CAR-CD34 (+) hematopoietic stem/progenitor cells produced in vivo protect against thoracic aortic aneurysm and dissection

Abstract

Thoracic aortic aneurysm and dissection is one of the most devastating cardiovascular diseases, with limited medical intervention options. This study aims to develop chimeric antigen receptor-engineered CD34⁺ hematopoietic stem/progenitor cells and evaluate their effects on repairing endothelial cell injuries in a murine model. All animal experiments were performed in male mice. We design a chimeric antigen receptor containing a single-chain variable fragment targeting VCAM-1 and a VEGFA activation domain. A ligand-mediated lipid nanoparticle delivery system was used to engineer circulating CD34⁺ cells for transient and tunable chimeric antigen receptor expression. In vitro studies show that the engineered cells exhibit improved differentiation, proliferation, migration, adhesion, and tube formation. They successfully restore endothelial function, strengthened cell junctions, suppressed inflammatory response, and blocked disease progression. This study demonstrates that chimeric antigen receptor technique can effectively equip CD34⁺ hematopoietic stem/progenitor cells to target injured vascular intima, offering a promising approach for treating cardiovascular diseases.

Thoracic aortic aneurysm and dissection is a devastating disease with limited treatment options. Here, the authors show that CAR-engineered stem/progenitor cells can target injured aortic endothelium, restore its function, and halt disease progression, offering a new therapeutic strategy

Introduction

Thoracic aortic aneurysm and dissection (TAAD) is a critical condition characterized by high morbidity and mortality rates¹. The management of TAAD typically includes open surgery, endovascular repair, and medicinal treatment². The development of endovascular instruments and surgical techniques has increased treatment efficacy and provided patients with a variety of benefits. Nevertheless, a number of patients are deemed unsuitable for surgical therapy due to comorbidities, seniority, and other factors. Consequently, they must be treated with conservative medicinal management, including anti-hypertensive and anti-platelet medications³. Furthermore, endovascular or surgical interventions may not be appropriate for certain early aortic intimal injuries, such as intramural hematoma and limited intimal tears⁴. Therefore, it is essential to develop medications that effectively facilitate the repair of early-stage injuries, inhibit the progression of TAAD, and prevent their rupture.

It has been shown that endothelial cells (ECs) play a vital role in the development of aortic dilatation disorders and may be a potential target of medical treatment^5,6. The aortic endothelium serves as a barrier that has anti-thrombotic properties and inhibits leukocyte adhesion⁷. In addition, it functions as a sensor for hemodynamics and metabolism^8,9. Research has shown that the loss of ECs and alterations in function might be the major initiators and underlying mechanisms of aortic dissection¹⁰. Early restoration of the endothelial barrier in TAAD may be crucial for preventing subsequent damage to the aortic media.

Circulating CD34⁺ hematopoietic stem/progenitor cells (HSPCs) have been proven to mobilize from the bone marrow and exert repairing and vascular regeneration effects in response to cardiovascular diseases^11,12. They may directly differentiate into ECs and release bioactive molecules like exosomes and PPAR agonizts to improve the activity of ECs^13–15. Decreased levels of circulating CD34⁺ HSPCs are associated with endothelial dysfunction, accelerated atherosclerosis, and unfavorable outcomes such as coronary artery disease and peripheral vascular disease¹⁶. Recent studies have demonstrated that the administration of CD34⁺ HSPCs can promote ECs growth on vascular grafts¹⁷. However, it was found that systemic administration of CD34⁺ HSPCs resulted in poor cell localization and impaired cell viability, herein leading to diminished therapeutic efficacy^18,19. To address these challenges, current studies have investigated ways to recruit CD34⁺ cells at the lesion site, such as using tailored scaffolds to improve CD34⁺ cell retention within the wound bed²⁰. Despite these efforts, effectively increasing the recruitment of CD34⁺ HSPCs to vascular lesions remains a considerable challenge in medical research, necessitating further refinement of current cell-based therapies. Our research aims to augment both the quantity and functionality of CD34⁺ HSPCs and to optimize their homing to the site of injury.

Chimeric antigen receptor T cells (CAR-T) and macrophages (CAR-M) have been extensively utilized in cancer-targeted therapy, showing remarkable therapeutic efficacy²¹. CAR technology may bestow HSPCs with targeting capability and therapeutic potential for cardiovascular diseases. Li et al. have used genetically engineered CAR-hematopoietic stem cells (CAR-HPSCs) to generate allogeneic HSC-engineered iNKT cells for effectively targeting tumor cells²². Sakiko et al. crafted a dual-antigen receptor T cell derived from induced CAR-pluripotent stem cells (CAR-iPSCs) to combat tumor evasion²³. CAR-mesenchymal stem cells (CAR-MSCs) modified with e-cadherin-targeted single-chain fragment variable (scFv) in mice were found to effectively target and mitigate graft-versus-host disease²⁴. These findings suggest that the application of CAR-based technology in stem cell therapy is both feasible and promising for future advancements.

The aim of this study is to produce CAR-modified CD34⁺ HSPCs that express vascular endothelial growth factor-A (VEGFA) and an scFv targeting vascular cell adhesion molecule-1 (VCAM-1) to enhance their homing and repair functions. Furthermore, we also evaluate the repairing effects of CAR-CD34⁺ HSPCs, which were genetically programmed via the CD34-ligand-mediated lipid nanoparticles (LNP) system in a TAAD murine model.

Results

Synthesis and characterization of CAR plasmids for lentivirus plasmid transfection and CD34-targeted CAR mRNA-loaded LNP (CD34/LNP-CAR mRNA)

To confer CD34⁺ HSPCs with the ability to target VCAM-1 expressed in aortic lesions and secrete ample VEGFA, thus activating the CD34⁺ cells and facilitating EC repair, we engineered a specialized cell termed VEGFA-expressing VCAM-1 CAR-CD34⁺ HSPCs (VEGFA-VCAM-1-CAR HSPCs). This tandem construct comprised a first-generation anti-VCAM-1 CAR construction and mouse VEGFA connected via a cleavable 2 A peptide sequence (Fig. 1a). Figure 1b illustrates the process of plasmid construct preparation and in vitro mRNA transcription.

Fig. 1. The structural characterization of CD34/LNP-CAR mRNA and its cellular uptake in CD34⁺ HSPCs.

a Schematic depicting the chimeric antigen receptor constructs. b Flow chart of in vitro mRNA transcription. Created in BioRender. Zhao, K. (2024) https://BioRender.com/s23r279. c TEM imaging of the CD34/LNP-CAR mRNA; and (d, e) DLS analysis of the CD34/LNP-CAR mRNA. Data in (c) are representative images from two independent experiments. f Time-dependent DLS size in PBS for 7 days. Data in (f) are from n = 3 biologically independent samples. g Workflow for CD34⁺ HSPCs isolation. Created in BioRender. Zhao, K. (2024) https://BioRender.com/x57e354. h Cell surface markers identified by flow cytometric analysis. Data in (h) are from n = 3 biologically independent samples. i Confocal images of CD34⁺ HSPCs treated with PBS, Fr-RNAs, LNP-CAR mRNAs, or CD34/LNP-CAR mRNA. mRNA was labeled with Cy5 (red), and the cell membrane and nuclei were counterstained with DIO (green) and DAPI (blue), respectively. j Cellular uptake of CD34/LNP-CAR mRNA by CD34⁺ HSPCs via flow cytometry analysis. Data in (j) are from n = 3 biologically independent samples. k Confocal images of subcellular compartments of CD34⁺ HSPCs incubated with CD34/LNP-CAR mRNA for 1, 3, and 6 h at 37 °C. l Quantitative analysis of co-localization of Cy5-mRNAs with endo/lysosomes labeled with Lysotracker Green. The cell nuclei were stained using Hoechst 33342 (blue), endo/lysosomes were stained using Lysotracker Green (green), and mRNAs were labeled with Cy5 (red). Data in (l) are from n = 3 biologically independent samples. m Viability of CD34⁺ HSPCs in different treatment groups at 24 h post-treatment. Data in (m) are from n = 5 biologically independent samples. All data are presented as mean ± s.d. Statistical significance and P-values of (j), and (m) were analyzed by one-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). Statistical significance and P- values of (l) were analyzed by two-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). *P < 0.05, **P < 0.01, ***P < 0.001. Scale bars, 200 or 100 nm (c); 10 μm (i); 10 μm (k). Source data are provided as a Source Data file.

CD34/LNP-CAR mRNA was synthesized through the self-assembly process involving Dlin-MC3-DMA cationic lipids and negatively charged mRNA, followed by encapsulation within helper lipids (DSPC and cholesterol) and a PEGylated lipid (DMPE-PEG₂₀₀₀) modified with CD34 antibody. Ethanol-based lipid solutions were generated with a composition of four lipid components [MC3: cholesterol: DSPC: DMPE-PEG₂₀₀₀] in a molar ratio of 50:38.5:10:1.5 mol%. Transmission electron microscope (TEM) imaging (Fig. 1c) demonstrated the average size of CD34/LNP-CAR mRNA to be about 100 nm, confirming a homogeneous spherical form. Furthermore, dynamic light scattering (DLS) analysis indicated that the size of CD34/LNP-CAR mRNA particles was around 130 nm, with a polydispersity index (PDI) of about 0.15 (Fig. 1d). Figure 1e revealed that the zeta potential of CD34/LNP-CAR mRNA was − 26.7 and provided an estimated mRNA encapsulation effectiveness (EE) of 86%. Notably, CD34/LNP-CAR mRNA displayed good stability in phosphate-buffered saline (PBS), as seen by insignificant changes in particle size over a 7-day period (Fig. 1f).

Cellular uptake of CD34⁺ HSPCs and CD34/LNP-CAR mRNA-mediated CAR expression in CD34⁺ HSPCs

Figure 1g is the flow chart showing that CD34⁺ HSPCs were isolated from the bone marrow of male C57BL/6 J mice, achieving a purity of approximately 96.16 ± 1.22% (n = 3) (Fig. 1h). The immunofluorescence (IF) results of CD34 are shown in Supplementary Fig. 1. Using mRNA tagged with a red fluorescent probe (Cy5), CAR mRNA encapsulated in LNPs was observed to be efficiently internalized into the cytoplasm within 240 min (Fig. 1i). In addition, the absorption of LNPs by CD34⁺ HSPCs was considerably enhanced when the CD34 antibody was modified, possibly owing to receptor-mediated endocytosis. These results were corroborated by flow cytometry (Fig. 1j). As indicated in Fig. 1k, the co-localization of red fluorescence of Cy5 and green fluorescence of Lysotracker Green in CD34⁺ HSPCs treated with CD34/LNP-CAR mRNA reduced with time, implying efficient release of CD34/LNP-CAR mRNA from endo/lysosomes. The veracity of this discovery was also verified by quantitative examination of the co-localization of CD34/LNP-CAR mRNA with endo/lysosomes in confocal fluorescence pictures employing Manders’ coefficients M1 and M2 (Fig. 1l).

In order to investigate the impact of CD34/LNP-CAR mRNA on CAR expression in CD34⁺ HSPCs, we evaluated cell viability following treatment with PBS, LNP, CD34 antibody-modified LNP without mRNA encapsulated (CD34/LNP), LNP with CAR mRNA encapsulated (LNP-CAR mRNA), and varying concentrations of CD34 antibody-modified LNP with mRNA encapsulated (CD34/LNP-CAR mRNA) (0.1, 0.2, 0.4, and 0.8 ug/ml, respectively). Our results, as demonstrated in Fig. 1m and Supplementary Fig. 2a, b, show that the cell viability in the LNP-CAR mRNA group was significantly higher than in the LNP group, indicating successful expression of VEGFA. Furthermore, compared to the LNP-CAR mRNA group (0.1 μg/ml), the cell viability in the CD34/LNP-CAR mRNA group (0.1 μg/ml) was significantly increased, confirming the targeting effect of the CD34 antibody. Notably, the viability of CD34⁺ HSPCs significantly increased at a concentration of 0.2 μg/ml after 24 and 48 h but decreased as the concentration was further increased.

Following a 24-hour in vitro incubation of CD34/LNP-CAR mRNAs with CD34⁺ HSPCs, the expression of anti-VCAM-1 scFv was quantified through flow cytometry, while VEGFA expression was evaluated via quantitative reverse transcription polymerase chain reaction (qRT-PCR) and Western blot analyses. Both CD34⁺ cells transfected with lentiviral vectors overexpressing VCAM-1-VEGFA-CAR and CD34/LNP-CAR mRNA successfully expressed anti-VCAM-1 scFv, whereas the PBS, LV-IgG-CAR, and CD34/LNP groups did not show this expression. Compared with the LNP-CAR mRNA group, the expression of anti-VCAM-1 scFv was remarkably higher in the CD34/LNP-CAR mRNA group (Fig. 2a, b). Furthermore, to validate the specificity of CD34 antibody targeting, we isolated CD34^- cells for analysis. Our findings indicated that the expression of anti-VCAM-1 scFv was significantly elevated in CD34⁺ cells compared to CD34^- cells following treatment with CD34/LNP-CAR mRNA (Supplementary Fig. 3). In this study, we also evaluated the efficiency of VCAM-1-VEGFA-CAR expression in CD34⁺ cells and optimized the transfection efficacy of CD34/LNP-CAR mRNA across four concentration gradients. Our results indicate that the expression of anti-VCAM-1 scFv in CD34⁺ cells reached its peak point at a concentration of 0.4 μg/ml, with a decline observed at higher concentrations (Supplementary Fig. 4a). Furthermore, the research investigated the stability of anti-VCAM-1 scFv expression, with data revealing that the expression of anti-VCAM-1 scFv was stable throughout a three-day period (Supplementary Fig. 4b). It was also revealed that the expression of VEGFA was significantly higher with the 0.1 μg/ml CD34/LNP-CAR mRNA treatment compared to the 0.1 μg/ml LNP-CAR mRNA treatment, underscoring the critical role of the CD34 antibody in facilitating LNP absorption and subsequent CAR-mRNA expression. The level of VEGFA expression exhibited a peak increase of 17.26-fold compared to the control group treated with PBS when exposed to a concentration of 0.2 ug/ml (Supplementary Fig. 4c). Based on the above results, we used 0.2 ug/ml CD34/LNP-CAR mRNA as the therapeutic concentration for subsequent functional cell experiments. Western blot analysis indicated a substantial increase in VEGFA expression in CD34⁺ HSPCs transfected with lentiviral vectors overexpressing the CAR plasmid (LV-CAR) and those treated with 0.2 μg/ml CD34/LNP-CAR mRNAs compared to the other groups (Fig. 2c).

Fig. 2. CAR expression of CAR-CD34⁺ cells transformed via lentivirus or CD34/LNP-CAR mRNA in vitro.

a Representative gating strategy for flow cytometry and analyses of VCAM-1-scFv-CAR expression in freshly isolated CD34⁺ cells after transfection with PBS, LV-IgG-CAR, CD34/LNP, LNP-CAR mRNA, LV-VCAM-1-CAR, and CD34/LNP-CAR mRNA. b Quantitative analyses of the VCAM-1 CAR expression in the CD34⁺ cells between groups. Data in (b) are from n = 3 biologically independent samples. c Western blot analyses of VEGFA expression. Data in (c) are from n = 3 biologically independent samples. d The EdU assay determined the EdU-positive CD34⁺ cell rate. Data in (d) are from n = 5 biologically independent samples. e Flow cytometry was performed to characterize the endothelial markers (CD31 and VE-CAD) in freshly isolated CD34⁺ cells. Data in (e) are from n = 3 biologically independent samples. All data are presented as mean ± s.d. Statistical significance and P-values were analyzed by using one-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). *P < 0.05, **P < 0.01, ***P < 0.001. Scale bars, 200 µm (d); 10 µm (k). Source data are provided as a Source Data file.

Functional evaluation of CAR-CD34⁺ HSPCs in vitro

VEGFA is hypothesized to facilitate the migration, differentiation, and proliferation of CD34⁺ HSPCs, crucial processes for angiogenesis^25,26. The results of the 5-Ethynyl-2′-deoxyuridine (EdU) assay demonstrated a significant increase in cell proliferation capacity in the LV-CAR group and those transfected with CD34/LNP-CAR mRNA, as compared to the remaining groups. Furthermore, the cell proliferation rate in the CD34/LNP-CAR mRNA group was slightly lower than that found in the LV-CAR group (Fig. 2d). After three days of overexpression of the CAR plasmid and treatment with CD34/LNP-CAR mRNA (0.2 ug/ml) in the freshly isolated CD34⁺ HSPCs, flow cytometry and immunofluorescence analysis revealed a significant increase in the expression of endothelial markers, including VE-cadherin (VE-CAD) and CD31, compared to the PBS and CD34/LNP groups (Fig. 2e). The immunofluorescence results also indicated that treatment with both LNP-CAR mRNA and CD34/LNP-CAR mRNA indeed increased the expression of the marker CD31 and VE-cadherin (Supplementary Fig. 5). According to the data provided in Fig. 3a, b, the wound-healing and transwell tests showed improved wound healing and more cell migration in the LV-CAR and CD34/LNP-CAR mRNA groups compared to the PBS and CD34/LNP groups. This observed influence may also be attributable to enhanced cell proliferation, since VEGFA is known to drive differentiation of CD34⁺ HSPCs into ECs and boost proliferation of both CD34⁺ cells and ECs^25–27. The results showed a considerable boost in the adhesion capacity of CAR-CD34⁺ HSPCs compared to control groups (Fig. 3c, d). The tube formation assay, assessing the angiogenesis potential of CD34⁺ HSPCs treated by the four groups, revealed that the LV-CAR and CD34/LNP-CAR mRNA groups exhibited greater total branching length and more junctions than the other groups (Fig. 3e, f). Nevertheless, the disparity between the LV-CAR and CD34/LNP-CAR mRNA groups was not statistically significant. Furthermore, in vitro mRNA sequencing was performed on CD34⁺ HSPCs after treatment with CD34/LNP-CAR mRNA and in the control group. The results showed that after treatment with CD34/LNP-CAR mRNA, the CD34⁺ HSPCs exhibited significantly enhanced functions related to angiogenesis, adhesion, and stem cell differentiation into endothelial cells (Supplementary Fig. 6). The protein-protein interaction (PPI) interaction network analysis identified VEGFA as a central hub after CD34/LNP-CAR mRNA treatment (Supplementary Fig. 7). In addition, to exclude the potential influence of CD34 antibodies of the CD34⁺ cells, we added the in vitro experiments assessing migration, adhesion, and colony-forming unit (CFU) assays, as well as a brief transcriptomic comparison (3 vs. 3 samples) of CD34⁺ cells treated with LNP-CAR mRNA and CD34/LNP-CAR mRNA. The transcriptomic profiles of CD34⁺ cells treated with CD34/LNP-CAR mRNA and LNP-CAR mRNA showed strong similarity. As shown in Supplementary Fig. 8a, the Pearson correlation heatmap revealed high intra-group and inter-group correlations, indicating consistent gene expression patterns between treatments. Violin plots further confirmed comparable global expression distributions across samples (Supplementary Fig. 8b). The volcano plot demonstrated that only a small number of genes were differentially expressed between the two groups, suggesting minimal transcriptomic alterations (Supplementary Fig. 8c). GO and KEGG enrichment analyses indicated that the few upregulated genes were mainly associated with metabolic regulation and immune-related signaling pathways, but no major functional pathway differences were observed (Supplementary Fig. 8d, e). The results showed no statistically significant differences in the numbers of migrated or adherent cells between the LNP-CAR mRNA and CD34/LNP-CAR mRNA groups (Supplementary Fig. 9), suggesting that CD34 antibody modification did not affect the function or gene expression of CD34⁺ cells.

Fig. 3. Functional evaluation of CAR-CD34⁺ cells transformed via lentivirus or CD34/LNP-CAR mRNA in vitro.

a Wound healing and transwell assays were used to evaluate CD34⁺ cells’ migration ability. b Quantitative data of the wound healing and transwell. Data in (b) are from n = 5 biologically independent samples. c Adhesion assays of the CD34⁺ cells. d Quantitative data for the adhesion assays. Data in (d) are from n = 5 biologically independent samples. e Tube formation assay of the CD34⁺ cells. f Tube formation was quantified by total tube formation length and branch junction. Data in (f) are from n = 5 biologically independent samples. All data are presented as mean  ± s.d. Statistical significance and P-values were analyzed by using one-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). *P < 0.05, **P < 0.01, ***P < 0.001. Scale bars, 200 μm (a); 100 μm (c); 200 μm (e). Source data are provided as a Source Data file.

Endothelial barrier function is disturbed in the early-stage of TAAD

To investigate the key time points in the TAAD progression and determine the most effective timing for intervention, we employed a β-aminopropionitrile (BAPN)-induced mouse model of TAAD and designated 7, 14, and 21 days as observation time points for echocardiographic analysis (Fig. 4a). Following 7 days of BAPN inducement, echocardiography indicated a progressive increase in the mean diameter of the aorta in the BAPN group, suggesting that day 7 may represent a critical time point (Fig. 4b). However, the difference between the control and BAPN groups was not found to be statistically different. Gross anatomy examinations indicated the onset of TAAD formation by day 14. As shown in Fig. 4c, the frequency of TAAD in the BAPN group was assessed at 7, 14, and 21 days. Notably, no TAAD was observed on day 7. Scanning electron microscopy (SEM) was applied to investigate the morphological alterations in the aorta endothelium throughout the early phases of TAAD development, especially at 7, 14, and 21 days post-inducement. The findings demonstrated a steady degradation of the endothelium barrier, with early disruption identified at 7 days post-BAPN inducement, advancing to severe damage by days 14 and 21 (Supplementary Fig. 10). The aorta’s Western blot indicated that CD31 expression declined modestly in the first seven days, then reduced dramatically after seven days, and only a small quantity remained after 14 days (Fig. 4d, e). The level of CD34 expression shows a gradual decline, possibly due to the existence of a number of CD34⁺ cells in the adventitia of the aorta²⁸. It was also shown that the VCAM-1 expression in the endothelium began to increase after 7 days. At 21 days, the aorta’s structure was disrupted. During the progression from early to late modeling stages, ECs gradually degrade, and the expression of VCAM-1 in endothelium was decreased, while the smooth muscle cell (SMC) began to express VCAM-1. By the 21st day, despite the loss of the aortic intima, there is significant expression of VCAM-1 by the SMC within the media (Fig. 4f) (Supplementary Fig. 11).

Fig. 4. The endothelial barrier is disrupted in the early stage of TAAD, and CD34/LNP-CAR mRNA successfully transforms CAR-CD34⁺ HSPCs in circulation to target the aortic endothelial lesion.

a Schematic scheme of the experimental design to assess the aortic endothelium injury and aortic dilation in the BAPN-induced TAAD murine model. Created in BioRender. Zhao, K. (2024) https://BioRender.com/z57v924. b Representative echocardiography imaging and quantitative analyses of the maximum ascending aortic diameter of the control group, as well as on days 7, 14, and 21 of the BAPN-inducement. Data in (b) are from 5 biologically independent samples. c Representative gross picture and quantitative analyses of the incidence of TAAD in the control group, as well as on days 7, 14, and 21 of the BAPN-inducement. Results in (c) are from 3 independent experiments. d Western blot analyses of the aortic CD31 and CD34 expressions. e Quantitative analyses of CD31 and CD34 expression. Data in (e) are from 3 biologically independent samples. f Representative VCAM-1 IHC imaging of the aorta from the control group and on days 7, 14, and 21 of the BAPN-inducement. Data in (f) are representative images from two independent experiments. g Flow cytometry was performed to characterize the time-dependent circulating CAR-CD34⁺ HSPCs between groups. h Quantitative analyses of the circulating CAR-CD34⁺ HSPCs. Results in (h) are from 3 independent experiments. i IF colocalization studies on Cy5 and CD34 of aorta from the control, LNP-CAR mRNA, CD34/LNP, and CD34/LNP-CAR mRNA groups. j Quantitative analyses of Cy5 intensity. Results in (j) are from 3 independent experiments. All data are presented as mean ± s.d. Statistical significance and P- values of (b, e, h, and j) were analyzed by using two-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). Statistical significance and P-values of (c) were analyzed by using one-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). *P < 0.05, **P < 0.01, ***P < 0.001. Scale bars, 20 μm (d); 20 μm (f); 10 μm (i). Source data are provided as a Source Data file.

CD34/LNP-CAR mRNA efficiently converted CAR-CD34⁺ HSPCs in vivo which targeted the aortic endothelial lesion

To assess the effectiveness of CAR-CD34⁺ HSPCs transformation in vivo, flow cytometry analysis was done on the peripheral blood of male C57BL/6 J mice exposed to BAPN inducement and varied treatments over a period of 12, 24, 36, and 48 h. Given the limitations of the FSC/SSC gating strategy, CD34⁺ HSPCs were first gated based on forward and side scatter to exclude debris, followed by selection of single cells and live cells using a viability dye (FL1-A::V450-A) (Supplementary Fig. 12a). The results showed that CD34 expression was largely maintained throughout the short time course (0–8 h), with minimal induction of VE-cadherin, an endothelial lineage marker (Supplementary Fig. 12b). The findings show that CD34/LNP-CAR mRNA treatment resulted in the successful generation of CAR-CD34⁺ HSPCs in the peripheral blood, while there was no circulating CAR expression detected following the saline, LNP-CAR mRNA, and CD34/LNP treatments (Fig. 4g, h).

It appears that CD34/LNP-CAR mRNA is well-tolerated and biocompatible, based on the absence of significant signs of injury in the heart, lungs, kidneys, liver, spleen, and brain, as illustrated in Supplementary Fig. 13. Besides, no significant difference in alanine transaminase (ALT), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), blood urea nitrogen (BUN), hemoglobin, or red blood cell (RBC) counts was observed between the mice that underwent saline, LNP-CAR mRNA, CD34/LNP and CD34/LNP-CAR mRNA treatments (Supplementary Fig. 14). IF analysis of the aorta at 7 and 21 days after BAPN inducement demonstrated a reduction in CD34 expression in both the LNP-CAR mRNA and CD34/LNP groups compared to the control group. However, CD34/LNP-CAR mRNA preserved the loss of CD34 (Fig. 4i). Furthermore, no CD34 or Cy5 fluorescence was detected in the LNP-CAR mRNA group and CD34/LNP groups at 21 days, implying that the aortic structure might have been destroyed. Specifically, the co-localization of Cy5 (red) fluorescence with CD34 (green) was exclusively observed in the CD34/LNP and CD34/LNP-CAR mRNA groups. The observed Cy5 fluorescence in the CD34/LNP group can likely be attributed to the non-specific targeting effects of the CD34 antibody, as well as the presence of CD34-positive ECs within the intima during the early stages of BAPN inducement. Nevertheless, the CD34/LNP group exhibited a substantially reduced intensity of Cy5 fluorescence compared to the CD34/LNP-CAR mRNA group (Fig. 4j).

Given the off-target effects of CD34 antibodies, which may directly affect the aorta and other organs, we established a CD34/LNP loaded with VEGFA mRNA (CD34/LNP-VEGFA mRNA) group in the in vivo experiments. No aortic targeting was observed in the saline and LNP-CAR mRNA groups at 12 and 24 h (Fig. 5a). The presence of bioluminescence was observed in the CD34/LNP-VEGFA mRNA and CD34/LNP-CAR mRNA groups. However, comparative analysis revealed that the bioluminescence intensity in the aorta of subjects treated with CD34/LNP-CAR mRNA was significantly higher than that observed in the CD34/LNP-VEGFA mRNA group, thereby confirming the aortic targeting properties of CAR-HSPCs (Fig. 5b) (Supplementary Fig. 15a). Animals treated with LNP-CAR mRNA, CD34/LNP-VEGFA mRNA, and CD34/LNP-CAR mRNA demonstrated expression of LNP-delivered mRNA in their spleens and livers, consistent with the expected hepatic clearance of LNPs²⁹. No statistical significance was observed in the bioluminescence intensity in the livers and spleens at 12 and 24 h between groups (Fig. 5c) (Supplementary Fig. 15b). Moreover, bioluminescence was observed in the pulmonary and bone marrow tissues of the right leg in subjects administered with CD34/LNP-VEGFA mRNA and CD34/LNP-CAR mRNA. Statistical analysis revealed no significant difference in bioluminescence intensity between the two treatment groups (Supplementary Fig. 15c, d). Conversely, bioluminescence was not detected in the lungs of the LNP-CAR mRNA group, suggesting that CD34 antibodies might elicit off-target effects within pulmonary tissues.

Fig. 5. CD34/LNP-CAR mRNA induces CAR-CD34⁺ HSPCs to gather at the aortic lesion and inhibit the development of TAAD.

a Bioluminescence imaging was conducted at 12 h and 24 h following saline, LNP-CAR mRNA, CD34/LNP-VEGFA mRNA, and CD34/LNP-CAR mRNA treatment, respectively. b The intensity of bioluminescence in the aorta at 12 h was analyzed. Data in (b) are from n = 3 biologically independent samples. c The intensity of bioluminescence in the liver, spleen, and lung at 12 h was analyzed. Data in (c) are from n = 3 biologically independent samples. d Schematic scheme of the experimental design to assess the therapeutic effect of CD34/LNP-CAR mRNA in the BAPN-induced TAAD murine model. Created in BioRender. Zhao, K. (2024) https://BioRender.com/z57v924. e Body weight changes during model induction. Data in (e) are from n = 3 biologically independent samples. f The Kaplan-Meier survival probability curve for the four groups. Data in (f) are from n = 12 biologically independent samples. g Representative echocardiographic images. h Data analysis of the maximum aortic diameter. Data in (h) are from n = 3 biologically independent samples. i The representative gross anatomy images of mice. j The incidence of TAAD. Data in (i) are from n = 3 independent experiments. All data are presented as mean ± s.d. Statistical significance and P-values of (b), and (j) were analyzed by using one-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). Statistical significance and P-values of (c, e, and h) were analyzed by using two-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). Data of (f) were analyzed by using the Kaplan–Meier method and compared by the log-rank (Mantel-Cox) test. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

CAR-CD34⁺ HSPCs repaired the endothelial injury and prevent TAAD from occurring

Due to the non-specific targeting effects of CD34 antibodies, it is plausible that organs such as the liver and lungs could also express CAR mRNA. This expression may lead to elevated levels of circulating VEGFA protein, potentially confounding the experimental outcomes. To eliminate the potential confounding effect of elevated circulating VEGFA, one cohort of male C57BL/6 J mice was administered the same dose of CD34/LNP-VEGFA mRNA. Five cohorts of 3-week-old male C57BL/6 J mice were randomly allocated to one of the following intervention groups: a non-BAPN-induced control group, a BAPN + saline group (saline group), a BAPN + CD34/LNP group (CD34/LNP group) serving as the solvent control, a BAPN + CD34/LNP-VEGFA mRNA group (CD34/LNP-VEGFA mRNA group), and a BAPN + CD34/LNP-CAR mRNA group (CD34/LNP-CAR mRNA group) (Fig. 5d). Enzyme-linked immunosorbent assay (ELISA) results indicated that, following 7 and 14 days of BAPN induction, the levels of circulating VEGFA proteins in the CD34/LNP-VEGFA mRNA and CD34/LNP-CAR mRNA groups were significantly elevated compared to the other three groups; however, no significant difference was detected between the CD34/LNP-VEGFA mRNA and CD34/LNP-CAR mRNA groups (Supplementary Fig. 16). As depicted in Fig. 5e, mice in the saline, CD34/LNP, and CD34/LNP-VEGFA mRNA groups exhibited a significant reduction in body weight compared with the control group. Conversely, administration of CD34/LNP-CAR mRNA significantly mitigated the weight inhibition effects of the BAPN. Compared with the saline, CD34/LNP, and CD34/LNP-VEGFA mRNA groups, the CD34/LNP-CAR mRNA treatment significantly lowered the mice mortality rate in 21 days (Fig. 5f). Figure 5g shows representative echocardiographic imaging for the five groups. Quantitative analysis shows the maximum aortic diameters of the remaining mice between the CD34/LNP and CD34/LNP-CAR mRNA groups were not statistically different (Fig. 5h), which might be due to the small number of survivors. In addition, CD34/LNP-CAR mRNA treatment demonstrated a notable decrease in the incidence of TAAD (Fig. 5i). Figure 5j shows representative gross samples from the five groups.

The subsequent investigation focused on ascertaining whether endothelial restoration could ameliorate aortic damage, as evidenced by the degradation of collagen and elastin fibers within the extracellular matrix (ECM). H&E, Masson, and Elastic Van Gieson (EVG) staining demonstrated a significant reduction in collagen and elastin degradation within the aortic media, accompanied by enhanced preservation of aortic structural integrity in the CD34/LNP-CAR mRNA group, in comparison to the saline, CD34/LNP, and CD34/LNP-VEGFA mRNA groups (Fig. 6a, b). Moreover, compared with the saline, CD34/LNP, and CD34/LNP-VEGFA mRNA groups, CD34/LNP-CAR mRNA therapy substantially decreased BAPN-induced Evans blue leakage in the aorta during 21-day BAPN inducement (Fig. 6c, d). SEM analyses conducted 21 days post-BAPN induction demonstrated that CD34/LNP-CAR mRNA therapy effectively restored the endothelial barrier to its original condition, as depicted in Fig. 6e. IF staining results further corroborate the recovery of the aortic endothelial barrier in the CD34/LNP-CAR mRNA group, evidenced by the presence of cells co-expressing CD34 and CD31(Fig. 6f, g). In addition, the mean density of CD34 within the intima was significantly greater than that observed in the saline, CD34/LNP, and CD34/LNP-VEGFA mRNA groups (Fig. 6h). No significant differences were detected in CD34 density within the adventitia across the groups. Furthermore, IF staining for VEGFA and VEGFR2 was conducted on the aortas of mice 21 days post-modeling. The findings indicated that VEGFR2 fluorescence was consistently observed across all layers of the aorta in all animal groups. In contrast, the blank control group exhibited no VEGFA fluorescence in the aorta. However, the saline, CD34/LNP, and CD34/LNP-VEGFA mRNA groups demonstrated a significant increase in VEGFA expression, with a substantial concentration in the outer membrane. Notably, the CD34/LNP-CAR mRNA group displayed markedly reduced VEGFA expression compared to the control group, with the majority localized within the intima and media (Supplementary Fig. 17).

Fig. 6. CD34/LNP-CAR mRNA treatment repairs the intima injury and protects the aortic structure.

a Representative image showing HE, MASSON, and EVG staining for aortic cross-sections. b Data of the collagen positive area and elastin fragmentation. Data in (b) are from n = 5 biologically independent samples. c Permeability of the aortic intimal barrier (to Evans blue dye). d The quantity data of percentage of endothelialization. Data in (d) are from n = 3 biologically independent samples. e Representative SEM imaging. Data in (e) are representative images from two independent experiments. f Representative immunofluorescence of CD34 and CD31 in the aortic sections. g Quantitative analyses of intima CD31 intensity Data in (g) are from n = 3 biologically independent samples. h Quantitative analyses of intima and adventitia CD34 intensity, Data in (h) are from n = 3 biologically independent samples. All data are presented as mean ± s.d. Statistical significance and P-values of (b, d and g) were analyzed by using one-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). Statistical significance and P-values of (h) were analyzed by using two-way ANOVA followed by Tukey’s multiple comparisons test (two-sided). *P < 0.05, **P < 0.01, ***P < 0.001. Scale bars, 200 or 20 μm (a); 20 μm (e); 10 or 20 μm (f). Source data are provided as a Source Data file.

Furthermore, ELISA analyses of peripheral blood on the 14th and 21st days following inducement of the TAAD model indicated that the matrix metalloproteinase-2 (MMP-2) level was significantly reduced in the CD34/LNP-CAR mRNA group compared to the saline, CD34/LNP, and CD34/LNP-VEGFA mRNA groups on the 14th day (Supplementary Fig. 18). Besides, tumor necrosis factor-alpha (TNF-α) in the CD34/LNP-CAR mRNA group was found to be substantially lower than that in the CD34/LNP and CD34/LNP-VEGFA mRNA groups on the 14th and 21st days (Supplementary Fig. 18c). Interleukin-6 (IL-6) in the CD34/LNP-CAR mRNA group was significantly lower than that in the CD34/LNP groups (Supplementary Fig. 18d). However, it was found that CD34/LNP may probably increase the risk of the release of inflammatory cytokines, such as TNF-α and IL-6, compared with the saline group.

This cumulative information implies that CD34/LNP-CAR mRNA therapy may offer protection against TAAD development and aortic rupture in mice.

CD34/LNP-CAR mRNA treatment reshapes aortic cellular landscape and enhances endothelial repair

To gain a more thorough understanding of the effects of CAR-CD34⁺ HSPCs on the aortic remodeling at the early stage, we performed a single-cell transcriptome sequencing on aortic tissues from BAPN-induced mice simultaneously treated with saline and CD34/LNP-CAR mRNA for 7 days. Cell clusters were visualized using uniform manifold approximation and projection (UMAP) to investigate the cellular landscape and the effect of CD34/LNP-CAR mRNA treatment on cell composition (Fig. 7a). A total of 10 major cell populations were identified across all samples, including ECs, epithelial cells, fibroblasts, macrophages, neutrophils, Prol cells, Schwann cells, SMCs, and T cells. Notably, SMCs, fibroblasts, and ECs constituted the predominant populations. To assess treatment-induced compositional changes, we compared the cellular proportions between the CD34LNP-CAR mRNA and saline groups (Fig. 7b). In comparison to the saline group, the CD34/LNP-CAR mRNA group demonstrated a significant increase in the proportion of ECs and a corresponding decrease in fibroblasts and SMCs. Furthermore, the treatment group exhibited a reduced presence of macrophages and neutrophils. These results suggest that CD34LNP-CAR mRNA administration alters the vascular microenvironment, potentially promoting EC restoration and modulating immune cell populations. The marker gene expression analysis revealed distinct transcriptional signatures across major cell types, as depicted in Fig. 7c. Cell-cell interaction analysis revealed significant alterations in intercellular communication networks following CD34/LNP-CAR mRNA treatment compared to the saline group (Fig. 7d). Saline-treated samples exhibited extensive interactions between inflammatory cells (neutrophils, macrophages) and vascular-wall cells (ECs, SMCs), while CD34/LNP-CAR mRNA treatment significantly suppressed inflammatory and vascular crosstalk compared to saline controls. To investigate the EC heterogeneity, we performed unsupervised clustering of single-cell transcriptomic data. UMAP dimensionality reduction identified 11 distinct clusters of ECs (Supplementary Fig. 19a), revealing a high degree of transcriptional diversity within the population. These clusters were subsequently annotated into five major endothelial subtypes based on canonical marker gene expression (Fig. 7e): arterial ECs (Bmx, and Fbln5), capillary arterial cells (Sema3g, and S100a4), capillary ECs (Ackr1 and Fabp4), lymphatic ECs (Prox1 and Flt4), and microvascular ECs (CD36 and Pparg) (Supplementary Fig. 19b). These gene expression signatures supported the functional annotation of each endothelial subtype and underscored the molecular distinctiveness of each cluster. Specifically, the group CD34/LNP-CAR mRNA exhibited a relative increase in arterial ECs and microvascular ECs compared to the saline group, suggesting the potential repairing effect of CD34/LNP-CAR mRNA (Fig. 7e). Gene ontology (GO) enrichment analysis of differentially expressed genes following CD34/LNP-CAR mRNA treatment revealed robust activation of endothelial repair programs essential for aortic intima regeneration. Key enriched pathways converged on vascular morphogenesis, dominated by angiogenesis and vasculogenesis, with positive regulation of angiogenesis ranking among the top five enriched terms; endothelial migration and barrier formation, characterized by activation of cell migration pathways, alongside upregulation of cell adhesion junction assembly (Fig. 7f).

Fig. 7. Single-cell transcriptomic profiling reveals vascular and immune modulation by CD34/LNP-CAR mRNA therapy.

a UMAP plot showing all cells isolated from aortic tissues of BAPN-induced mice treated with either saline or CD34/LNP-CAR mRNA for seven days. b Relative proportions of major cell clusters in the aortic samples from the saline and CD34/LNP-CAR mRNA groups. c Violin plots depicting the expression levels of representative marker genes across different cell types. d Cell–cell interaction networks inferred under saline and CD34/LNP-CAR mRNA treatment conditions. e UMAP projection of aortic endothelial cells, color-coded by molecularly defined subtypes, demonstrating compositional differences in endothelial subclusters between saline- and CD34/LNP-CAR mRNA–treated aortic samples, and the relative proportions of each subtype. f GO enrichment analysis of differentially expressed genes within the endothelial cell clusters following CD34/LNP-CAR mRNA treatment. Dot size corresponds to the number of genes; color represents the -log₁₀(P-value). Analysis was performed using the clusterProfiler R package based on a two-sided hypergeometric test, with P-values adjusted for multiple comparisons using the Benjamini-Hochberg (FDR) method. Only terms with an adjusted P < 0.05 are shown. Source data are provided as a Source Data file.

To further substantiate the reparative effects of CD34/LNP-CAR mRNA, a transcriptomic analysis was conducted on aortic tissues following a 21-day induction with BAPN. The results demonstrated that treatment with CD34/LNP-CAR mRNA led to the upregulation of 3796 genes and the downregulation of 1870 genes, as depicted in Supplementary Fig. 20a. The findings revealed that treatment with CD34/LNP-CAR mRNA resulted in an increase of 3796 genes and a reduction of 1870 genes, as illustrated in Supplementary Fig. 20b. The enrichment of gene sets involved with the cell adhesion molecules pathway was especially noticeable following CD34/LNP-CAR mRNA treatment (Supplementary Fig. 20c). It suggests that the treatment may have an effect on reducing inflammation in endothelial cells. The gene ontology (GO) cellular component analysis revealed enrichment in cell junctions, as depicted in Supplementary Fig. 20d. Furthermore, a reduction in mRNA expression of adhesion molecules, specifically VCAM-1, P-selectin, ICAM-1, and E-selectin, was noted in the CD34/LNP-CAR mRNA group, as illustrated in Supplementary Fig. 20e. Reduced protein levels of VCAM-1, ICAM-1, P-selectin, and E-selectin were identified by Western blot analysis, which validated this result and was consistent with the sequencing data (Supplementary Fig. 20f, g). Supplementary Fig. 20h–j demonstrates that the CD34/LNP-CAR mRNA group exhibited higher levels of mRNA and protein expression of the cell junction markers ZO-1 and VE-CAD, according to qPCR and Western blot analyses. Furthermore, alongside endothelial injury, the GO biological process analysis identified enrichment of the inflammatory response, suggesting the participation of a substantial quantity of inflammatory cells (Supplementary Fig. 20k). After conducting qRT-PCR studies, it was found that the CD34/LNP-CAR mRNA group exhibited lower mRNA levels of MMP-9, TNF-α, IL-6, and interleukin beta (IL-1β) (Supplementary Fig. 20l). The protein expression levels of IL-1β, IL-6, TNF-α, and MMP-9 were consistent with the qRT-PCR results (Supplementary Fig. 20m, n).

Given the potential for CAR-CD34 HSPCs to secrete reparative extracellular vesicles (EVs) that facilitate endothelial repair, we conducted an evaluation of the differential miRNA expression patterns in the aortas of mice between the saline-treated group and the CD34/LNP-CAR mRNA-treated group. Our analysis revealed that, compared to the saline group, CD34/LNP-CAR mRNA treatment upregulated 91 miRNAs and downregulated 342 miRNAs (Supplementary Fig. 21a, b). Among the upregulated miRNAs, we identified four highly differentially expressed ones (miR-143-5p, miR-127-3p, miR-206-3p, and miR-137-3p) based on a fold change value ≥ 2 (Supplementary Fig. 21b). To validate these miRNA sequencing results, we measured the expression of four miRNAs in the two groups’ aortas using qRT-PCR. Our findings showed that CD34/LNP-CAR mRNA treatment significantly upregulated these four genes, with miR-127-3p exhibiting the highest upregulation (Supplementary Fig. 21c).

Discussion

At present, there is a lack of effective medical therapies in clinical settings for early-stage aortic endothelial impairment, thereby avoiding possible aortic dissection and delaying aortic rupture. In this study, a unique approach is explored, concentrating on targeted therapy to address intimal damage in the early phases of TAAD development. In the TAAD murine model, we have successfully engineered CAR-CD34⁺ HSPCs in vivo to specifically target aortic endothelium exhibiting elevated levels of VCAM-1 expression. Furthermore, the engineered cells possess the ability to secrete substantial quantities of VEGFA, which facilitates their proliferation, differentiation, and migration, thereby promoting reendothelialization and the restoration of intimal integrity. Consequently, CAR-CD34⁺ HSPCs may represent a promising therapeutic tool for the treatment of early TAAD. (Fig. 8).

Fig. 8. In vivo transformation of CAR-CD34⁺ HSPCs was achieved through the targeted delivery of CD34/LNP-CAR mRNA in a TAAD murine model.

Created in BioRender. Zhao, K. (2024) https://BioRender.com/o67q634.

The comprehension of the underlying mechanisms of TAAD remains elusive, presenting a significant challenge in the clinical management of this highly lethal condition³⁰. TAAD has been proven to be associated with heritable mutations, ECM, and SMC dysfunction, all of which lead to the fragile aortic wall^31,32. Recently, ECs have gained a lot of attention, which is believed to be the origin of TAAD³³. Diagbouga et al. discovered that ECs lacking SMAD6 exhibit dysregulation of genes related to cell junctions and impaired endothelial barrier function in the development of TAAD³⁴. The inhibition or alteration of ROBO4 expression in ECs leads to compromised barrier integrity and contributes to the pathogenesis of aortic dissection³⁵. Furthermore, a study reported that AT-1001, as an endothelial tight junction sealant, can ameliorate barrier dysfunction, thus blocking the incidence of TAAD⁶. In summary, the above studies indicated that the disruption of ECs may be a crucial early incident preceding TAAD formation. Developing approaches targeting ECs and repairing the endothelial barrier at this initial stage may be an effective way to prevent the progression of TAAD.

The maintenance of EC hemostasis is reportedly achieved by the presence of a stable population of CD34⁺ HSPCs in the bloodstream³⁶. Circulating progenitor cells play a role in maintaining endothelial homeostasis and have the potential to act as a reservoir for endothelial repair in various pathological states^37,38. The efficacy of EVs derived from MSCs and progenitor/stem cells has been demonstrated to yield the most favorable outcomes following myocardial infarction³⁹. These EVs mitigate adverse cardiac remodeling by downregulating fibrosis and promoting angiogenesis. Research showed that augmenting the migration and differentiation of Sca1⁺ vascular progenitor cells toward an EC lineage facilitates angiogenesis and mitigates vascular injury in ischemic tissues⁴⁰. Cheng et al. illustrated that the co-culturing of iPSC-derived ECs with cardiomyocytes in the post-myocardial infarction heart resulted in a substantial augmentation of graft size, heightened arterial blood supply, and enhanced cardiac function⁴¹. The powerful endothelium repair capacity exhibited by CD34⁺ HSPCs renders them a valuable tool in cardiovascular disease research. Numerous studies have established that vascular progenitor cells situated within the aortic adventitia contribute to the restoration of damaged aortic endothelium by migrating towards the endothelium and undergoing differentiation into an EC lineage⁴². Both endogenous resident cells and circulating HSPCs play a significant role in vascular remodeling after vascular injury⁴⁰. Employing circulating HSPCs might be an effective approach to repairing the early-stage endothelial barrier injury of TAAD.

Based on prior research, the direct infusion of progenitor cells may not result in a sufficient quantity of cells reaching the lesion site, thereby limiting the therapeutic efficacy of this approach^19,43. Furthermore, numerous vascular diseases, such as TAAD, are associated with reduced progenitor cells in the peripheral blood and compromised progenitor cells repair functions^44,45. To overcome these challenges, a number of studies have utilized engineering approaches to enhance their therapeutic efficacy in cardiovascular diseases. Overall, there are two primary ways to achieve this goal. One strategy is to enhance the functionalities of progenitor cells, such as proliferation and differentiation, etc. A recent study demonstrated that the sequential activation of Wnt, BMP4, VEGF, and NFATc1 signals induced iPSCs to generate substantial heart valve cells⁴⁶. Agonizts targeting these pathways may potentially promote the differentiation of progenitor/stem cells into cardiovascular cells. Yu et al. also used nanocarriers to deliver CXCR4 and VEGFA genes simultaneously to improve the proliferation, differentiation, and migration capabilities of endothelial progenitor cells (EPCs) and promote reendothelialization of carotid arteries²⁶. Another approach is to increase the number of progenitor cells at the lesion site, namely by enhancing their targeting and homing abilities. A 3D PGS-PCL vascular graft has been designed to attract CD93⁺/CD34⁺ circulating cells for reendothelialization, increasing antithrombogenicity and regeneration of small arteries⁴⁷. Shen et al. have utilized CD34-targeting antibody-conjugated platelets to capture circulating EPCs, leveraging the inherent infarct-homing properties of platelets to enhance EPC-mediated tissue repair⁴⁸.

CAR therapy is renowned for enabling cells to target specific molecules with high precision while simultaneously improving cell function⁴⁹. A study has utilized CAR-HPSCs to generate abundant and more stable CAR-neutrophils, thereby enhancing their anti-tumor effects⁵⁰. Furthermore, these cells derived from CAR-stem/progenitor cells were found to be devoid of graft versus host disease and cytokine release syndrome⁵¹. CAR-stem/progenitor cells have made significant strides in tumor treatment and are now showing potential in initial applications for inflammation-related diseases. To alleviate immune diseases, Olivia et al. designed CAR-modified MSCs to target tissues with abundant E-cadherin expression²⁴. This alteration led to the upregulation of immunosuppressive genes and receptors involved in T-cell inhibition. Given the feasibility of CAR for stem cells, we hypothesize that CAR-CD34⁺ HSPCs may exert vital effects in the treatment of TAAD. The CAR framework in our study was meticulously designed to align with the features of vascular inflammation. Specifically, the scFv was engineered to target VCAM-1. VCAM-1 has been generally recognized as a molecular marker of ECs for multiple cardiovascular disorders, such as atherosclerosis and acute myocardial infarction⁵². Our immunohistochemistry (IHC) and IF analyses revealed a significant increase in VCAM-1 expression within the intimal layer beginning on day 7, consistent with findings from previous studies⁵³. To enhance repair efficacy, the VEGFA sequence was chosen as the intracellular domain with the objective of upregulating VEGFA protein expression. This strategy is intended to augment the proliferation, differentiation into ECs, migration, and adhesion capabilities of CD34⁺ HSPCs^25,44,54,55. Moreover, the VEGFA domain contributes to preventing HSPCs from differentiating into immune cells, which may amplify the inflammatory response. Our in vitro experimental results demonstrated a significant enhancement in the reparative function of CD34⁺ HSPCs. The integrated transcriptomic and PPI network analysis reveals that after CD34/LNP-CAR mRNA treatment, a comprehensive regulatory landscape through which VEGFA orchestrates the differentiation of CD34⁺ HSPCs toward the endothelial lineage. The clustered heatmap demonstrates a distinct upregulation of metabolic and proliferation-associated genes in VEGFA-treated cells, including PKM, PGK1, ALDOA, and LDHA, which are key glycolytic enzymes⁵⁶. Glycolysis related genes play a key role in angiogenesis, research has revealed that the activation of PKM could promote angiogenesis and the differentiation of vascular resident EPCs into ECs by regulating glycolysis, mitochondrial fission, and fusion⁵⁷. The equilibrium between oxidative and non-oxidative glucose metabolism is essential for the proliferation and differentiation of HSPCs. The deficiency of PKM may lead to a reduction in metabolic intermediates necessary for biosynthesis, thereby impairing progenitor cell function. Conversely, the absence of LDHA markedly inhibits the functionality of HSPCs during the process of hematopoietic differentiation⁵⁸. These findings are consistent with previous reports indicating that VEGFA not only promotes endothelial identity, but also reprograms cellular metabolism to meet the bioenergetic and biosynthetic demands of differentiation and angiogenesis⁵⁹. The PPI network positions VEGFA at the core of a highly interconnected signaling module. Notably, VEGFA engages with several functionally distinct hubs, including PIK3R1, a critical effector of the PI3K/Akt pathway, as well as FGFR1 and JUN, which are linked to MAPK/ERK signaling. These interactions are collectively associated with the cellular proliferation of EPCs. Research has demonstrated that the proliferative effect of VEGFA on progenitor cells necessitates the activation of the ERK and Akt signaling pathways⁶⁰. In addition, it has been observed that CCND1 expression is upregulated in cells treated with CD34/LNP-CAR mRNA. CCND1 is known to facilitate cell cycle progression, promote the transition from the G0 to the G1 phase, and reduce the duration of the G1 phase, thereby leading to clonal expansion of EPCs⁶¹. Furthermore, the network underscores the significance of RELN, THBS1, ITGA3, and SDC1 as pivotal downstream effectors implicated in the remodeling of the extracellular matrix (ECM), cell adhesion, and vascular stabilization. These molecules are likely instrumental in the structural integration of ECs into emerging vascular networks. Notably, THBS1 acts as a convergence point between VEGFA signaling and ECM components, potentially facilitating the transition from a progenitor state to a functional vascular assembly⁶². In addition, our analysis identifies a metabolic circuit comprising SHMT2, MTHFD2, PHGDH, and PSAT1, which is enriched in one-carbon and serine-glycine pathways. These pathways are recognized for their roles in supporting nucleotide synthesis, redox homeostasis, and epigenetic programming⁶³, all of which are essential for the rapid proliferation and fate commitment of endothelial progenitors. The co-regulation of these metabolic enzymes with VEGFA signaling indicates a closely linked axis of metabolism-driven endothelial differentiation. Overall, these findings endorse a model wherein VEGFA not only initiates traditional angiogenic pathways, such as PI3K/Akt and MAPK/ERK, but also reconfigures cellular metabolism and cell-matrix interactions to promote effective endothelial lineage specification. The identified hub genes—PIK3R1, THBS1, RELN, and SHMT2—emerge as promising regulatory targets for augmenting endothelial differentiation and therapeutic vascularization.

In constructing the transmembrane (TM) and hinge domains (HD) within our CAR framework, we consulted established literature indicating that CD8, CD28, or IgG are commonly utilized for the HD, while CD8 or CD28 are typically employed for the TM domain in immunocytes⁶⁴. Research has demonstrated that the inclusion of a CD28 domain enhances the immunosuppressive effect following antigen-specific stimulation in CAR Tregs and CAR-MSCs^24,65. Consequently, we employed a similar strategy by incorporating the CD28 HD and TM domains into our CAR design to mitigate potential immune activation.

This study sought to achieve in vivo conversion of CAR-CD34⁺ HSPCs for precise targeting of aortic intimal injury. Although CAR-based therapies have demonstrated remarkable treatment efficacy, their applications are hindered by complex and costly manufacturing processes, as well as the risk of cytokine release syndrome^64,66. To address these issues, Yang et al. utilized the technique of dual mRNA co-delivery to directly generate phagocytosis-enhanced CAR-M for intensified liver cancer therapy⁶⁷. Similarly, to eliminate multidrug-resistant bacterial infections in septic mice, Tang et al. employed transient in situ conversion technology to generate CRV/LNP-CAR mRNA macrophages, enhancing the efficacy of CAR-M⁶⁸. Significant progress has also been made in the treatment of cardiac fibrosis. Researchers have injected LNPs expressing CD5 antibody and CAR-mRNA into the blood, resulting in the transient activation of anti-fibrotic CAR-T cells, which decreased post-injury fibrosis and restored heart function⁶⁹. These investigations underline the enormous promise of in vivo conversion, which not only minimizes adverse effects but also enables improved control over the therapeutic response⁴⁹. Besides, it is established that allogeneic MSCs, progenitor cells, and stem cells are prone to being taken as immunogens, potentially leading to immunological reactions⁷⁰. Conversely, the in vivo transformation of autologous CAR-CD34⁺ HSPCs could lead to increased functioning and a lower frequency of unfavorable consequences. In vivo genome editing of stem/progenitor cells has gained a lot of attention in recent years. A recent study has developed CD117/LNP mRNA targeting CD117 on HSPCs⁷¹. The delivery of this gene-editing system has demonstrated near-complete correction of hematopoietic sickle cells. Building on this concept, we employed an LNP-based mRNA delivery method to program CAR-CD34⁺ HSPCs in vivo for TAAD therapy. In vitro experiments demonstrated that a concentration of 0.2 µg/mL CD34/LNP-CAR mRNA optimally supports the survival of CAR-CD34⁺ HSPCs while ensuring consistent expression of anti-VCAM-1 scFv over a period of three days. Echocardiography was employed to assess the diameter of the aorta, facilitating the determination of the optimal timing for intervention using the LNP system. Notably, significant aortic dilation was observed by day 14, while Western blot and SEM analyses indicated the onset of endothelial barrier degradation as early as day 7. Consequently, targeted therapies were initiated on day 7. Flow cytometry analysis revealed the disappearance of CAR expression in peripheral blood 48 hours post-intravenous administration. Moreover, bioluminescence imaging demonstrated a substantial decrease in aortic fluorescence at the 48 h mark, suggesting that a two-day period may represent an optimal time window for intervention. The findings of the study also demonstrated favorable outcomes, with no animals subjected to in vivo conversion exhibiting adverse CAR reactions or major organ damage attributable to LNP administration.

However, in addition to the aorta, we also detected CAR-mRNA expression in the bone marrow, liver, lungs, and spleen via in vivo fluorescence imaging. This observation may be attributed to the non-specific targeting effects of CD34-targeting antibodies, suggesting that a certain amount of VEGFA proteins might be released from the lungs, spleen, and other blood vessels. As illustrated in Supplementary Fig. 16, the VEGFA levels in the peripheral blood of both the CD34/LNP-VEGFA mRNA and CD34/LNP-CAR mRNA groups were found to be significantly elevated compared to the control group, thereby confirming the presence of non-targeting effects. VEGFA insufficiency has been demonstrated to have a significant correlation with the initiation of aortic dilation diseases. Systemic administration of VEGF pathway inhibitors, including bevacizumab, has been associated with an elevated risk of aortic dissection⁷². Consequently, we utilized CD34/LNP-VEGFA mRNA in animal models to investigate the effects of circulating VEGFA on the progression of TAAD. Subsequent experiments in these models indicated that the CD34/LNP-VEGFA mRNA group neither inhibited aortic dilation nor promoted intimal repair. These findings suggest that increasing circulating VEGFA levels may have limited therapeutic efficacy in aortic repair.

IF staining results demonstrated that CD31 was localized within the intima of the normal mouse aorta, whereas CD34 exhibited high expression levels not only in the intima but also in the adventitia of the aorta. In saline, CD34/LNP, and CD34/LNP-VEGFA mRNA groups, the expression of CD31 and CD34 in the aortic intima was significantly diminished, while the expression of CD34 in the adventitia remained largely unaffected (Fig. 6h). These findings suggest the presence of an “intimal repair barrier” in addition to the intimal barrier, implying that “endothelial progenitor cells” may reside within the aortic ECs, capable of repairing the endothelium upon damage. Furthermore, BAPN not only compromises the inner membrane barrier but also disrupts the intrinsic intimal repair barrier. Following treatment with CD34/LNP-CAR mRNA, cells co-expressing CD34 and CD31 in the endothelium were restored. It is important to note that the fluorescence intensity of adventitia CD34 did not exhibit a significant difference between the two groups. This observation suggests that the treatment has a minimal effect on adventitia progenitor cells and predominantly promotes intimal repair.

However, the MC3-LNP employed in this study represents a relatively basic formulation that has received FDA approval for siRNA delivery, and is also widely used as a formulation for mRNA delivery^73–75. It must be acknowledged that it is an imperfect plant. On the fourteenth day following BAPN induction, the concentrations of TNF-α, IL-6, and MMP-2 were markedly increased in the group receiving CD34/LNP treatment compared to the saline-treated group (Supplementary Fig. 18c, d). It suggests that the CD34/LNP carrier itself may induce certain inflammatory responses, potentially attributable to the liposome nanoparticle characteristics of the MC3 formulation, warranting further investigation and optimization. Consequently, when employing the CD34/LNP delivery system, it is imperative to thoroughly assess its potential effects on the inflammatory response and to investigate safer and more efficacious therapeutic strategies. More advanced LNP formulations, such as SM-102/ALC-0315⁷⁶, have been developed, offering enhanced suitability for extrahepatic targeting. In this investigation, we utilized a CD34 antibody modification to mitigate the limitations associated with the extrahepatic targeting capacity of MC3-LNP. The primary objective of this study was to conduct a preliminary assessment of the feasibility and efficacy of LNP-mediated conversion of CAR-CD34 HSPCs in vivo, without incorporating the more recent SM-102/ALC-0315 formulations. Future research will focus on evaluating the impact of various LNP formulations on mRNA delivery efficiency. In addition, the development of nanomaterials that bind to antibodies targeting HPSCs, such as CD117, may mitigate the non-target effect, antibody-modified LNPs designed for targeting CD34⁺ cells in bone marrow, such as those facilitating LNP-mediated mRNA delivery to CD34⁺ cells in rhesus monkeys⁷⁷, have emerged and may offer improved efficacy for bone marrow targeting. These innovative materials will be integrated into future experiments to further optimize targeted delivery efficiency.

After excluding the potential confounding effects of VEGFA, our results demonstrated that treatment with CAR-CD34⁺ HSPCs restored intimal integrity, reduced elastin degradation, decreased collagen breakdown, and lowered the incidence of TAAD. Our single-cell transcriptomic analysis provides compelling evidence that CD34/LNP-CAR mRNA treatment profoundly reshapes the aortic cellular landscape in BAPN-induced mice, promoting vascular repair and modulating inflammation (Fig. 7). Compared to saline controls, CAR-CD34⁺ cell treatment increased the proportion of ECs while reducing fibroblasts, SMCs, macrophages, and neutrophils, suggesting enhanced endothelial restoration and suppression of pro-fibrotic and inflammatory responses. Marker gene and cell-cell interaction analyses further revealed that CD34/LNP-CAR mRNA mitigated the extensive inflammatory crosstalk observed in saline-treated samples, particularly between innate immune cells and vascular wall components, which likely contributes to a more reparative microenvironment. GO enrichment analysis highlighted robust activation of endothelial regenerative pathways, including angiogenesis, vasculogenesis, and cell adhesion dynamics—critical processes for intimal healing. Moreover, endothelial heterogeneity analysis uncovered an expansion of arterial and microvascular EC subtypes in the treatment group, implying a shift toward more functionally mature and reparative EC populations. In addition, the miRNA expression profiles in our study revealed four significantly differentially expressed miRNAs (miR-143-5p, miR-127-3p, miR-206-3p, and miR-137-3p) in the intervention group. The literature has documented that HSPCs attenuate the inflammatory response and promote repair through the secretion of EVs containing numerous miRNAs⁷⁸. Previous studies have discovered that miR-127-3p and miR-206-3p can be released by progenitor/stem cells and possess anti-inflammatory and reparative functions⁷⁹. miR-143-5p was also found to modulate PCSK9 expression by binding to its 3’UTR, thereby effectively attenuating the progression of atherosclerotic cardiovascular diseases⁸⁰. The miRNAs in EVs released by HSPCs may play a crucial role in the repair of the endothelium and the inhibition of TAAD inflammation. Developing techniques that target the aforementioned miRNAs offers tremendous potential for enhancing TAAD therapy. In summary, utilizing CD34-binding LNPs to anchor HSPCs resulted in the effective generation of CAR-CD34⁺ HSPCs and dramatically boosted the effectiveness of progenitor/stem cell treatment in aortic repair. Nevertheless, the fundamental mechanisms by which HSPCs enable early damage restoration are not yet entirely investigated, underscoring the need for future research to develop progenitor/stem cell-based therapeutic approaches.

Study limitations

Firstly, the FSC/SSC gating strategy used to define CD34⁺ cells has inherent limitations that may impact the accuracy of our analysis, this approach was primarily employed as a preliminary tool for morphological enrichment and screening rather than a highly precise method for antigen-specific sorting. Incorporating more refined, antigen-based sorting methods, such as fluorescence-activated cell sorting (FACS), in subsequent experiments may address these methodological constraints and enhance the accuracy of future analyses. Secondly, the therapeutic efficacy and safety profile of this targeted therapy necessitate prolonged treatment across various TAAD animal models. Thirdly, the non-specific target effects of CD34 antibodies may induce the secretion of VEGFA in organs such as the liver and lungs. It remains unconfirmed whether this has any impact on individuals with other comorbidities. The development of nanomaterials that bind to antibodies targeting HPSCs, such as CD117, may mitigate this effect. Finally, it is yet to be ascertained whether this targeted therapeutic strategy can be extrapolated to other cardiovascular diseases.

Methods

Design of the CAR plasmid construct and lentiviral transduction

It has been found that CAR-MSCs incorporating the CD28 hinge and TM exhibit enhanced immunosuppressive properties at inflammation sites associated with graft-versus-host disease²⁴. In this study, we selected CD28 for the TM to avoid potential activation of the immune response. The CAR construct was comprised of an scFv targeting VCAM-1 (IgG-scFv as control), a CD28 hinge, a CD28 transmembrane, and a VEGFA activation domain (Supplementary Fig. 22a, b). A lentivirus was derived from HEK293T cells’ supernatant. The CAR construct was co-transfected with lentiviral mix (10 L) and transgene reagents (60 L) using a lentiviral packaging kit (Genomeditech, Shanghai, China) for a duration of 2 days. Subsequently, viral stocks were obtained from the culture medium and filtered to exclude non-adherent cells. To obtain CD34⁺ HSPCs consistently expressing CAR constructs, cells were grown in StemPro-34 medium supplemented with 10% fetal bovine serum (FBS) and 2.5 μg/ml puromycin for 48 h after lentiviral transduction. Following cell selection, stable, infected pooled clones were collected for in vitro experiments.

CAR mRNA in vitro transcription and encapsulation in LNP

The CAR plasmids harboring the luciferase (Luc) genes utilized for in vitro CAR mRNA transcription were provided by VectorBuilder (Guangzhou, China). These plasmids were engineered utilizing the pmRVac vector, which encompasses the T7 promoter, 5’UTR, 3’ UTR, and poly A tail (Supplementary Fig. 23). The synthesis of mRNAs from linearized plasmids was conducted in a laboratory environment using MEGAScript T7 kits supplied by VectorBuilder in Guangzhou, China. During the transcription process, UTP was substituted with m1Ψ-5’-triphosphate (TriLink, N-1081), and a 5’ cap was subsequently appended utilizing CleanCap (TriLink, N-7113). The synthesized mRNAs were then subjected to analysis via denaturing agarose gel electrophoresis. Then the resulting CAR mRNA was subsequently purified by a Megaclear kit. A Cy5-labeled Luc-mRNA was chosen as a representative mRNA to assess the impact on LNP encapsulation, endo-lysosomal escape, and subsequent mRNA transfection efficiency.

LNPs used in this study encapsulating mRNA typically include several major components: cholesterol, helper lipids, and ionizable lipids⁸¹. mRNA strands were further purified by cellulose columns, stored at − 20 °C, and encapsulated into LNPs using Ignite NanoAssembler (Precision Nano Systems Inc., Vancouver, Canada). In particular, mRNAs were prepared in 50 mM sodium citrate buffer (pH = 4), and lipids consisted of DLin-MC3-DMA, DSPC, cholesterol, and DSPE-PEG₂₀₀₀ in a molar ratio of 50:10:38.5:1.5, which has been approved by the Food and Drug Administration (FDA) for siRNA delivery⁸², were dissolved in ethanol. The Ignite NanoAssembler was applied to coalesce mRNA and lipids at a 4:1 ratio (aqueous phase: organic phase) while maintaining a total flow speed of 6 ml/min. CD34-antibody-modified LNP was synthesized by the post-insertion method⁸³. Specifically, the pre-formed LNP were incubated with the CD34 monoclonal antibody at a 1:1 molar ratio (antibody to lipid) in a final volume of 2 mL of PBS buffer (pH 7.4). This antibody was pre-conjugated to a lipid moiety (DSPE-PEG₂₀₀₀-CD34). The antibody was inserted into the LNP surface by incubating the mixture at room temperature for 2 h, ensuring that the antibody interacted with the lipid bilayer via hydrophobic or electrostatic interactions. During this process, the pH of the solution was maintained around 7.4, and the temperature was kept at 25 °C to facilitate efficient insertion of the antibody into the LNPs membrane. After the incubation, the LNPs were purified to remove unincorporated antibody using size exclusion chromatography to isolate the CD34 antibody-modified LNPs.

Subsequently, the combination underwent dialysis against PBS for 24 hours with three buffer replacements, followed by dilution with water. Based on Zetasizer Nano software (Malvern Panalytical, England), we analyzed the DLS and zeta potential of the synthesized LNPs post-dialysis. A Qubit fluorometer (Thermo Fisher Scientific, Q33226) was used to determine the quantity of mRNA contained in the LNP. To mimic in vivo stability, LNP was evaluated in 10% FBS at 37 °C to measure its size and PDI. To preserve LNP stability over time, we combined it with sucrose at a concentration of 10% and stored it at − 80 °C for short-term storage (up to one month).

Cell lines

Mononuclear cells were extracted from the femoral marrow of male C57BL/6 J mice using density gradient centrifugation (Cedarlane, Burlington, Canada). The CD34⁺ cells were tagged with MACS CD34 MicroBeads (Miltenyi Biotec, Auburn, Shanghai, China) and isolated using an LS column. For further investigation of CD34/LNP-CAR mRNA effects on cellular function, CD34⁺ cells were cultured in Dulbecco modified Eagle medium (DMEM) (Gibco, US), which was supplemented with 10 ng/mL VEGFA for 2 days to simulate the interaction between CD34⁺ cells and the vascular endothelial microenvironment. Select cells that have already adhered to the substrate for subsequent experimental procedures.

Safety of LNP transfection

96-well plates were incubated with CD34⁺ cells at a concentration of 1 × 10⁴ cells per well and allowed to adhere overnight in preparation for cytotoxicity testing. The cells were exposed to various treatments, such as PBS, LNP, CD34/LNP, LNP-CAR mRNA, and CD34/LNP-CAR mRNA at doses ranging from 0.1 to 0.8 μg/ml. Following a 6-hour exposure, fresh complete media (200 μl per well) was introduced, and the cells were then incubated for an additional 18 h. Cell viability was tested using the CCK-8 kit according to the directions supplied by the manufacturer.

Examination of endolysosomal escape and cellular uptake

LNP-CAR mRNAs encapsulated in Cy5 labeling were applied to CD34⁺ cells for 1, 3, and 6 h, followed by 30 min of Hoechst 33342 and Lysotracker Green incubation at 37 °C. The cells were immobilized using 4% paraformaldehyde (PFA) and then examined by confocal laser scanning microscopy (CLSM) (Nikon, Tokyo, Japan). To quantify the co-localization of LNP-CAR mRNAs and Lysotracker Green inside the cells, Manders’ co-localization coefficients were determined using ImageJ software. A 24-well plate was initially seeded with CD34⁺ cells and permitted to adhere overnight. Subsequently, PBS, free mRNA (Fr-mRNA), LNP-CAR mRNA, and CD34/LNP-CAR mRNA were incubated with the CD34⁺ cells for 4 h. Following this incubation time, the cells underwent three washes with PBS before being submitted to examination by flow cytometry. DIO (green) and DAPI (blue) were applied as fluorescent dyes to identify the cell membrane and nucleus, respectively, prior to CLSM examination for qualitative assessment.

CD34/LNP-CAR mRNA transfection in vitro and CAR expression analysis

The CD34⁺ cells in the logarithmic growth phase were injected into 96-well plates at a density of 10⁴ cells per well and transfected with PBS, lentiviral vectors overexpressing the IgG-scFv-CAR plasmid (LV-IgG-CAR), lentiviral vectors overexpressing the anti-VCAM-1-scFv-CAR plasmid (LV-VCAM-1-CAR), CD34/LNP (0.1 μg/ml), and LNP-CAR mRNA (0.1 μg/ml). Then, different concentrations of CD34/LNP-CAR mRNA were administered to assess the optimal transfection efficacy. Subsequently, flow cytometry was utilized to determine the transfection efficiency, with CAR expression determined by recombinant mouse VCAM-1 (c-Fc), which was visualized with anti-IgG (Fc)-FITC. The expression of VEGFA was evaluated by Western blot analysis using the anti-VEGFA antibody.

Detection of cell proliferation and differentiation

To measure the proliferation of cells, a 50 nM EdU solution was utilized and then fixed in 4% PFA according to the manufacturer’s method (Beyotime, Beijing, China), followed by staining with Hochest 33342 staining nuclei (1 mg/ml, Beyotime, Beijing, China). A fluorescent microscope was used to capture images of cells that were positive for EdU.

To assess the influence of CAR plasmid and CD34/LNP-CAR mRNA on the CD34⁺ cell differentiation in vitro, we transfected 5 × 10⁵ freshly isolated cells with PBS, CD34/LNP, CAR plasmid lentivirus, and CD34/LNP-CAR mRNA. After 3 days of inducement, these cells were prompted to undergo differentiation. Flow cytometry was used to analyze the number of endothelial markers of CD31 and VE-CAD expression in CD34⁺ cells.

In vitro migration of CAR-CD34⁺ cells

Wound-healing and transwell assays were performed to determine the migration of CD34⁺ cells. The wound-healing assay consists of a culture of 2 × 10⁵ CD34⁺ cells in six-well plates, with a p20 pipette applied to induce a scratch in the monolayer. CD34⁺ cells without CAR plasmid transfection were treated with the same operation, followed by washing once and supplementation with culture media containing PBS, CD34/LNP (0.2 mg/ml), and CD34/LNP-CAR mRNA (0.2 mg/ml). An optical microscope was employed to analyze transmigrated cells after a 12 h incubation period at 37 °C. Transwell experiments comprised the digestion and suspension of CD34⁺ cells in serum-free media, followed by the loading of 2 × 10⁴ cells into the top chambers of the transwell chamber with 8 μm-sized holes (Corning, China). The bottom chambers were filled with PBS-containing culture medium, CD34/LNP (0.2 μg/ml), and CD34/LNP-CAR mRNA (0.2 μg/ml). Subsequently, each well was stained with crystal violet after a 6 h incubation at 37 °C, and cell counts were undertaken in five random areas.

In vitro adhesion and tube formation of CAR-CD34⁺ cells

The CD34⁺ cell adhesion test was conducted using fibronectin-coated dishes. Different groups of CD34⁺ cells were stimulated with PBS, CD34/LNP, and CD34/LNP-CAR mRNA for 6 hours at 37°C. CD34⁺ cells transfected with the CAR plasmid underwent the same procedures. In each trial, unattached cells were eliminated using PBS, while adherent cells were immobilized with 4% PFA and then stained with 0.3% crystal violet. A growth factor-depleted Matrigel (Corning, USA) was thawed overnight at 4 °C and subsequently evenly distributed in 96-well plates at a consistent level to induce tube formation. CD34⁺ cells at a concentration of 2 × 10⁴ were resuspended in culture media and subsequently loaded onto the top of the Matrigel. The experimental groups were comprised of three wells, which were observed under the microscope following a 12 h incubation at 37 °C. The average quantity of tubules in each well was calculated based on observations from five randomly selected fields.

Functional evaluation of CD34⁺ cells by CFU

To quantify CFUs, CD34⁺ cells were examined under an inverted microscope after seven days of culture. The results were normalized to the mean number of colonies generated per 10⁶ plated cells. The average colony count was determined manually by two independent observers from ten replicate wells using a light microscope.

Echocardiography and SEM

Echocardiograms were conducted on days 7, 14, and 21 to assess the maximal diameter of the aorta in mice. The measurement of each indicator was taken meticulously across three cardiac cycles, and the mean value was later determined.

The ascending or thoracic aorta from a mouse was dissected into 3 mm rings and then immersed in a 100 mmol/L cacodylate buffer solution with a pH of 7.4, supplemented with 3% glutaraldehyde for a duration of three hours. Subsequently, the tissues underwent a series of processing steps: washing in cacodylate buffer, post-fixation in 1% OsO₄, dehydration in an ethanol series, infiltration with propylene oxide, and embedding in Epon 812. Ultrathin sections were then cut using an ultramicrotome (LKB Ultramicrotome, Bromma, Sweden) and examined with scanning electron microscopy (JEOL, Tokyo, Japan).

Animal models

Male C57BL/6 J mice aged 3 weeks were purchased from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and housed at the Navy Medical University’s Animal Experiment Center in Shanghai, China. The mice were kept under standard laboratory conditions with free access to food and water. The environment was maintained on a 12 h light-dark cycle, with temperature controlled at 18–22 °C and humidity at 50–70%. The current study received approval from the Animal Research Ethics Committee of the Navy Medical University First-Affiliated Hospital (CHEC2021-010), and all protocols and regulations pertaining to animal experimentation were strictly adhered to.

A mouse model of TAAD was established using BAPN purchased from Sigma-Aldrich (St. Louis, MO, USA). Male C57BL/6 J mice aged three weeks were given BAPN (1 g/kg/day) for 21 consecutive days to establish the TAAD model⁸⁴, whereas mice aged three weeks were given regular chow. Aortic dissection is defined as an accumulation of blood within the medial layer of the aorta, leading to the appearance of a false lumen. An aortic aneurysm is defined as the diameter of the aorta surpassing 50% of its baseline measurement.

In our study, we identified three specific time points for detection (days 7, 14, and 21) and chose days 7 and 14 to symbolize the initial and early phases preceding the formation of TAAD, based on our experimental findings. In the end, for histology detection, we injected mice with an overdose of sodium pentobarbital, perfused their hearts with PBS, and fixed them with 4% PFA; for WB, qRT-PCR, or mRNA sequence analysis, the aortas were stored at − 80 °C after cryopreservation. Animals that died during the TAAD model inducement were also managed timely, as described above.

To investigate the in vivo effects of CD34/LNP-CAR mRNA, male C57BL/6 J mice underwent BAPN inducement and were simultaneously treated with saline, CD34/LNP (10 μg), CD34/LNP-VEGFA mRNA (10 μg), or CD34/LNP-CAR mRNA (10 μg) via intravenous injection through the tail vein (n = 12 per group), with mice without BAPN inducement as the control group.

Western blot analysis

The cell or tissue lysates underwent electrophoresis on an 8% SDS-PAGE gel for Western blot analysis, followed by transfer onto nitrocellulose membranes. The proteins were then separated on the gel and transferred onto a PVDF membrane (Millipore, Shanghai, China) using an electric current of 350 mA for 70 min. Subsequently, the membranes were blocked with 5% skim milk (Solarbio, Beijing, China) in PBST and incubated overnight at 4 °C with primary antibodies in 1% skim milk. Finally, the membranes were subjected to secondary antibodies. The ChemiDoc Imaging System made by Bio-Rad was applied for image capture, and the chemiluminescence (ECL) kit (Beyotime, China) was employed to improve the quality of the images. The membrane was incubated overnight at 4 °C with primary antibodies, including anti-VEGFA from Rabbit (1:1000, ab46154, abcam), anti-CD34 (RM2052) from Rabbit (1:1000, ab317589, abcam), anti-ICAM-1 (EPR16608) from Rabbit (1:1000, ab179707, abcam), ZO-1 from Rabbit (1:1000, ab96587, abcam), anti-VE-CAD (EPR18229) from Rabbit (1:1000, ab205336, abcam), anti-IL-6 (EPR23819-103) from Rabbit (1:1000, ab290735, abcam), anti-TNF-α(EPR19147) from Rabbit (1:1000, ab183218, abcam), anti-P-selectin (EPR17012-27) from Rabbit (1:1000, ab253240, abcam), anti-E-selectin from Rabbit (1:1000, ab185698, abcam), anti-GAPDH from Rabbit (1:3000, ab8245, abcam), anti-IL-1β(E7-2-hIL1β) from Mouse (1:1000, sc-32294, Santa Cruz Biotechnology), anti-MMP-9 (7-11 C) from Mouse (1:1000, sc-13520, Santa Cruz Biotechnology), anti-VCAM-1 (E-10) from Mouse (1:1000, sc-13160, Santa Cruz Biotechnology), and anti-CD31 from Mouse (1:1000, AF3628, R&D Company). Next, the membrane was incubated with HRP-conjugated Goat anti-rabbit IgG (1:5000, #SA00001-2, Proteintech) and HRP-conjugated Goat anti-mouse IgG (1:5000, #SA00001-1, Proteintech).

HE, Masson, and EVG staining

Tissues from the aorta of male C57BL/6 J mice were fixed with 4% PFA for over 24 h. The tissues were then embedded in paraffin and sectioned into 4 μm slices. HE staining was performed on these sections, and Masson’s trichrome staining with a Trichrome Stain (Masson) Kit (Sigma-Aldrich) was applied to the aortic media to indicate the existence of collagen. Elastin fibers in the tissue sections were visualized using EVG staining. The elastin fibers are specifically stained with resorcin-fuchsin (Sigma-Aldrich, Shanghai, China) and methylene blue (Sigma-Aldrich, Shanghai, China) solutions.

An evaluation of pathological scores for elastin degradation in the aorta was conducted blindly. Scores were assigned according to specific criteria: a score of 1 for degradation levels below 25%, a score of 2 for degradation levels between 25% and 50%, a score of 3 for degradation levels between 50% and 75%, and a score of 4 for degradation levels exceeding 75%⁸⁵. The breakdown of elastin was determined in non-adjacent sections in the core region of each sample (every fifth section). Each specimen within a category was averaged to determine the average score.

Histochemical staining

Tissues from the aorta of male C57BL/6 J mice were fixed with 4% PFA for over 24 h. The tissues were then embedded in paraffin and sectioned into 4 μm slices. We stained them with VCAM-1 (sc-13160) for the IHC analyses⁸⁶. For the IF assay, aorta sections were stained with antibodies against CD34 (Abcam, ab317589) and CD31 (R&D, AF3628) using a fluorescence immunostaining procedure. FBS in PBS was used to prepare and block the sections for one hour. Primary antibodies were incubated overnight at 4 °C, followed by fluorescent secondary antibodies for 45 min, then DAPI/Hoechst for 10 min at room temperature. Ultimately, the results were evaluated using Image Pro Plus 3.0 (Nikon, Tokyo, Japan).

Flow cytometry

For cells in vitro, CD34⁺ cells subjected to various treatments were incubated with the specified antibodies as mentioned before in a 100 ul MACS buffer (comprising PBS, 0.5 mM EDTA, and 0.5% BSA) at 4 °C for a duration of 20–30 min, subsequently washed, and reconstituted in a 300 μl MACS buffer. Flow cytometric analysis was conducted using FlowJo 10.8 software (BD Biosciences). We drew peripheral blood for flow cytometry to validate the transformation of CAR-CD34⁺ cells in circulation. CD34⁺ cells in the blood were isolated in the same way as from the bone marrow, and cells were treated using the same procedures as in vitro cytometry. The expression of the mouse His-tag was detected using a rabbit monoclonal anti-His tag antibody. The gating strategies employed in the flow cytometry analyses for the cell markers CD31-FITC and VE-CAD-PE, and VCAM-1 CAR conjugated with Alexa Fluor 647 are illustrated in Supplementary Fig. 24a, b. The following antibodies were used: anti-CD34-FITC(581) from mouse (ab131589, Abcam), recombinant VCAM-1 (C-Fc) from mouse (Cat. No.:CU39, Novoprotein), anti-IgG (Fc)-FITC from mouse (F5387, Sigma), anti-CD31-FITC from mouse (#561813, BD), anti-VE-CAD-PE from mouse (12-1441-82, eBioscience™), and anti-6xHis tag-Alexa Fluor 647(EPR20547) from rabbit (ab237337, Abcam). Cell viability was assessed using the LIVE/DEAD™ Fixable Violet Dead Cell Stain Kit (Thermo Fisher Scientific, Cat. No. L34955) according to the manufacturer’s instructions. Briefly, cells were washed twice with PBS and incubated with the viability dye (FL1-A:: V450-A) for 30 min at room temperature, protected from light. After staining, cells were washed and resuspended in flow cytometry buffer for subsequent analysis. The fixable nature of the dye allowed for downstream fixation and permeabilization steps without loss of viability discrimination.

Tissue luciferase assay

A steady-glo luciferase assay (Promega) was used to assess luciferase expression in Tissues from male C57BL/6 J mice tissues 24 and 48 h following saline, LNP-CAR mRNA, CD34/LNP-VEGFA mRNA, and CD34/LNP-CAR mRNA treatments. We weighed and homogenized the tissues three times in 500 ml of GLO lysis buffer using a FastPrep homogenizer (ThermoScientific) set to speed “6”. Samples homogenized in GLO lysis buffer were diluted and read on a microplate reader (Bio-Rad, Shanghai, China) using a new white plate.

In vivo biocompatibility evaluation

To investigate the possible in vivo adverse effects of the targeted CD34/LNP-CAR mRNA, different major organs, including the heart, kidney, lung, liver, brain, and spleen, were harvested 24 h post-administration of saline, LNP-CAR mRNA, CD34/LNP, and CD34/LNP-CAR mRNA in male C57BL/6 J mice. HE staining was applied in the histological study of tissue slices from these organs. In addition, blood samples were drawn from male C57BL/6 J mice via the retro-orbital route 24 h following intravenous injection of saline or CD34/LNP-CAR mRNA at a dose of 10 μg per mouse. Serum levels of ALT, AST, LDH, and BUN were then analyzed using a biochemical analyzer (Pointcare M4). Hemoglobin and RBC counts were assessed by an automatic blood analyzer (Sysmex XE 5000, Kobe, Japan).

ELISA assay

Aorta tissues from male C57BL/6 J mice were processed for analysis using ELISA to quantify levels of MMP-9 (MMPT90), TNF-α (MTA00B-1), IL-6 (M6000B-1), and IL-1β (MLB00C-1). In brief, the tissues were homogenized and sonicated in normal saline to generate a 2.5% homogenate, which was then centrifuged at 3000 × g for 10 min to collect the supernatant for analysis. ELISA was carried out using ELISA kits manufactured by RD Company (Minnesota, USA).

RNA extraction and qRT-PCR

TRIzol (Invitrogen) was employed for the extraction of RNA from male C57BL/6 J mice aortic tissues, followed by the generation of cDNA from the entire RNA using the High-Capacity cDNA Reverse Transcription Kit (Invitrogen) from Applied Biosystems. Gene expression in aorta tissue was measured via qRT-PCR with SYBR Green (Yeasen, Shanghai, China), employing DNA primers specific to each gene based on the mouse genome. The quantification of all cDNAs was done using the Bio-Rad MyiQ single-color real-time PCR equipment, with experiments completed in triplicate to confirm the reproducibility of the results. Supplementary Table 1 provides primer sequences.

Single cell sequencing and miRNA sequencing analysis

Aortas were isolated from male C57BL/6 J mice induced with BAPN, with one group concurrently receiving saline (n = 3; 6 aortas per sample) and another group receiving CD34/LNP-CAR mRNA treatment for a duration of 7 days (n = 3; 6 aortas per sample). Following scRNA-seq, one sample from the CD34/LNP-CAR mRNA treatment group (C3) was identified as an outlier due to poor data quality (characterized by an extremely low number of captured cells and a high mitochondrial gene percentage) and was excluded from downstream analysis. All procedures related to single-cell capture and library preparation were performed by LC-Bio Technology Co., Ltd. (Hangzhou, China). Initially, murine aortic tissues were rinsed with ice-cold normal saline supplemented with 3% fetal bovine serum (FBS; Thermo Fisher Scientific, Inc.) to eliminate residual blood clots. Subsequently, the tissues were transferred to sterile, RNase-free culture dishes containing ice-cold, calcium- and magnesium-free 1 × PBS. The tissues were then finely minced and enzymatically dissociated into single-cell suspensions according to the standard protocol of the 10 × Genomics platform. Following centrifugation, the supernatant was removed, and the cell pellet was resuspended in red blood cell lysis buffer (Solarbio) and incubated on ice for 2 min to lyse erythrocytes. Total RNA was extracted and purified from aortic tissue using TRIzol reagent (Invitrogen, CA), followed by quality assessment of RNA quantity and integrity. Small RNA libraries were constructed according to standard protocols and subjected to deep sequencing using the Illumina platform (NovaseqTM 6000; LC Bio-Technology Co., Ltd., Hangzhou, China). Sequencing data were processed and analyzed to quantify miRNA expression levels. Differential expression analysis was performed using normalized read counts and various statistical methods, including Fisher’s exact test, Student’s t test, ANOVA, the Chi-squared 2 × 2 test, and the Chi-squared N × N test. miRNAs with high expression levels were identified based on a significance threshold of p < 0.05.

Statistical analysis

The data in this study are reported as the mean ± standard deviation. In each experiment, n defines the number of mice included. Statistical analyses included a two-tailed Student’s t test for comparing differences between two groups; a one-way ANOVA followed by Tukey’s post hoc test for comparisons among more than two groups; and a two-way time-varying ANOVA and t-tests for analyzing time-related differences between groups. Statistical significance was defined as p < 0.05.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

J.Z. and K.Z. conceived the project. K.Z., Y.H. and R.Y. designed the experiments. Y.H. performed most of the in vitro experiments. K.Z., Y.H. and S.L. wrote the manuscript. K.Z., J.N. and L.K. performed the animal experiments. P.D. provided critical reagents and scientific input. H.Z., Z.Z., R.Z. and T.L. contributed to the in vivo study design. Conceptualization, Writing – Review & Editing, and Supervision: Z.J. All authors have read and approved the final manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data supporting the findings of this manuscript are available within the manuscript and its supplementary information. Sequencing data have been deposited in the NCBI SRA public database under the accession number PRJNA1191614. The mRNA sequence accession numbers are SRR31572590; SRR31572591; SRR31572592; SRR31572593; SRR31572594 and SRR31572595, and the miRNA sequence accession numbers are SRR31572588 and SRR31572589. The raw single-cell RNA sequencing data have been deposited in the SRA at the NCBI under BioProject accession PRJNA1448865. The scRNA-seq accession numbers are SRX32830225, SRX32830226, SRX32830228, SRX32830229, and SRX32830230. Any additional raw data will be available from the corresponding author upon reasonable request. Source data are provided in this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-72203-3.