Smurf1 exhibits complex regulatory functions in TGF-β/BMP signaling cascades, early studies focused predominantly on its E3 ubiquitin ligase activity targeting Smad1/5 for proteasomal degradation, thereby blocking intracellular BMP signals through ubiquitination and proteasomal degradation19. However, subsequent research has revealed that Smurf1 interactions extend beyond these targets. For instance, Multiple studies have demonstrated that Smurf1 promotes Smad7 ubiquitination and facilitates its cytoplasmic retention. Despite these findings, the regulatory mechanisms of Smurf1 on Smad2 remain unclear and somewhat ambiguous. While Smurf1 significantly affects Smad1 and Smad5 levels, its impact on Smad2 is less pronounced. Specifically, Smurf1 only affects Smad2 protein levels at the highest levels of Smurf1 expression19. Previous studies also have contradictory conclusion. Notably, Smurf1 expression failed to produce any detectable changes in total Smad2/3 abundance or their activated phosphorylated forms (pSmad2/3) in the lens epithelial cell system, further underscoring the selective nature of Smurf1’s regulatory role16. Smurf1 significantly decreases endogenous Smad5 protein levels in C2C12 cells, while having no detectable effect on Smad2, Smad3, or Smad722. In experimental conditions where Smurf1 expression showed minimal impact on Smad2 and Smad3 protein levels, particularly in estrogen receptors α (ERα)-deficient contexts23. Additionally, in kidney disease, hypoxia can lead to the increased protein levels of TGF-β1, Smurf1, and Smad2/324. When Smad2 is co-expressed with Smurf1, higher degrees of Smad4 downregulation are observed25. Smurf1 is a key molecule in TGF-β/BMP signaling and other cellular processes, with its dysfunction linked to critical physiological and pathological outcomes. While its effects on Smad1, Smad5, and Smad7 are well-documented, its regulatory impact on Smad2 remains less clear. Future studies should systematically examine Smurf1-mediated Smad2 regulation and its dual roles in homeostasis and disease pathogenesis. Based on the above research objectives, we further explored its mechanism in HCAEC and found that Smurf1 regulates Smad2 levels in a dose-dependent manner, with overexpression of Smurf1 reducing Smad2 levels and knockdown increasing them. Smurf1 shortened the half-life of Smad2, promoting its degradation, while Smurf1 depletion prolonged Smad2 stability. The physical interaction between Smurf1 and Smad2 was consistently demonstrated through complementary approaches: (1) co-immunoprecipitation assays revealed molecular association in protein complexes, and (2) immunofluorescence microscopies confirmed their spatial co-localization in cellular compartments. Furthermore, Smurf1 catalyzed K48-linked ubiquitination of Smad2, targeting it for proteasomal degradation, with lysine 420 identified as the primary ubiquitination site. The E3 ubiquitin ligase function of Smurf1 proved critical for mediating this biological process, as an enzymatically inactive mutant (Smurf1 C699A) failed to ubiquitinate Smad2. The TGF-β/Smad signaling pathway represents a pleiotropic regulatory system that plays critical roles in both physiological homeostasis and pathological processes within the cardiovascular system. Such as myocardial infarction (AMI), Kawasaki disease (KD), pulmonary hypertension, atherosclerosis have been found to have pathological processes of EndMT26,27. For example, Ubiquitin-specific protease 7 (USP7) promotes EndMT in HCAECs through deubiquitination-mediated stabilization of Smad2 and Smad3 proteins18. Apolipoprotein A-I (ApoA-I) exerts a protective effect on endothelial function by suppressing EndMT in HCAECs, decreasing phosphorylated Smad3 and Smad228. Consistent with these findings, Ghrelin suppresses TGF-β1-mediated EndMT in HCAECs. This inhibitory effect occurs through the GHSR-1a/AMPK signaling axis, which upregulates Smad7 expression, thereby blocking Smad2/3 phosphorylation29. This investigation presents the initial experimental confirmation that Smurf1 mediates K48-linked ubiquitination of Smad2 at lysine 420 to restrain EndMT in HCAECs, as evidenced by decreased that of EndMT markers (Fn1, N-cadherin, snail). These findings demonstrate that Smurf1 regulates Smad2 stability through ubiquitination and degradation, providing mechanistic insights into its role in TGF-β signaling and EndMT. This study highlights Smurf1 as a potential target for intervention in diseases involving EndMT, such as fibrosis and cardiovascular disorders (Supplementary Fig. S1). Supplementary Information Supplementary Information 1. Supplementary Information 2. Supplementary Information 3. Supplementary Information 4. Supplementary Information 5. Supplementary Information 6. Supplementary Information 1. Supplementary Information 2. Supplementary Information 3. Supplementary Information 4. Supplementary Information 5. Supplementary Information 6.