Sustained suppression of choroidal neovascularization by intraocularly stable tetrahedral network encapsulated miR-22–3p

Abstract

Choroidal neovascularization (CNV), characterized by abnormal vessel growth and vascular leakage, is the hallmark of wet age-related macular degeneration (wAMD) and a leading cause of irreversible vision loss. Although anti-vascular endothelial growth factor (VEGF) therapies remain the current standard, their frequent administration and limited long-term efficacy highlight the need for novel treatments. Here, we developed a miR-22-3p-loaded tetrahedral framework nucleic acids (tFNAs-miR22) nanostructure and evaluated its efficacy in CNV suppression. The nanocomplex was structurally validated, exhibiting high assembly fidelity and superior intraocular stability compared to serum conditions. In a laser-induced CNV mouse model, a single intravitreal injection of tFNAs-miR22 significantly reduced lesion size and leakage by day 10, with efficacy comparable to aflibercept. In a rat model of stable and long-lasting CNV, tFNAs-miR22 demonstrated durable inhibition of vascular leakage by Day 21, showing greater persistence compared to aflibercept. This effect was dose-dependent, with the high-dose group outperforming aflibercept in suppressing leakage. Transcriptomic profiling of hypoxia-challenged HUVECs further revealed that tFNAs-miR22 modulates angiogenic pathways, including suppression of the phosphoinositide 3-kinase (PI3K)/protein kinase B (AKT)/mechanistic target of rapamycin (mTOR) axis. These findings demonstrate the potent and long-lasting therapeutic effects of tFNAs-miR22, supporting its promise as a next-generation, gene-regulatory nanotherapy for sustained inhibition of CNV.

Graphical abstract

Highlights

• •tFNAs serve as a stable DNA nanocarrier for miR-22–3p intraocular delivery.
• •Single injection of tFNAs-miR22 reduces CNV lesion size and leakage in mice.
• •Dose-dependent and durable efficacy demonstrated in chronic rat CNV model.
• •tFNAs-miR22 modulates angiogenesis via suppression of PI3K/AKT/mTOR signaling.

1. Introduction

Choroidal neovascularization (CNV), characterized by abnormal vessel growth and vascular leakage, is the hallmark of wet age-related macular degeneration (wAMD), which is the leading cause of irreversible central vision loss among individuals over the age of 50. This condition significantly impairs quality of life and imposes a substantial growing burden on global healthcare systems [1,2]. Epidemiological studies estimate that with the accelerating aging of the global population [3], the prevalence of wAMD is expected to increase sharply in the coming decades [4]. The abnormal choroidal neovascularization are fragile and prone to leakage, hemorrhage, fibrosis, and eventual photoreceptor degeneration and scarring, and these changes may be associated with abnormal vascular endothelial growth as well as RPE inflammation and oxidative stress response [5,6].

Currently, intravitreal injection of anti-vascular endothelial growth factor (VEGF) therapies remains the gold standard treatment for wAMD, although these therapies have dramatically improved visual outcomes, they are not without limitations [7]. A substantial proportion of patients exhibit suboptimal or incomplete responses, and the therapeutic effects are often transient, necessitating repeated, lifelong injections to maintain visual stability [8,9]. Moreover, VEGF blockade may disrupt normal vascular homeostasis and contribute to geographic atrophy or macular ischemia over time [10,11]. These challenges underscore the need for alternative or adjunctive therapeutic strategies that provide more sustained efficacy with fewer interventions.

MicroRNAs (miRNAs) are small, non-coding RNAs that fine-tune gene expression post-transcriptionally and regulate a wide array of cellular processes, including angiogenesis, oxidative stress, and inflammation-all central to the pathogenesis of CNV in wAMD [12]. Among them, miR-22–3p has attracted growing interest due to its multifaceted biological functions [13]. It has been demonstrated that miR-22–3p can inhibit angiogenesis by downregulating proangiogenic factors such as VEGF [14], suppress oxidative stress by targeting Reactive Oxygen Species (ROS)-generating pathways [15] and modulate inflammatory cascades [16]. These pleiotropic effects make miR-22–3p an attractive candidate for therapeutic intervention in wAMD. However, the clinical translation of miRNA-based therapies remains constrained by delivery challenges, including enzymatic degradation, limited cellular uptake, and poor tissue specificity [17].

To address these obstacles, Tetrahedral Framework Nucleic Acids (tFNAs) have gained attention as innovative nanocarriers for nucleic acid delivery [18]. These self-assembled, three-dimensional DNA nanostructures are biocompatible, highly stable, and capable of encapsulating and delivering nucleic acids to cells with high efficiency [18,19]. Their nanoscale architecture allows for efficient cellular uptake without transfection reagents, while their modular design enables functionalization with targeting moieties or therapeutic cargos [20]. Furthermore, recent studies have suggested that tFNAs themselves may exhibit bioactive properties, such as antioxidative and anti-inflammatory effects, which could further enhance their therapeutic potential [21]. While preliminary investigations have demonstrated the potential of tFNAs in ocular drug delivery [[22], [23], [24]], their application in posterior segment diseases, especially those involving CNV such as wAMD, remains at an early stage and warrants further systematic exploration [25].

In this study, we investigate the therapeutic potential of a miR-22-3p-loaded tFNA nanocomplex (tFNAs-miR22) for the treatment of CNV. Using a laser-induced CNV mouse model, we evaluate the efficacy of tFNAs-miR22 in reducing neovascularization, a hallmark of wAMD [26]. In parallel, we examine its effects on oxidative stress in ARPE-19 cells and anti-angiogenic activity in human umbilical vein endothelial cells (HUVECs), both of which are widely employed in vitro models for studying wAMD pathophysiology [27,28]. To gain mechanistic insights, we selected HUVECs for transcriptomic analysis before and after tFNAs-miR22 treatment, identifying key differentially expressed genes and pathways, because they serve as a widely recognized in vitro model for angiogenesis [29]. (Fig. 1). Based on these findings, we further explore the molecular mechanisms.

Fig. 1.

Schematic illustration of the assembly and therapeutic application of tetrahedral framework nucleic acids loaded with miR-22–3p (tFNAs-miR22). Proposed mechanism of action: following cellular uptake, tFNAs-miR22 downregulates the PI3K/AKT/mTORC signaling pathway by inhibiting phosphorylated PI3K, AKT, and mTORC components.

2. Results and discussion

2.1. Synthesis and characterization of tFNAs-miR22

The successful production of equimolar quantities of four single-stranded DNA (ssDNA), as indicated in Fig. 2A, led to the formation of the miR-22-loaded tetrahedral nanostructure (tFNAs-miR22). Each ssDNA was designed to fold into three complementary segments that self-assembled into a well-defined triangular framework via precise Watson–Crick base pairing. This structural programmability ensures uniformity and reproducibility, essential for downstream biomedical applications [30]. A schematic of the assembly process is shown in Fig. 2B. In the single-stranded DNA molecules, we engineered the S1 strand to contain the miR-22–3p sequence in subsequent experimental analyses. This design facilitates targeted delivery, improving the reliability of biological assessments [31]. To confirm the formation of tFNAs and tFNAs-miR22, high-performance capillary electrophoresis (HPCE) confirmed that four ssDNA molecules were present, aligning with previous theoretical values (Fig. 2C), indicating high structural integrity of the nanocomplex [32]. This result reflects the successful and precise hybridization essential for maintaining the three-dimensional architecture of tFNAs [33]. The formation of tFNAs-miR22 was further validated by 8 % polyacrylamide gel electrophoresis (PAGE), which revealed distinct migration patterns corresponding to the assembled nanostructures (Fig. 2D). The agreement between HPCE and PAGE results confirms both purity and assembly fidelity, providing strong evidence of structural consistency across characterization methods [32,34]. Transmission electron microscopy (TEM) imaging revealed that the synthesized tFNAs-miR22 retained a uniform, triangular geometry consistent with prior reports (Fig. 2E), reinforcing the predictable morphology of tFNAs [35]. The particle size of tFNAs-miR22 was approximately 12.84 nm, slightly larger than that of unloaded tFNAs (11.91 nm), indicating successful incorporation of miR-22–3p into the tetrahedral framework (Fig. 2F). In parallel, zeta potential measurements revealed a shift from −11.3 mV (tFNAs) to −18.8 mV (tFNAs-miR22), consistent with the addition of negatively charged miRNA (Fig. 2G). This increased surface charge enhances colloidal stability by promoting electrostatic repulsion, thereby preventing undesired aggregation and ensuring that the nanocomplex remains structurally stable and functionally intact under physiological conditions [36].

Fig. 2.

Synthesis, structural validation, and stability of tFNAs-miR22. (A) Each ssDNA was designed with three complementary segments. (B) Schematic illustration of the one-step annealing of four single-stranded DNAs (S1-S4, S1 carrying miR-22–3p) into a tetrahedral framework nucleic acid and the analytical techniques used for characterization (capillary electrophoresis, PAGE, TEM, HPLC). (C) HPCE showing the expected size for each individual strand, the unloaded tetrahedron (tFNAs), and the tFNAs-miR22. (D) 8 % polyacrylamide gel electrophoresis confirming successful assembly, the slower-migrating bands correspond to fully formed tFNAs and tFNAs-miR22. (E) Transmission electron microscopy images displaying uniform triangular morphology for tFNAs and tFNAs-miR22. (F–G) Representative HPLC chromatograms and summary graphs of hydrodynamic diameter and zeta potential, confirming miRNA loading and increased colloidal stability. (H) Workflow for HR-MS/MS and restriction-enzyme digestion used to verify strand sequence and folded architecture, together with the metabolic-stability assay design.

2.2. Structural validation and intravitreal stability of tFNAs-miR22

To ensure the fidelity and stability of the synthesized tFNAs-miR22 nanocomplex, we performed a series of structural and metabolic analyses (Fig. 2H). High-resolution tandem mass spectrometry (HR-MS/MS) confirmed the sequence accuracy of all four single-stranded oligonucleotides (S1mR, S2, S3, S4), with coverage rates exceeding 98 % (Fig. S1). A representative isotopic distribution of the S1mR strand is shown (Fig. S2). These findings verified that all component strands were synthesized with high sequence fidelity, a critical prerequisite for precise self-assembly into the tetrahedral nanostructure [33].

To evaluate the structural integrity of the assembled tFNAs-miR22, we employed restriction endonuclease digestion using Acil, Cac8I, and ScrFI-enzymes recognizing specific double-stranded DNA motifs-and monitored cleavage fragments via HR-MS. Representative digestion patterns of the S1mR strand following Acil treatment are shown in Fig. S3, consistent with the theoretical cleavage sites. Similar results were obtained for S2–S4, indicating that the assembled tFNA possessed the expected hybridized conformation [37]. These results confirm the correct spatial folding and base pairing required for functional tFNA architecture.

Given the therapeutic application in ocular tissues, we further assessed the metabolic stability of tFNAs-miR22 in different biological environments. Ultra Performance Liquid Chromatography-High Resolution Mass Spectrometry (UPLC-HRMS) analysis demonstrated that the S1mR strand was rapidly degraded when incubated in plasma-disappearing within 15 min (Fig. S4), whereas it remained detectable after 1-h incubation in vitreous (Fig. S5). Notably, the scaffold DNA strands (S2–S4) exhibited relative stability in both vitreous and plasma environments. These results highlight the rapid metabolism of the RNA-modified strand in systemic environments and underscore the advantage of intravitreal delivery, which offers a stable ocular environment that preserves the structural integrity of the tFNAs platform and supports its sustained therapeutic function.

2.3. tFNAs-miR22 inhibit ARPE-19 cell proliferation in vitro

ARPE-19 cells were cultured under normoxic conditions and treated with vehicle, 1 μg/μL aflibercept (AFL), 100 nmol/L tFNAs, miR-22, or tFNAs-miR22 for 24 h, followed by a 6-h exposure to H₂O₂. Cell proliferation was evaluated using the EdU assay (Fig. 3A). The results demonstrated a significant inhibition of ARPE-19 cell proliferation following H₂O₂ exposure. However, cells in the tFNAs group (27.99 % ± 6.76 %) and tFNAs-miR22 group (36.88 % ± 7.98 %) exhibited significantly higher proliferative activity compared to the control group (20.62 % ± 5.45 %), with tFNAs-miR22 showing the most pronounced effect (Fig. 3B–G). These findings indicate that tFNAs-miR22 treatment effectively mitigates the proliferation suppression induced by oxidative stress.

Fig. 3.

TFNAs-miR22 mitigates oxidative stress in ARPE-19 cells and inhibits hypoxia-induced angiogenic behaviour in HUVECs. (A) Experimental workflow for the EdU incorporation assay. (B) Representative fluorescence images (EdU, red; nuclei, blue) and corresponding 3-D surface plots of ARPE-19 cells (upper rows) and HUVECs (lower rows) under the indicated treatments. (C) Schematic of the tube-formation assay. (D) Phase-contrast micrographs (4X and 10X) of HUVEC capillary-like networks after 24 h. (E) Diagram of the scratch-wound migration assay. (F) Time-lapse images showing wound closure at 0, 24, and 48 h under normoxic and hypoxic conditions. (G) Violin plots quantifying EdU-positive rate, total tube length, branch points, wound closure percentage, and migrated distance (mean ± SD, n = 6 independent experiments). Statistical significance was assessed by one-way ANOVA test; *P < 0.05, **P < 0.01, ***P < 0.001. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

2.4. tFNAs-miR22 inhibit HUVECs proliferation, tube formation and migration in vitro

To assess the anti-angiogenic potential of tFNAs-miR22, HUVECs were exposed to either vehicle, aflibercept (AFL, 1 μg/μL), miR-22, tFNAs, or tFNAs-miR22 (all at 100 nmol/L), under normoxic or hypoxic conditions (37 °C, 1 % O₂, 5 % CO₂) for 24 h. Cellular proliferation was evaluated using the 5-Ethynyl-2′-deoxyuridine (EdU) incorporation assay (Fig. 3A). Under hypoxia, a marked increase in EdU-positive cells was observed in the control group, while treatment with tFNAs or tFNAs-miR22 significantly reduced the proportion of proliferating cells, with tFNAs-miR22 yielding the most pronounced suppression (51.91 % ± 2.57 % vs. control: 79.58 % ± 2.36 %, Fig. 3B–G). These data indicate that tFNAs-miR22 effectively counteracts hypoxia-induced endothelial cell proliferation.

We next examined the impact of treatment on capillary-like tube formation. After 24 h of incubation under normoxic or hypoxic conditions, HUVECs were assessed for total tube length and number of branch points (Fig. 3C). Hypoxic stimulation promoted tube formation compared to normoxia (tube length: 33,196 ± 3345 μm/field; branch points: 1261 ± 79.28 vs. normoxia: 28,386 ± 3834 μm/field; 1139 ± 85.51, Fig. 3G). Notably, treatment with AFL, tFNAs, or tFNAs-miR22 significantly reduced both parameters, with tFNAs-miR22 showing the greatest inhibitory effect (24,573 ± 2602 μm/field; 1032 ± 59.07, Fig. 3D–G), suggesting impaired endothelial morphogenesis.

Cell migration was further evaluated using a scratch wound assay at 24- and 48-h post-treatment under hypoxic conditions (Fig. 3E). Compared to normoxia, hypoxia significantly enhanced wound closure rates (45.27 % ± 1.84 % at 24 h and 65.11 % ± 3.55 % at 48 h vs. normoxia: 62.91 % ± 5.91 % at 24 h and 100.00 % ± 0 at 48 h, Fig. 3F and G). However, treatment with tFNAs-miR22 resulted in a substantial reduction in wound healing capacity (29.37 % ± 2.51 % at 24 h and 48.97 % ± 2.47 % at 48 h, Fig. 3F and G), and outperformed AFL, tFNAs, or miR-22 alone (p < 0.05 for all comparisons), indicating superior inhibition of hypoxia-induced endothelial migration.

2.5. tFNAs-miR22 effectively reduces CNV lesion and leakage in mouse model

To preliminarily assess the therapeutic efficacy of tFNAs-miR22, we first employed a rapid-onset, self-resolving CNV model using laser-induced injury in C57BL/6 J mice (Fig. 4A), in which CNV typically develops as early as day 4 and regresses spontaneously within 14 days [26]. (Fig. 4B). Intravitreal injections of vehicle (VHCL), aflibercept (AFL, 40 mg/mL), miR22, tFNAs, or tFNAs-miR22 (55 μg/mL) were administered on D4, when CNV lesions were established. Lesion size and leakage was monitored at D0, D4, D7, and D10, using fundus photography (FP), optical coherence tomography (OCT) and fluorescein fundus angiography (FFA) (Fig. 4C).

Fig. 4.

TFNAs-miR22 reduces laser-induced CNV and vascular leakage in mice. (A) Timeline of procedures: laser photocoagulation on day 0 (D0), intravitreal injection on D4, and imaging schedule for FP, FFA and OCT. (B) Cartoon depicting the natural history of the mouse CNV model, with peak neovascularization at D4 and spontaneous regression by D10. (C) Representative FP, FFA, and OCT images at D0, D4, D7, and D10 for VHCL, AFL (40 mg/mL), miR22, tFNAs, and tFNAs-miR22 (55 μg/mL); dotted circles highlight lesions. (D–G) Violin plots summarising percentage change in OCT-derived lesion area and FFA leakage area over time (n = 6 eyes per group). One-way ANOVA test was used; *P < 0.05, **P < 0.01, ***P < 0.001.

On D7, compared to those on D4, OCT imaging showed that tFNAs-miR22 treatment led to a 34.29 ± 1.60 % reduction in lesion area, compared with VHCL (21.97 ± 0.65 %), miR22 (23.46 ± 0.91 %), tFNAs (21.61 ± 1.32 %), and matching AFL (34.05 ± 1.15 %). By D10, the cumulative lesion reduction reached 56.39 ± 2.83 % in the tFNAs-miR22 group, matching AFL (55.44 ± 2.46 %), and superior to other controls, including VHCL (27.19 ± 0.47 %), miR22 (40.05 ± 1.36 %) and tFNAs (40.80 ± 1.52 %). These results demonstrate that tFNAs-miR22 acts rapidly and matches the gold-standard therapy in reducing CNV lesion burden (Fig. 4F and G). Consistent with the OCT results, FFA data further confirmed these findings. On D7, compared to D4, leakage areas were decreased by 68.17 ± 3.27 % in the tFNAs-miR22 group, compared to 36.47 ± 3.10 % (VHCL),32.31 ± 4.52 % (AFL),33.44 ± 3.09 % (miR22), and 28.17 ± 1.50 % (tFNAs), respectively. By D10, tFNAs-miR22 group achieved an 81.52 ± 1.52 % reduction in leakage area, markedly greater than that observed in AFL (55.81 ± 3.46 %) and other controls including VHCL(56.71 ± 1.76 %), miR22 (54.45 ± 4.35 %) and tFNAs (49.74 ± 1.55 %) (Fig. 4D and E). These data suggest that tFNAs-miR22 exerts a robust anti-leakage effect following a single injection.

2.6. tFNAs-miR22 sustainably reduces CNV lesion and leakage in a dose-dependent manner in rat model

To further assess the duration of efficacy and potential dose-response relationship, we then employed a chronic, stable CNV model in Brown Norway (BN) rats, in which CNV peaks on day 7 and persists for up to 22 weeks [38] (Fig. 5A). This model is particularly suitable for evaluating long-term therapeutic effects and dosage optimization (Fig. 5B). Intravitreal injections were performed on D7 with VHCL, AFL (40 mg/mL), or tFNAs-miR22 at escalating doses: 0.1 μmol/L (low dose, LD), 1 μmol/L (medium dose, MD), and 5 μmol/L (high dose, HD) (n = 6 per group). Serial FFA was conducted on D7, D14, and D21 to track vascular leakage (Fig. 5C).

Fig. 5.

Dose-responsive and durable inhibition of CNV by tFNAs-miR22 in Brown Norway rats. (A) Experimental schedule: laser injury on D0, single intravitreal injection on D7, and serial FFA on D7, D14, and D21. (B) Illustration of the chronic rat CNV model, which peaks at one week and persists for >20 weeks. (C) Representative FFA images at each time point for VHCL, AFL, and tFNAs-miR22 at low (LD, 0.1 μmol/L), medium (MD, 1 μmol/L), and high (HD, 5 μmol/L) doses; lesions outlined in red. (D–G) Violin-plots of leakage area and fluorescence score, lesion area at D14 and D21 (n = 6 eyes per group). One-way ANOVA test; *P < 0.05, **P < 0.01, ***P < 0.001. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

On D7, FFA grading scores and leakage areas showed no significant differences among groups (P > 0.05), indicating uniform baseline lesion severity (Fig. 5C, D, E). On D14, significantly reductions of leakage area were observed in the MD (2.10 ± 0.45 mm²), HD (2.30 ± 0.58 mm²), and AFL (2.29 ± 0.51 mm²) groups versus VHCL (3.92 ± 0.36 mm²) (Fig. 4C–J). By D21, the HD group showed the lowest leakage area (1.52 ± 0.58 mm² vs. 4.36 ± 0.39 mm² in VHCL, P < 0.05). When reduction rate is calculated, leakage areas were reduced to 34.61 % (LD), 50.63 % (MD), and 48.51 % (HD) at D14, and to 45.21 %, 53.93 %, and 67.73 %, respectively at D21, which indicating a clear dose-dependent therapeutic benefit of tFNAs-miR22. Notably, the HD group outperformed AFL (39.30 %) by D21, suggesting a potential advantage over the current standard.

Fluorescence score mirrored these trends (Fig. 5F and G). On D14, fluorescence scores in the VHCL group remained high (3.53 ± 0.21), while significant reductions were observed in the MD (2.75 ± 0.20), HD (2.86 ± 0.33), and AFL (2.78 ± 0.23) groups (P < 0.05), indicating that tFNAs-miR22 achieved a therapeutic effect comparable to AFL within one week. By D21, scores declined further, with the HD group (2.17 ± 0.34) showing greater suppression than AFL (2.58 ± 0.11), suggesting enhanced and sustained efficacy at higher doses.

Together, these results illustrate a rapid and sustained anti-leakage and inhibition of CNV across both self-resolving and chronic CNV models. The treatment exhibited clear dose-dependent effects, with the high-dose group outperforming aflibercept in both lesion suppression and leakage control. These results underscore the superior efficacy of tFNAs-miR22, supporting its potential as a promising therapeutic candidate for next-generation treatment of choroidal neovascular diseases.

2.7. Transcriptomic profiling reveals PI3K/AKT/mTOR as a potential downstream pathway of tFNAs-miR22 under hypoxia

To gain mechanistic insights into the anti-angiogenic effects of tFNAs-miR22 under hypoxic conditions, bulk RNA sequencing was performed on HUVECs cultured under hypoxia with or without tFNAs-miR22 treatment (Fig. 6A). Principal component analysis (PCA) and hierarchical clustering revealed distinct transcriptomic profiles between the two groups, indicating substantial gene expression alterations induced by tFNAs-miR22 (Fig. 6B).

Fig. 6.

Transcriptomic profiling and Western-blot validation demonstrate that tFNAs-miR22 suppresses PI3K/AKT/mTOR signaling under hypoxia. (A) Workflow for RNA isolation, cDNA synthesis, next-generation sequencing, and bioinformatic analysis in hypoxic HUVECs with or without tFNAs-miR22. Workflow for protein extraction and Western blotting. (B) Bioinformatic outputs: principal-component analysis, hierarchical clustering dendrogram, Pearson-correlation heat-map (C) differential-expression heat-map (D)volcano plot (tFNAs-miR22 vs hypoxia), and (E) KEGG enrichment plot highlighting PI3K/AKT pathway suppression. (F) Protein levels of PI3K, p-PI3K, AKT, p-AKT, mTOR and p-mTOR were quantified by Western blotting, with β-actin serving as the loading control. Data are mean ± SD from three independent experiments; **P < 0.01, ****P < 0.0001 b y one-way ANOVA.

Differential expression analysis identified a robust set of genes modulated by tFNAs-miR22 treatment (adjusted p < 0.05, |log₂FC| > 0) (Fig. 6C and D). Notably, KEGG pathway analysis revealed significant enrichment in the PI3K/AKT signaling pathway, a central axis governing endothelial cell survival, angiogenic activation, and stress adaptation (Fig. 6E). The suppression of this pathway in the tFNAs-miR22 group suggested a potential molecular mechanism by which the nanocomplex exerts its anti-angiogenic and antioxidative effects under hypoxic stress.

To validate the transcriptomic findings at the protein level, we performed Western blot analysis to quantify phosphorylation states of key components within the PI3K/AKT/mTOR cascade (Fig. 6F). Compared to the hypoxia-only group, tFNAs-miR22 treatment resulted in marked reductions in the phosphorylation ratios of PI3K (p-PI3K/PI3K), AKT (p-AKT/AKT), and mTOR (p-mTOR/mTOR), indicating effective suppression of pathway activation [39]. Total protein levels of PI3K, AKT, and mTOR remained largely unchanged, suggesting that tFNAs-miR22 primarily interferes with signal transduction rather than protein expression (Fig. 6G).

Together, these results support a model in which tFNAs-miR22 attenuates hypoxia-induced angiogenic signaling by dampening the PI3K/AKT/mTOR axis, thereby contributing to its observed inhibitory effects on endothelial proliferation, migration, and neovascularization. Given the central role of this pathway in vascular remodeling and oxidative homeostasis, its modulation may represent a critical mechanism through which tFNAs-miR22 confers therapeutic benefit in CNV-related pathologies such as wAMD.

3. Conclusion

In summary, this study demonstrates that tFNAs-miR22, a structurally defined and intraocularly stable DNA nanocarrier loaded with miR-22–3p, offers a potent and durable therapeutic strategy for choroidal neovascular diseases. Through comprehensive in vivo assessments in both transient and sustained CNV models, tFNAs-miR22 effectively reduced lesion burden and vascular leakage, with dose-dependent efficacy surpassing that of conventional anti-VEGF treatment. Transcriptomic analyses further revealed that the therapeutic benefit is mediated via modulation of PI3k/Akt/mTOR pathways under hypoxic conditions. The dual advantages of biological efficacy and ocular retention position tFNAs-miR22 as a strong candidate for clinical translation in the treatment of wAMD and other vision-threatening neovascular disorders.

4. Methods

4.1. Synthesis of tFNAs-miR22

Four single-stranded DNAs (ssDNAs), listed in Table 1, were synthesized and verified by Genescript (Nanjing, China). A TM buffer (pH 8.0) was prepared using Tris-HCl and MgCl₂. Equal concentrations of the four ssDNAs, with one strand carrying microRNA22 (miR22), were combined in the buffer. The solution was then thoroughly mixed, centrifuged, heated to 95 °C for 10 min, and cooled to 4 °C for 20 min to form the tFNAs-miR22 structure. The specific sequences of the ssDNAs are provided in Table 1.

Table 1.

The sequences of the four ssDNAs and microRNA22.

ssDNA	Base sequence (5′-3′)
S1	ATTTATCACCCGCCATAGTAGACGTATCACCAGGCAGTTGAGACGAACATTCCTAAGTCTGAA
S2	ACATGCGAGGGTCCAATACCGACGATTACAGCTTGCTACACGATTCAGACTTAGGAATGTTCG
S3	ACTACTATGGCGGGTGATAAAACGTGTAGCAAGCTGTAATCGACGGGAAGAGCATGCCCATCC
S4	ACGGTATTGGACCCTCGCATGACTCAACTGCCTGGTGATACGAGGATGGGCATGCTCTTCCCG
microRNA22	AAGCUGCCAGUUGAAGAACUGU

After synthesis, 100 μL of the tFNAs-miR22 sample underwent pre-purification via HPLC using varying mobile phases. The samples were loaded onto a DNA PacTMPA100 column (ThermoFisher Scientific, USA) and purified at a flow rate of 10 mL/min. Mobile phase A consisted of 25 mM Tris-HCl, and mobile phase B contained 25 mM Tris-HCl and 375 mM NaClO₄. The purity of the tFNAs-miR22 was determined by analyzing the main peak collected after HPLC purification.

4.2. Characterization of tFNAs-miR22

To confirm the successful synthesis of tFNAs-miR22, we employed several characterization techniques, as described in previous studies. Polyacrylamide gel electrophoresis (PAGE) and HPCE were used to assess the molecular weight differences between tFNAs-miR22 and its control samples. Additionally, a nanoparticle size analyzer was utilized to measure the zeta potential and particle size distributions of the tFNAs-miR22, further confirming the integrity of the synthesized product.

4.3. Cell culture and treatment

ARPE-19 cells and HUVECs were sourced from ATCC and cultured in 1 × DMEM/F-12 medium (Gibco, USA) supplemented with 10 % fetal bovine serum (FBS) (Gibco, USA) and a complete antibiotic solution containing 10,000 μg mL⁻¹ streptomycin and 10,000 U mL⁻¹ penicillin (Gibco, USA). HUVECs were maintained under normoxic conditions (37 °C, 5 % CO₂) for 24 h. Subsequently, cells were exposed to either normoxia (37 °C, 5 % CO₂) or hypoxia (37 °C, 1 % O₂, 5 % CO₂) for an additional 24 h in the presence of different treatment conditions (vehicle control, tFNAs at 100 nmol L⁻¹, or AFL at 1 μg μL⁻¹), using DMEM/F-12 medium with appropriate FBS levels.

For oxidative stress modeling, ARPE-19 cells were treated with 200 μM hydrogen peroxide (H₂O₂) for 6 h under normoxic conditions (37 °C, 5 % CO₂). Experimental treatments were administered simultaneously with H₂O₂ stimulation to assess protective or modulatory effects.

4.4. Cell immunofluorescence assay

The expression of proteins related to angiogenesis and apoptosis was assessed using immunofluorescence staining. ARPE-19 cells were cultured in 6-well plates (Corning, USA) at a density of 9.6 × 10⁵ cells per well and treated as previously described. After treatment, the cells were fixed in PBS containing 4 % paraformaldehyde (PFA) for 15 min, followed by permeabilization and blocking with a solution of 0.5 % Triton X-100 (Biofroxx, Germany) and 1 % BSA (Biosharp, Anhui, China) for 20 min. The cells were then incubated overnight at 4 °C with the primary antibody. Afterward, the cells were incubated with a secondary antibody against rabbit IgG (diluted 1:1000; catalog number: 4413; Cell Signaling Technology, CST) for 2 h at room temperature. The nuclei were stained with Hoechst 33,342 (Solarbio, Beijing, China), and the cytoskeleton was labeled with phalloidin (Solarbio, Beijing, China). Finally, the samples were mounted with anti-fade reagent to preserve fluorescence.

4.5. Protein extraction and western blot analysis

HUVECs were cultured and divided into different experimental groups. After a 24-h incubation period, proteins were extracted from the cells using RIPA buffer (Solarbio, Beijing, China) supplemented with protease inhibitors. Protein concentration was determined using the BCA Protein Quantification Kit (KeyGEN, Nanjing, China) with a protein standard solution as a reference. Equal amounts of protein (20 μg) from the supernatant were heated at 95 °C for 5 min and then separated by electrophoresis on SurePAGE™ Bis-Tris gel (GenScript, USA). The separated proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore, Merck, Germany).

Following blocking with 5 % non-fat milk in TBS-Tween-20 (TBST) at room temperature for 1 h, the membranes were incubated overnight at 4 °C with primary antibodies: PI3K (1:1000, 4257 S, CST), p-PI3K (1:1000, 4228 S, CST), AKT (1:1000, ET1609, huabio), p-AKT (1:1000, ET1607, huabio), mTOR (1:1000, 2983 S, CST), p-mTOR (1:1000, 5536 S, CST), β-actin (1:1000, 4967 S, CST). After washing five times with TBST, the membranes were incubated with an anti-rabbit IgG-HRP conjugated secondary antibody (1:3000; 7074, CST) for 2 h at room temperature. Protein visualization was performed using an enhanced chemiluminescence detection system (ProteinSimple, USA).

4.6. EdU cell proliferation assay

To assess the inhibitory effect of tFNAs-miR22 on cell proliferation, the Alexa Fluor 488 Click-iT EdU Imaging Kit (ThermoFisher Scientific, USA) was used to detect 5-Ethynyl-2′-deoxyuridine (EdU) incorporation. ARPE-19 cells and HUVECs were seeded in 6-well plates and incubated for 24 h before being treated as described in the ‘Cell Culture and Treatment' section. The EdU proliferation assay was carried out according to the manufacturer's instructions. Fluorescent images of cells were captured using confocal laser microscopy (Carl Zeiss, Oberkochen, Germany). Statistical analysis was performed using ImageJ software, with each experiment repeated at least three times.

4.7. Cell migration assay and treatment

The ARPE-19 human retinal pigment epithelium cell line was purchased from the American Type Culture Collection (Manassas, VA, USA). Cells were cultured in a 1:1 mixture of Dulbecco's modified Eagle's medium (DMEM) and Ham's F12 medium, supplemented with 10 % fetal bovine serum and standard antibiotics (Invitrogen/Gibco, Gaithersburg, MD, USA), in a humidified incubator set to 37 °C with 5 % CO₂. The cells were authenticated using short tandem repeat DNA profiling (Supplementary Fig. S1) and were used for experiments within three passages post-thawing. To induce oxidative stress, ARPE-19 cells were exposed to H2O2 at the specified concentrations for 6 h. After treatment, cell lysates and supernatants were collected for subsequent analysis.

4.8. Laser-induced CNV model

All experimental procedures involving animals were approved by Ethics Committee for Basic and Clinical Research of Sichuan Provincial People's Hospital (The assigned approval number: LS2025429).

Mouse model. Seven-week-old male C57BL/6 J mice (SLAC, Shanghai) were anesthetized with 0.2 mL of 2 % pentobarbital sodium intraperitoneally, and pupils were dilated with 1 % tropicamide. CNV was induced by 532-nm laser photocoagulation (120 mW, 100 ms, 50 μm; Carl Zeiss Meditec, Dublin, Ireland). Four burns were applied around the optic disc, each producing a rapid bubble indicative of Bruch's-membrane rupture. Lesions were evaluated by fluorescein fundus angiography (FFA) or optical coherence tomography (OCT) at the indicated time points.

Rat model. Male Brown Norway rats (6–8 weeks) were anesthetized with intraperitoneal pentobarbital after 1 % tropicamide dilation. Six 532-nm laser burns (150 mW, 100 ms, 60 μm) were placed circumferentially around the optic disc, each producing a characteristic bubble indicating Bruch's-membrane rupture. On day 7, baseline FFA confirmed lesion formation; rats then received a single 4 μL intravitreal injection of vehicle, aflibercept (40 mg/mL), or tFNAs-miR22 (5.5, 55, or 275 μg/mL) via a 33-gauge Hamilton syringe. CNV leakage was quantified by FFA on days 14 and 21.

4.9. Intravitreal injection

Mice were first anesthetized with 1 % pentobarbital sodium (50 mg kg⁻¹). A total of 1 μL of AFL (10 μg μL⁻¹), miR22 (10 nmol L⁻¹) without transfection reagent, tFNAs (1 μmol L⁻¹), or tFNAs-miR22 (1 μmol L⁻¹) was then injected into the vitreous cavity using a 33-gauge Hamilton syringe (Hamilton, USA), under a stereomicroscope (M620 F²⁰, Leica Microsystems, France) to avoid lens damage. These injections were performed on day 0 (D0) in the CNV mice. As a control, 1 μL of vehicle solution was injected intravitreally. Following the procedure, tobramycin eye ointment was applied to prevent potential infections.

4.10. RNA sequencing and transcriptomic analysis in HUVECs

To elucidate the molecular mechanisms underlying the therapeutic effects of tFNAs-miR22 under hypoxic stress, bulk RNA sequencing (RNA-seq) was performed on human umbilical vein endothelial cells (HUVECs) cultured under two conditions: hypoxia (1 % O₂, 5 % CO₂), and hypoxia with tFNAs-miR22 treatment (Hy-tFNAs-miR22).

Total RNA was extracted using the FastPure Cell/Tissue Total RNA Isolation Kit V2 (Vazyme, China), with genomic DNA contamination removed by gDNA-Filter Columns III. RNA was purified by ethanol precipitation followed by sequential washes with Buffer RW1 and RW2 and eluted in RNase-free ddH₂O. RNA quantity and purity were measured by NanoDrop spectrophotometry, and samples were stored at −85 °C to −65 °C until further processing. Only samples with an RNA integrity number (RIN) ≥7 were used for library construction.

Polyadenylated mRNAs were enriched using oligo (dT) magnetic beads, fragmented by divalent cations, and reverse-transcribed into double-stranded cDNA using random hexamer primers. After end repair, dA-tailing, and ligation of Illumina-compatible adapters, libraries were purified, size-selected, and amplified by PCR. The final libraries were sequenced on an Illumina NovaSeq 6000 platform to generate paired-end reads (2 × 150 bp), achieving a sequencing depth of at least 20 million reads per sample.

Clean reads were aligned to the Homo sapiens reference genome (GRCh38) using HISAT2. Transcript-level expression was quantified using both TPM and raw read counts via featureCounts. Quality control was assessed using FastQC, followed by principal component analysis (PCA), hierarchical clustering, and Pearson correlation heatmaps based on log₂-counts per million (log₂CPM).

Differential expression analysis was performed using the DESeq2 package. Genes with an adjusted p value < 0.05 and absolute log₂ fold change >0 were defined as differentially expressed. Functional enrichment of differentially expressed genes was conducted using the clusterProfiler R package, with significance thresholds set at Benjamini-Hochberg-adjusted p < 0.05 for Gene Ontology (GO) biological processes and nominal p < 0.05 for KEGG pathway analysis.

To explore the transcriptional effects of tFNAs-miR22 under hypoxic stress, bulk RNA sequencing was performed on HUVECs cultured under normoxia, hypoxia, or hypoxia plus tFNAs-miR22 treatment. Total RNA was extracted using Novizan FastPure Cell/Tissue Total RNA Isolation Kit V2, and only samples with RNA integrity number (RIN) ≥7 were used. mRNA was enriched via poly(A) selection using olig dT. For library construction, purified mRNA was randomly fragmented with divalent cations, and double-stranded cDNA was synthesized using random oligonucleotides as primers. Purified double-stranded cDNA was subjected to end repair, dA-tailing, and sequencing adapter ligation. The ligation products were purified, size-selected, and amplified by PCR, followed by another round of purification to obtain the sequencing library. The library was then sequenced on an Illumina platform with paired-end 150 bp reads (≥20 million reads per sample). Quality control was performed using FastQC, and reads were aligned to the human genome (GRCh38) with STAR. Gene expression was quantified using featureCounts, and differential expression analysis was conducted with DESeq2 (adjusted p < 0.05, |log₂FC| > 0). Functional enrichment was performed using ClusterProfiler to identify pathways and biological processes regulated by tFNAs-miR22 under hypoxia.

4.11. Statistical analysis

The statistical analysis was conducted using SPSS Statistics 25.0, and the results from multiple experiments were presented as means ± standard deviation (SD). To assess significant differences between groups, a one-way ANOVA test was employed with P values below 0.05 indicating statistical significance. The number of repetitions and/or total animals involved can be found in either the figure legends or directly on the figures themselves.

CRediT authorship contribution statement

Xinyu Liu: Writing – original draft, Software, Methodology, Investigation, Formal analysis, Data curation. You Wang: Writing – original draft, Software, Methodology, Formal analysis, Data curation. Junyang Huang: Writing – original draft, Software, Methodology, Formal analysis, Data curation. Li Chen: Validation, Software, Formal analysis. Tao Cai: Visualization, Software, Formal analysis. Qiong Wang: Visualization, Software, Formal analysis. Shiqi Li: Methodology, Investigation, Data curation. Guoqiang Yu: Methodology, Investigation, Data curation. Yue Chen: Methodology, Investigation, Data curation. Delun Luo: Methodology, Investigation, Data curation. Xiaoyan Ding: Writing – review & editing, Supervision, Methodology, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability statement

The data from this study are available upon request with sufficient reason and permission from the corresponding author.