Aptamer-based nanoparticle drug delivery system designed for targeting GPC3-positive liver cancer

Abstract

Nucleic acid aptamers, short single-stranded DNA (ssDNA) or RNA molecules generated via systematic evolution of ligands by exponential enrichment (SELEX), have emerged as promising alternatives to traditional antibodies owing to their exceptional specificity and affinity. These unique properties render them invaluable tools in both scientific research and clinical theranostics. In recent years, aptamer-functionalized nanomaterials have demonstrated substantial potential as tumor-targeted drug delivery systems, offering notable advantages such as reduced systemic toxicity and enhanced therapeutic efficacy. In this study, we developed a novel targeted drug delivery system specifically designed for glypican-3 (GPC3)-positive hepatocellular carcinoma (HCC) by conjugating the GPC3-targeted aptamer AP613-1 to liposomes encapsulating doxorubicin (DOX). To enhance the biostability and binding affinity of the aptamer, we constructed an aptamer-embedded triangular nanostructure (mTriangle-AP613-1) using a DNA origami technique. Characterization results revealed that the mTriangle-AP613-1 nanostructure exhibited a significantly improved dissociation constant (K_D) of 15.2 ± 0.21 nM and enhanced biosafety profiles compared to free aptamers. Both in vitro and in vivo evaluations demonstrated that DOX-loaded mTriangle-AP613-1-liposomes achieved superior tumor growth inhibition in GPC3-positive tumor models compared to control groups. These findings highlight the potential of mTriangle-AP613-1-liposomes (DOX) as a highly effective nanomedicine platform for targeted cancer therapy. Notably, our results underscore the translational value of aptamer-functionalized nanostructures in advancing precision oncology through optimized drug delivery systems.

Graphical abstract

Preparation and perfomance of mTriangle-AP613-1-liposomes (DOX).

Highlights

• •An aptamer-based nanoparticle drug delivery system targeting GPC3-positive tumors was developed.
• •The nanostructure showed an improved K_D of 15.2 ± 0.21 nM.
• •The nanoparticle drug delivery system inhibited GPC3-positive tumor growth effectively in vivo.
• •Aptamer-functionalized nanostructures show great potential for precision oncology advancement.

1. Introduction

Hepatocellular carcinoma (HCC) represents a significant global health challenge, accounting for 80%–90% of all primary liver cancer cases [1]. Despite advances in conventional treatments such as surgery, radiation, and chemotherapy, these approaches remain limited by factors including toxicity and lack of specificity. This has driven the development of targeted therapies, which rely on identifying effective biomarkers to improve treatment efficacy. Among potential targets, glypican-3 (GPC3) has emerged as a particularly promising candidate for HCC therapy.

GPC3 is a cell-surface glycoprotein that is overexpressed in more than 70% of HCCs while being absent in benign hepatic lesions, cirrhosis, hepatitis, and healthy adult tissues [2]. Its involvement in multiple cell-signaling pathways during HCC progression, combined with its specific expression pattern, positions GPC3 as an ideal biomarker for targeted treatment. Notably, GPC3 overexpression correlates with undifferentiated, proliferating malignant hepatic cells, more aggressive HCC phenotypes, poor prognosis, and reduced overall survival. Patients with GPC3-positive HCC exhibit significantly lower 5-year survival rates compared to those with GPC3-negative tumors, and multivariate analysis has identified GPC3 as an independent prognostic factor [3]. However, traditional targeting strategies using conventional antibodies face significant challenges including immunogenicity, potential contamination, batch-to-batch variability, and high production costs.

To address these limitations, aptamers, i.e., short single-stranded DNA (ssDNA) or RNA molecules selected via systematic evolution of ligands by exponential enrichment (SELEX) [4,5], offer a promising alternative. These molecules can specifically recognize various targets with high affinity due to their unique secondary and tertiary structures. Compared to antibodies, aptamers possess distinct advantages such as ease of chemical modification, low or no immunogenicity, and rapid in vitro synthesis [6]. These properties make them particularly attractive for precise targeting applications in cancer therapy, including the potential to target GPC3-expressing HCC cells.

While aptamers provide an effective targeting mechanism, their delivery to cancer cells remains a challenge. Nanomaterials, especially liposomes, have been widely explored as drug delivery vehicles due to their ability to self-assemble into vesicles capable of carrying various therapeutic agents [7]. However, the non-selective nature of conventional liposomes leads to off-target effects, limiting their clinical application. The integration of aptamers into liposomal systems has been proposed as a strategy to enhance targeting specificity, creating a synergistic approach that combines the advantages of both technologies.

Recent advances in DNA nanotechnology, particularly DNA origami technology [8], have further expanded the possibilities for targeted drug delivery systems. Notably, the development of single-strand assembly methodologies (ssDNA origami) and self-cleaving catalytic DNA systems has effectively addressed key limitations of traditional DNA origami, enabling the efficient fabrication of complex DNA nanostructures with enhanced precision and scalability [[9], [10], [11]]. Traditional DNA origami requires precise design of long scaffold DNA strands and numerous short staple DNA oligonucleotides, which becomes increasingly problematic as structural complexity increases, leading to exponential growth in the number of staple DNAs, exorbitant synthesis costs, and tedious sequence verification. Additionally, the large quantity of staple DNAs necessitates purification via gel electrophoresis, a cumbersome process with low yield that hinders large-scale production. These limitations have motivated the exploration of alternative approaches for constructing DNA nanostructures.

Building upon these technological advancements, our research group has developed a multifunctional drug delivery system that integrates GPC3 targeting, DNA nanostructure technology, and liposomal drug delivery. In our previous work, we identified a GPC3-bound ssDNA aptamer named AP613-1, consisting of 37 nucleotides, using capillary electrophoresis (CE)-SELEX [12]. Expanding upon this foundation, we engineered a drug delivery system by functionalizing liposomes with AP613-1 and encapsulating doxorubicin (DOX) to specifically target GPC3-positive tumors. To enhance the targeting efficiency and stability, AP613-1 was further integrated into a long self-cleaving catalytic ssDNA strand that self-assembles into a triangular nanostructure, termed mTriangle-AP613-1. This innovative design enables large-scale production using biotechnology with the assistance of helper phages (Fig. 1).

Fig. 1.

Schematic illustration of (A) the synthesis of mTriangle-AP613-1-liposomes doxorubicin (DOX) and (B) the proposed antitumor mechanism. dsDNA: double-stranded DNA; ssDNA: single-stranded DNA; GPC3: glypican-3.

The mTriangle-AP613-1 nanostructure demonstrated a dissociation constant (K_D) of 15.2 ± 0.21 nM, indicating enhanced binding affinity to GPC3 compared to the original aptamer. In vitro evaluations confirmed that mTriangle-AP613-1-liposomes specifically targeted GPC3-positive cells, with cytotoxicity assays revealing that liposomes (DOX) had a lower half-maximal inhibitory concentration (IC₅₀) than free DOX. Importantly, mTriangle-AP613-1 itself showed no toxicity to tumor cells, confirming its biocompatibility. In vivo studies using Huh-7 tumor-bearing nude mice further validated the superiority of our drug delivery system. Compared to control groups, mTriangle-AP613-1-liposomes (DOX) significantly inhibited tumor growth. Tumor tissue imaging demonstrated specific accumulation of mTriangle-AP613-1-liposomes (DOX) in tumor tissues within 24 h, confirming the targeting efficiency. Moreover, safety assessments indicated no obvious adverse effects on liver function, suggesting good biosafety profiles.

Collectively, our findings demonstrate that the mTriangle-AP613-1-liposomes (DOX) represent an effective and safe nanomedicine platform for targeted therapy against GPC3-positive liver cancer. This study highlights the potential of integrating aptamer-functionalized nanostructures with liposomal drug delivery systems to advance precision oncology, offering a promising approach to improve the treatment outcomes for HCC patients.

2. Materials and methods

2.1. Cell lines and cell culture

In this study, three human cell lines were utilized: MGC-803, HepG2, and Huh-7. The MGC-803 cell line, characterized by GPC3-negative expression, is derived from human gastric adenocarcinoma. In contrast, HepG2 and Huh-7 cell lines, both exhibiting GPC3-positive expression, originate from human HCC. All cell lines were procured from the Shanghai Cell Bank of the Chinese Academy of Sciences (Shanghai, China). For cultivation, the cells were maintained in Dulbecco's modified Eagle medium (DMEM) obtained from Gibco Life Technologies (Carlsbad, CA, USA), supplemented with 10% fetal bovine serum (FBS) (Gibco, Life Technologies), at 37 °C in an atmosphere containing 5% CO₂.

2.2. Design of the recombinant p3024 phagemid vector

The p3024 phagemid vector, which contains the ssDNA, was generously provided by Prof. Hongzhou Gu Laboratory at the School of Chemistry and Chemical Engineering, Shanghai Jiao Tong University (Shanghai, China). The target ssDNA is strategically designed with restriction sites at both the 5′ and 3′ ends: G/GATCC for BamH I recognition and A/AGCTT for Hind III recognition, where the "/" denotes the restriction enzyme cutting site. Embedded within the target ssDNA are two highly active self-cleaving deoxyribozymes (I-R1a), which form hairpin secondary structures on either side when the DNA becomes single-stranded. In the presence of zinc ions, I-R1a self-cleaves at the conserved site (5′-GTTGA/AG-3′), facilitating the transformation of the target ssDNA into triangular nanostructures (mTriangle) through interaction with complementary oligonucleotides. To construct mTriangle-AP613-1 nanostructures, two AP613-1 aptamers are incorporated into the target ssDNA. Subsequently, the designed target double-stranded DNA (dsDNA) is inserted into the p3024 phagemid vector to create the recombinant p3024 phagemid vector. This construction process was carried out by GeneRay Biotech (Shanghai) Co., Ltd. (Shanghai, China).

2.3. Production of the ssDNA amplicons by phagemid amplification

Escherichia coli (E. coli) JM109 competent cells (100 μL per tube; Solarbio Life Sciences, Beijing, China) were transformed with 0.5 μL of the recombinant p3024 phagemid vector, which was prepared at concentrations of 315.8 and 452.72 ng/μL. Initially, the transformation mixture was incubated on ice for 30 min. Following this, the mixture underwent a heat shock at 42 °C for 90 s, after which it was immediately chilled on ice for additional 2 min. To facilitate recovery, 900 μL of Luria-Bertani (LB) medium (Sangon Biotech, Shanghai, China) was added to the mixture, and it was incubated in a shaker at 180 rpm and 37 °C for 1 h. The cells were then centrifuged at 2000 rpm for 5 min, leaving approximately 100 μL of cell pellets, which were subsequently resuspended. These resuspended cell pellets were evenly spread onto a pre-warmed (37 °C) LB agar plate containing ampicillin at a concentration of 100 μg/mL. The plate was then incubated upside down at 37 °C for approximately 12 h, or overnight, until visible colonies formed.

To expand the bacterial population, a single colony was initially selected and cultured in 20 mL of LB medium containing ampicillin (100 μg/mL) in a shaker set at 220 rpm and 37 °C overnight. The following day, 3 mL of this bacterial suspension was used to inoculate two separate bottles, each containing 300 mL of 2 × yeast extract-tryptone (YT) medium (composed of 16 g/L tryptone, 10 g/L yeast extract, 5 g/L NaCl, and 5 mM MgCl₂, with pH at 7.0; Sangon Biotech). These cultures were incubated under the same shaking conditions until the optical density at 600 nm (OD₆₀₀) reached approximately 0.5, corresponding to a cell density of about 4.0 × 10⁸ cells/mL. Subsequently, M13KO7 helper phage (Shuji Biotech, Shanghai, China) was introduced into each 300 mL culture at a multiplicity of infection (MOI) of approximately 10. Given the phage titer of 10¹² pfu/mL, 1.2 mL of M13KO7 helper phage was added to achieve the desired MOI (calculated as 4.0 × 10⁸ cells/mL × 300 mL × 10/10¹² pfu/mL = 1.2 mL). After incubation for 30 min, kanamycin (70 μg/mL) (Sangon Biotech) was added to each bottle to select for cells infected by the M13KO7 helper phage, and the cultures were further incubated for an additional 3–4 h. In the presence of the helper phage, the recombinant p3024 phagemid vector was amplified in a single-stranded form and packaged into phagemid particles, which were predominantly secreted into the medium. The cell cultures were then centrifuged at 4000 rpm for 15 min at 4 °C, and the supernatant was collected and transferred to a clean beaker containing polyethylene glycol 8000 (PEG 8000) (40 g/L; Sangon Biotech) and NaCl (30 g/L). This mixture was stirred for 10 min and incubated on ice for 30 min to precipitate the phagemid particles. Following centrifugation at 6500 rpm for 30 min at 4 °C, the supernatant was discarded, and the phagemid pellets were resuspended in an appropriate volume of TE buffer until fully dissolved. To remove any bacterial residue, the phagemid suspension was further centrifuged at 12,000 rpm for 10 min at 4 °C, and the resulting supernatant was transferred to clean bottles.

To remove proteins from the recombinant p3024 phagemid particles, a two-step process was employed. Initially, a solution consisting of 200 mM NaOH (Sangon Biotech) and 1% sodium dodecyl sulfate (SDS) (Sangon Biotech) was added to the phagemid solution at twice its volume. This mixture was gently swirled for 3 min to facilitate protein stripping. Subsequently, 1.5 volumes of 3 M potassium acetate (KOAc) (pH 5.5, adjusted with glacial acetic acid; both reagents from Sangon Biotech) were added into the mixture. The combined solution was then thoroughly mixed by inverting the container several times. Following this, the mixture was incubated on ice for 10 min to enhance protein precipitation. Finally, centrifugation was performed at 12,000 rpm for 30 min at 4 °C to separate the protein pellets. The recombinant p3024 phagemid ssDNA remained in the supernatant, ready for further analysis.

To isolate the ssDNA, the supernatant was first combined with twice its volume of 100% ethanol (Sangon Biotech) and then incubated on ice for 30 min. Following incubation, the mixture was centrifuged at 12,000 rpm for 30 min at 4 °C. The resulting supernatant was discarded, leaving behind pellets that contained the recombinant p3024 phagemid ssDNA. These pellets were subsequently washed with an adequate amount of 75% ethanol to remove residual salts and centrifuged again at 12,000 rpm for 10 min at 4 °C. Once the pellets were completely dry, indicating the absence of ethanol, they were dissolved in approximately 1 mL of double-distilled water (ddH₂O) (Sangon Biotech) for further analysis. The concentration of the phagemid ssDNA was then quantified using a Nanodrop One spectrophotometer (Thermo Fisher Scientific Inc, Waltham, MA, USA).

2.4. Isolation of the target ssDNA from the phagemid ssDNA by DNA-cleaving deoxyribozymes

The initial cleavage of the phagemid ssDNA by deoxyribozymes was conducted on a small scale. Specifically, 2 μL of ssDNA was combined with 48 μL of buffer 1, which consisted of 50 nM 4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid (HEPES) (Sangon Biotech) and 100 mM NaCl at pH 7.0. This mixture was incubated at 80 °C for 5 min and subsequently allowed to cool to room temperature. Following this, the cooled mixture was added to 50 μL of buffer 2, containing 50 nM HEPES, 100 mM NaCl, 50 mM MgCl₂, and 5 mM ZnCl₂ (Sangon Biotech), also at pH 7.0. The resulting solution was then incubated in a water bath at 37 °C for 1 h. To evaluate the cutting efficiency of the DNA-cleaving deoxyribozymes, a 10 μL sample was subjected to electrophoresis in a 1% denatured agarose gel. The gel was prepared with 0.25 g of agarose (Sangon Biotech), 3 g of urea (Sangon Biotech), 0.5 μL of GelRed (Sangon Biotech), and 25 mL of 1 × Tris-acetate-ethylenediaminetetraacetic acid (TAE) buffer (Sangon Biotech). Electrophoresis was conducted at room temperature for 45 min at 80 V, using 1 × TAE buffer and supplemented with 10 μL of loading buffer (8M urea, 10% (m/V) sucrose, 0.025% (m/V) bromophenol blue, 0.025% (m/V) xylene cyanol FF, and 0.05% (m/V) SDS) (Sangon Biotech). The gel was subsequently analyzed for accuracy and efficiency under ultraviolet (UV) light using a gel imaging system (Tanon 1600; Tanon, Shanghai, China).

To effectively isolate the target ssDNA from the phagemid ssDNA on a large scale, approximately 120 pmol of the sample was initially incubated with buffer 1 at 80 °C for 15 min in each bottle. Following this high-temperature incubation, the mixture was allowed to cool to room temperature. Once cooled, an equivalent volume of buffer 2, matching the volume of the mixture, was added to each bottle. The samples were then incubated overnight in a water bath maintained at 37 °C.

The following day, the procedure for collecting phagemid ssDNA was employed, as previously described. The ssDNA, digested by DNA-cleaving deoxyribozymes, was combined with a two-fold volume of 100% ethanol and incubated on ice for 30 min. Subsequently, the mixture underwent centrifugation at 12,000 rpm for 30 min at 4 °C. After discarding the supernatant, the pellets were thoroughly washed with 75% ethanol and subjected to an additional centrifugation at 12,000 rpm for 10 min at 4 °C to eliminate residual salts. Once completely dry, each pellet was directly dissolved in 0.6 mL of LB and mixed using a pipette. The ssDNA solution was then subjected to electrophoresis in a 1% denatured agarose gel, composed of 0.5 g of agarose, 6 g of urea, 1 μL of GelRed, and 50 mL of 1 × TAE buffer, for 45 min at room temperature (80 V) using 1 × TAE buffer. To prevent damage to the target ssDNA, the gel was visualized under light-emitting diode (LED) blue light (OSE-470; TIANGEN Biotech (Beijing) Co., Ltd, Beijing, China) rather than UV exposure. Two distinct ssDNA bands were observed on the gel, with the lower band identified as the target ssDNA. This target ssDNA was subsequently collected into a new container and stored at 4 °C for further analysis.

To recycle the target ssDNA, the extraction process begins by mixing the ssDNA with binding buffer NTI, using a ratio of 200 μL of buffer NTI for every 100 mg of agarose gel. This mixture is incubated in a water bath at 50 °C for 10–15 min until the gel slices are completely dissolved. Once dissolved, 700 μL of the sample is loaded into a clean-up column placed within a 2 mL collection tube. The tubes are then centrifuged at 10,000 rpm for 1 min, and the flow-through is discarded. If necessary, the remaining sample is loaded and centrifuged repeatedly. Subsequently, 700 μL of wash buffer NT3 is added to each clean-up column, which is then returned to the collection tube. The tubes undergo another centrifugation at 10,000 rpm for 1 min, with the flow-through discarded. This washing step is recommended to be repeated twice to effectively remove chaotropic salts. Following this, the tubes are centrifuged at 10,000 rpm for an additional 2 min to ensure complete removal of residual ethanol from the wash buffer NT3. The columns are then transferred to new 1.5 mL microcentrifuge tubes and left exposed to air at room temperature for at least 30 min. Finally, 30–50 μL of preheated ddH₂O at 65 °C is added to each column and incubated at room temperature for 2 min. The tubes are centrifuged at 10,000 rpm for 2 min to collect the flow-through, which contains the isolated target ssDNA from the phagemid ssDNA. The target ssDNA is subsequently quantified using a Nanodrop One (Thermo Fisher Scientific Inc.) and stored at 4 °C for further analysis. The purity of the target ssDNA can be verified through electrophoresis.

The binding buffer NTI and wash buffer NT3 were both included in NucleoSpin kit from MACHEREY-NAGEL (Düren, Germany).

2.5. Assembly of the target ssDNA nanostructures with complementary oligonucleotides

Based on the principle of base complementarity, ssDNA can self-assemble into triangular nanostructures when incubated with complementary oligonucleotides. In this process, the molar ratio of target ssDNA to complementary oligonucleotides is maintained at 1:6. Both components are initially dissolved in an 8 × hydration buffer (HB), which consists of 200 mM HEPES, 3.2 M KCl (Sangon Biotech), and 80 mM MgCl₂ at pH 7.0. Subsequently, this solution is diluted with deionized water (ddH₂O) to achieve a final buffer concentration of 1 × HB.

To verify the successful self-assembly of the nanostructures, the target ssDNA was first quantified to a concentration of approximately 10 nM using an 8 × HB solution. This ssDNA was then programmed with complementary oligonucleotides using a polymerase chain reaction (PCR) thermal cycler (BIO-RAD, Hercules, CA, USA). The mixture underwent a series of temperature changes: initially heated to 95 °C for 5 min, then gradually cooled to 75 °C for 10 min, 60 °C for 10 min, 45 °C for 10 min, and finally to 22 °C for 10 min. Following this thermal cycling, the resulting triangular nanostructures were subjected to electrophoresis in a 6% native gel. The gel composition included 1.5 mL of 40% Arc-Bis (Sangon Biotech), 1 mL of 10 × TAE buffer, 100 μL of 1 M CH₃(COO)₂Mg (Sangon Biotech), 100 μL of 10% ammonium persulfate (APS) (Sangon Biotech), 7 μL of N,N,N′,N′-tetramethylethylenediamine (TEMED) (Sangon Biotech), and deionized water to a total volume of 10 mL, with 1 × TAE buffer (10 mM CH₃(COO)₂Mg) used as the running buffer. The electrophoresis was conducted at 80 V for 3.5–4 h. As a control, the target ssDNA was also run on the same 6% native gel. Both samples were prepared with a loading buffer consisting of 50% glycerol (Sangon Biotech) mixed in a 1:1 volume ratio with 1 × TAE buffer. Additionally, mTriangle-AP613-1 at a concentration of 10 nM was deposited onto a mica sheet and analyzed using atomic force microscopy (AFM) with a Dimension ICON instrument (Bruker, Karlsruhe, Germany).

Fluorescein and cholesterol could be modified on complementary oligonucleotides at the 5′ and 3′ ends respectively. This procedure was manipulated by GeneRay Biotech (Shanghai) Co., Ltd..

2.6. Production of liposomes

The lipids 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) (31.8 mM), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) (13.4 mM), 1,2-dioleoyl-sn-glycero-3-phospho-L-serine (DOPS) (12.3 mM), and 1,2-dioleyl-sn-glyceryl-3-phosphate ethanolamine-N-(lissamine rhodamine B sulfonyl) (Liss Rhod PE) (0.768 mM) were procured from Avanti® Polar Lipids (Alabaster, AL, USA) for the preparation of liposomes. To achieve a total lipid concentration of 5 mM in a 600 μL liposome solution (equivalent to 3 μmol of lipids), specific volumes of each lipid were combined: 55.9 μL of DOPC, 67.2 μL of DOPE, 24.4 μL of DOPS, and 31.2 μL of Liss Rhod PE. These lipids were dissolved in chloroform and mixed in a glass tube. The lipid mixture was then evaporated under a stream of nitrogen gas for 5 min to form a thin lipid film at the bottom of the tube. Subsequently, 600 μL of pre-heated (NH4)₂SO₄ (Sangon Biotech) solution (maintained at 65 °C, pH 5.5) was added to resuspend the lipid film. The resulting suspension was stirred using a magnetic stirrer for at least 30 min, ensuring the temperature remained at 65 °C throughout the process. To further reduce large lipid aggregates, the suspension was subjected to freeze-thaw cycles using liquid nitrogen, repeated 5–10 times. The liposomes were then sequentially extruded through membranes with pore sizes of 200, 100, and 50 nm, with each extrusion performed 10–15 times. The extruded liposomes were subsequently transferred to a new tube and stored at 4 °C. Finally, the zeta potential of the liposomes was assessed using a Malvern Zetasizer Nano ZS90 (Malvern, Worcestershire, UK).

2.7. ssDNA nanostructure-assisted liposomes sorting

To prepare the ssDNA nanostructures coated liposomes, 280 μL of cholesterol/cyanine 5 (Cy5) modified ssDNA nanostructures (1 μM) were incubated with 21 μL of liposomes, which consisted of 5% 200 nm, 45% 100 nm, 45% 50 nm, and 5% sonicated liposomes, at 37 °C for 2 h. The ratio was maintained at 40 pmol DNA to 3 μL liposomes. Concurrently, an iodixanol density gradient was prepared by diluting iodixanol (Sangon Biotech) in 1 × HB to achieve concentrations of 20%, 18%, 15%, 12%, 9%, 6%, 3%, and 0% (v/v). The ssDNA nanostructures coated liposomes were then mixed with 180 μL of 60% iodixanol to create a 22.5% iodixanol solution, which was subsequently placed at the bottom of a 4 mL ultracentrifugation tube. Following this, 460 μL of each iodixanol dilution was carefully layered into the ultracentrifugation tube, starting with the 18% solution at the bottom and ending with the 0% solution at the top. The density gradient mixture was then subjected to ultracentrifugation at 55,000 rpm for 5 h at 23 °C.

To assess the sorting efficacy of the liposomes, the collected samples were subjected to gel electrophoresis. Specifically, approximately 2 μL of the sample from each well was loaded onto a 3.5% agarose gel containing 0.05% SDS. The electrophoresis was conducted at 80 V for 2 h. The sorted liposomes were initially collected by pipetting 200 μL of solution from top to bottom into a 96-well plate. Following electrophoresis, the gel was imaged using a Typhoon 9500 scanner (General Electric Company, Marlborough, MA, USA) to evaluate the sorting results.

To remove iodixanol, the samples were centrifuged at 10,000 rpm for 10 min using a 30 kDa molecular weight cut-off (MWCO) (Amicon; Merck Millipore, Burlington, MA, USA) filter and washed three times with 1 × HB. Additionally, the concentrated samples could be treated with DNase I (Beyotime Biotechnology, Shanghai, China) to eliminate the ssDNA coating.

2.8. Transmission electron microscope (TEM) imaging

To prepare the liposome samples for analysis, approximately 5 μL of liposomes was deposited onto a glow-discharged, carbon-coated copper grid (Electron Microscopy Sciences, Hatfield, PA, USA) and allowed to sit at room temperature for 1 min. Following this, the sample was blotted with filter paper and briefly washed with 5 μL of uranium acetate (Sangon Biotech). The excess liquid was then blotted again, and the sample underwent a second wash with 5 μL of uranium acetate for an additional 1 min. Subsequently, the grid was left to air-dry at room temperature for 5 min. The size of the liposomes was then measured using a TEM (FEI Talos L120C microscope; Thermo Fisher Scientific Inc.) operating at 120 kV.

2.9. Encapsulation of DOX

The dissociation of (NH₄)₂SO₄ within the liposomes results in the formation of NH₄⁺ and SO₄²⁻. Subsequently, NH₄⁺ can further dissociate into NH₃ and H⁺. When the external (NH₄)₂SO₄ is removed using a 9.4% sucrose solution at pH 6.3, NH₃ diffuses out of the liposomes due to its higher transmembrane rate and permeability coefficient compared to SO₄²⁻. This diffusion creates an accumulation of H⁺ within the liposomes, thereby establishing a pH-gradient system. To load DOX into the liposomes, a mixture of DOX at a 1:10 mass ratio to liposomes was incubated at 65 °C for 15 min. Following incubation, the mixture was centrifuged at 12,000 rpm for 3 min using ultrafiltration devices to remove excess DOX. The liposomes were then washed with 9.4% sucrose (pH 6.3) (Sangon Biotech) and centrifuged 3–5 times until the filtrate was completely transparent. The collected flow-through was used to measure the OD value at a wavelength of 480 nm, and the encapsulation efficiency was calculated based on established standards.

2.10. Half maximal inhibitory concentration (IC₅₀) and cytotoxicity analysis

To determine the IC₅₀ values, all three types of tumor cells were initially cultured in 96-well plates for 24 h (7 × 10³ cells per well). Subsequently, the cells were treated with varying concentrations of DOX, following a two-fold dilution series: 100, 50, 25, 12.5, 6.25, 3.125, 1.56, 0.78, and 0 μg/mL. On the following day, the cells were incubated with 100 μL of medium containing 10 μL of Cell Counting Kit-8 (CCK-8) reagent (Sangon Biotech) for a duration of 2–4 h at 37 °C. After incubation, the 96-well plates were analyzed using a microplate reader (Thermo Fisher Scientific Inc.) to assess cell viability, with measurements taken at a wavelength of 450 nm.

In the cytotoxicity experiment, three types of tumor cells were cultured in 96-well plates and treated with either liposomes at a concentration of 0.15 mM or ssDNA nanostructures at 100 pmol. Following inoculation for varying durations of 24, 48, and 72 h, 100 μL of medium containing 10 μL of CCK-8 reagents was added to each well. Subsequently, the cells were co-incubated for 2–4 h at 37 °C. Finally, the 96-well plates were analyzed using a microplate reader to assess cell viability, with measurements taken at a wavelength of 450 nm. The cell viability is calculated by using the formula shown below:

Cell viability (%) = [(A_s − A_b)/(A_c − A_b)] × 100%, where A_s, A_b, and A_c indicates the absorbance of experimental group, blank group, and control group, respectively.

2.11. Surface plasmon resonance (SPR) experiment

2.11.1. AP613-1

In this study, the aptamer AP613-1 was modified with biotin at its 5′ end, as provided by GeneRay Biotech (Shanghai) Co., Ltd.. Preliminary experiments determined that approximately 1 μM of the biotin-modified AP613-1 was optimal for loading onto streptavidin (SA) biosensors (ForteBio, Fremont, CA, USA) for 10 min. The analytes used were His-tag GPC3 proteins (ACROBiosystems, Newark, DE, USA) prepared in a 96-well plate with concentrations following a two-fold dilution series: 1,600, 800, 400, 200, 100, 50, 25, and 0 nM. Both the aptamers and proteins were dissolved in a running buffer composed of 25 mM HEPES, 0.4 M KCl, 10 mM MgCl₂, and 0.005% Tween-20 (Sangon Biotech), adjusted to pH 7.4. To remove any unbound AP613-1, the AP613-1-conjugated biosensors were washed in the running buffer for 2 min. Subsequently, these biosensors were immersed in the GPC3 solution for 10 min to facilitate the association with GPC3. Following this, a 5-min wash in the running buffer was conducted to promote the dissociation of GPC3 from the biosensors. The K_D was then automatically calculated using the Octet 96e instrument (ForteBio).

2.11.2. mTriangle/mTriangle-AP613-1

In the preliminary experiments, approximately 400 nM of His-tagged GPC3 proteins were immobilized on nitrilotriacetic acid (NTA) biosensors (ForteBio) for a duration of 10 min. The analytes, mTriangle and mTriangle-AP613-1, were prepared at varying concentrations in a 96-well plate, employing a two-fold dilution series: 62.5, 31.25, 15.63, 7.81, 3.91, 1.95, 0.98, and 0 nM. Both the ssDNA nanostructures and the proteins were dissolved in the running buffer. To remove any unbound GPC3, the GPC3-conjugated biosensors were washed in the running buffer for 2 min. Subsequently, the biosensors were exposed to the GPC3 solution for 10 min to facilitate the association with mTriangle/mTriangle-AP613-1. Following this, the biosensors underwent an additional 5-min wash in the running buffer to promote dissociation from mTriangle/mTriangle-AP613-1. The K_D was automatically calculated using the Octet 96e instrument.

2.12. Tumor cells imaging

Three types of tumor cells were initially seeded into glass-bottom cell culture dishes (Φ 15 mm; NEST Biotech, Wuxi, China). Following a 24-h incubation period, the cells underwent a triple wash with phosphate-buffered saline (PBS) to ensure cleanliness. Subsequently, they were stained with 100 pmol of Cy5.5-labeled mTriangle-AP613-1-liposomes, maintaining a ratio of 40 pmol DNA to 3 μL liposomes, and incubated at 37 °C in the dark for 1 h. For comparison, mTriangle-liposomes served as a negative control. After incubation, the cells were washed three times with PBS to remove any unbound nanostructures. They were then stained with Hoechst 33342 for 5 min in the dark to facilitate nuclear visualization. Following thorough washing, the cells were imaged using a laser-scanning confocal microscope (Leica SP8; Leica Microsystems, Richmond Hill, ON, Canada). Additionally, as a positive control, the tumor cells were stained with 5 μg/mL mouse antihuman GPC3 antibody (Abcam, Cambridge, MA, USA).

2.13. Flow cytometry analysis

In this experiment, three types of tumor cells were seeded in a six-well plate. Approximately 1 × 10⁶ tumor cells were incubated at 37 °C in the dark for 1 h with 100 pmol of fluorescein isothiocyanate (FITC) labeled mTriangle-AP613-1-liposomes, which consisted of 40 pmol DNA to 3 μL liposomes, in a 200 μL PBS solution. Following incubation, the tumor cells underwent two washes with PBS to remove any unbound nanostructures and were subsequently resuspended in 0.5 mL PBS. The fluorescence signal intensity of the cells was then analyzed using a flow cytometer (FACScan; Becton, Dickinson and Company, Mansfield, MA, USA).

2.14. Western blot

Three types of tumor cells were lysed using radioimmune precipitation assay (RIPA) buffer (Beyotime Biotechnology) containing protease and phosphatase inhibitors. The concentration of extracted proteins was quantified using bicinchoninic acid (BCA) Protein Assay Kit (Beyotime Biotechnology). Equivalent amounts of protein extracts were loaded onto the 10% SDS-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to polyvinylidene difluoride (PVDF) membranes (Millipore). After blocking with 5% non-fatty milk (Beyotime Biotechnology) in Tris-buffered saline with Tween 20 (TBST) (Beyotime Biotechnology) for 2 h, the membranes were incubated with primary antibodies specific against GPC3 (Abcam) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Proteintech, Rosemont, IL, USA) overnight at 4 °C, the blots were incubated with appropriate horseradish peroxidase (HRP)-conjugated secondary antibody (Affinity, Golden, CO, USA) for 1 h at room temperature. Bands were visualized using the enhanced chemiluminescence (ECL) regent (Tanon) and finally analyzed using the ImageJ software.

2.15. Nuclease degradation assay

Based on their molecular weight and conformation, AP613-1 and mTriangle-AP613-1 were tested for their stability under nuclease digestion. Equal amount of the assembled ssDNA nanostructure and aptamer (about 1 μg) were individually incubated with DNase I (1 U/μL) in 30 μL reaction solution (10 mM Tris-HCl, 2.5 mM MgCl₂, and 0.1 mM CaCl₂) at 37 °C for certain time. At different time points (t = 0, 5, 10, 15, 20, 25, and 30 min), 9 μL of the sample was taken out and mixed with 1 μL of a stop buffer (25 mM EDTA) at 65 °C for 10 min to terminate the digestion. After all samples were collected, they were run into 1% agarose gels containing a 1:10,000 dilution of GelRed in 1 × TAE buffer at 70 V for 1 h. After electrophoresis, the gels were imaged and analyzed with a gel imaging system (Tanon 1600).

2.16. Safety assessment and antitumor efficacy in vivo

For safety assessment, prior to the in vivo application of mTriangle-AP613-1-liposomes, a comprehensive safety assessment was conducted due to the propensity of liposomes to accumulate in the liver. To evaluate potential hepatic effects, Bagg Albino (BALB/c) nude mice were administered mTriangle-AP613-1-liposomes via the tail vein. Subsequently, whole blood samples were collected from the mice and placed in 1.5 mL tubes. These samples were then centrifuged at 5000 rpm for 15 min at 4 °C to obtain serum. The serum was analyzed for liver function indicators, which were assessed by Servicebio (Wuhan, China).

In vivo antitumor efficacy was assessed using BALB/c nude mice, into which approximately 1 × 10⁷ Huh-7 cells were subcutaneously injected into the right upper flank. Tumor volume was calculated using formula 1/2 × length × width², and the weight of the mice was recorded every other day to monitor their health. Once the tumors reached a volume of approximately 100 mm³, which typically occurred after about one week, the mice were randomly divided into three groups. Each group received treatment via tail vein injection every two days: one group received saline, another received mTriangle-liposomes (DOX), and the third group received mTriangle-AP613-1-liposomes (DOX). After a treatment period of 10 days, the mice were euthanized, and the tumors were collected for further analysis.

The procedures for care and use of animals were approved by the Ethics Committee of Zhongshan Hospital of Fudan University (Shanghai, China) and all applicable institutional and governmental regulations concerning the ethical use of animals were followed (Approval No.: Y2024-613).

2.17. Statistical analysis

The data were analyzed using ImageJ, Origin 2021, and GraphPad Prism 9 software. Quantitative variables were presented as means ± standard deviation (SD) and assessed with one-way analysis of variance (ANOVA). Results were deemed statistically significant at P < 0.05.

3. Results

3.1. Catalytic DNA-assisted biotechnological production of mTriangle-AP613-1

Our research commenced with the design of mTriangle-AP613-1. Initially, the target ssDNA was engineered to be flanked by two restriction sites at the 5′ and 3′ ends (BamH I and Hind III). This ssDNA was synthesized as a double-stranded pseudogene fragment and subsequently inserted into a p3024 phagemid vector [11] (Fig. 2A and Table S1). To enable the spontaneous detachment of the target ssDNA from the phagemid ssDNA, two conserved recognition sequences (5′-GTTGAAG-3′) for the class I Zn²⁺-dependent DNA-cleaving deoxyribozymes (I-R1a) [13] were also embedded within both ends of the target ssDNA, which would fold into a hairpin secondary structure for in cis [13] cleavage (Fig. 2B). Upon cleavage from the phagemid ssDNA, the target ssDNA was capable of self-folding into a triangular nanostructure with the complementary oligonucleotides, which could be modified with various chemical moieties (Fig. 2C), termed mTriangle (Fig. S1). Based on this framework, we integrated two AP613-1 aptamer sequences into the target ssDNA (Fig. 2B). The recombinant target ssDNA, also synthesized as a double-stranded pseudogene fragment (760 bp), was then inserted into a p3024 phagemid vector to create a recombinant p3024. Similar to the original design, when cleaved by self-cleaving deoxyribozymes, the recombinant target ssDNA could fold into a triangular nanostructure with the same complementary oligonucleotides, resulting in the formation of mTriangle-AP613-1 (Figs. 2C and S2).

Fig. 2.

Characteristics of mTriangle-AP613-1. (A) Detailed information about the recombinant p3024 (3869 bp) phagemid vector: 1, 10,000 bp DNA ladder; 2, restriction enzyme digestion, inserted double-stranded DNA (dsDNA) (shown in green box). (B) Detailed information about the recombinant target single-stranded DNA (ssDNA). (C) The recombinant target ssDNA combines with complementary oligonucleotides, which are modified with fluorescein and cholesterol, to form mTriangle-AP613-1 through polymerase chain reaction (PCR) assembly.

The phagemid-helper-phage system [9,11] has emerged as a promising biotechnological method for the efficient and cost-effective production of large quantities of long-strand ssDNA. In our study, we utilized the recombinant p3024 phagemid vector, which was amplified in E. coli JM109 competent cells. These cells were co-incubated with the M13KO7 helper phage to generate phagemid ssDNA amplicons through an in vivo rolling circle amplification (RCA) mechanism (Fig. 3A). To evaluate the cutting efficiency and accuracy of DNA-cleaving deoxyribozymes, about 2 μL of the phagemid ssDNA was initially incubated in a cleaving buffer (50 nM HEPES, 100 mM NaCl, 50 mM MgCl₂, and 5 mM ZnCl₂; pH 7.0) at 37 °C for 1 h. The digested products were analyzed using a denaturing agarose gel, which allowed us to confirm the size of the recombinant target ssDNA by comparison with two types of DNA ladder (1 kb and 100 bp). An electrophoretic band, identified as the recombinant target ssDNA, could be observed cleaving from the phagemid ssDNA (Fig. 3B). Following these preliminary results, a large quantity of the phagemid ssDNA (approximately 1 mg) was cleaved using the same method, and the digested products were collected and purified (Fig. 3C). The purified ssDNA underwent further analysis via denaturing agarose gel, confirming its identity as the recombinant target ssDNA (Fig. 3D). Subsequently, the recombinant target ssDNA was assembled with the complementary oligonucleotides to form mTriangle-AP613-1 through assembly PCR (95 °C for 5 min, 75 °C for 10 min, 60 °C for 10 min, 45 °C for 10 min, and 22 °C for 10 min). The PCR products were analyzed using a 6% native gel, which revealed a significant band shift for mTriangle-AP613-1 compared to the recombinant target ssDNA, indicating the successful formation of the nanostructure (Fig. 3E). These nanostructures were also visualized under AFM (Fig. S3). Additionally, mTriangle was assembled using the same method (Figs. S4A–C). Electrophoresis demonstrated that the original target ssDNA moved faster than the recombinant target ssDNA due to its smaller molecular weight (Fig. S4D). Based on a preliminary estimation, the recovery rate of the target ssDNA throughout the entire process, from enzymatic digestion to purification, was approximately 43.4% (Fig. S4E). Collectively, these results demonstrated the successful application of DNA-cleaving deoxyribozymes in the biotechnological production of mTriangle-AP613-1.

Fig. 3.

Biotechnological production of mTriangle-AP613-1. (A) The catalytic DNA-assisted biotechnological production of mTriangle-AP613-1. (B) Deoxyribozyme-assisted digestion of phagemid single-stranded DNA (ssDNA) (small scale, red box points to the recombinant target ssDNA): 1, 1 kb DNA ladder; 2, 100 bp DNA ladder; 3, ssDNA before deoxyribozyme digestion; and 4, ssDNA after deoxyribozyme digestion. (C) Deoxyribozyme-assisted digestion of the ssDNA (large scale, white box points to the recombinant target ssDNA). (D) Certification of the purified recombinant target ssDNA (red box points to the recombinant target ssDNA): 1, 1 kb DNA ladder; 2, 100 bp DNA ladder; 3, ssDNA after deoxyribozyme digestion; and 4, the purified recombinant target ssDNA. (E) Assembling of the recombinant target ssDNA with complementary oligonucleotides: 1, 100 bp DNA ladder; 2, the purified recombinant target ssDNA; 3, mTriangle-AP613-1, with a molar ratio of purified recombinant target ssDNA to complementary oligonucleotides of 1:6, with cholesterol labeled on the complementary oligonucleotides; and 4, mTriangle-AP613-1, with the same molar ratio but without cholesterol labeling. dsDNA: double-stranded DNA; E. coli: Escherichia coli; YT: yeast extract-tryptone; RCA: rolling circle amplification; PEG 8000: polyethylene glycol 8000; OD₂₆₀: optical density at 260 nm; PCR: polymerase chain reaction; LED: light-emitting diode.

3.2. Production of liposomes and ssDNA nanostructure-assisted liposomes sorting

In our study, we mixed various commercial lipids to produce liposomes (Fig. 4A and Table S2). The composition of liposomes included 59.2% DOPC, 30% DOPE, 10% DOPS, and 0.8% Liss Rhod PE (Figs. 4B–E), all of which were stocked in chloroform respectively. The lipid mixture was subjected to repeated liquid nitrogen freeze-thaw treatments to prevent accumulation. They were then sequentially extruded through the membranes with pore sizes of 200, 100, and 50 nm to achieve a final liposome size between 50 and 200 nm (Fig. 4A). Additionally, we utilized a sonicator to generate liposomes with a mean diameter of 30 nm. The zeta potential of these liposomes was measured at approximately −3.58 mV and the average particle size of the liposomes detected by dynamic light scattering (DLS) was 144.7 nm (Figs. S4F and G).

Fig. 4.

Production of liposomes. (A) Construction of mTriangle-AP613-1-liposomes. (B−E) Chemical structures of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), 1,2-dioleoyl-sn-glycero-3-phospho-L-serine (DOPS), and 1,2-dioleyl-sn-glyceryl-3-phosphate ethanolamine-N-(lissamine rhodamine B sulfonyl) (Liss Rhod PE).

According to the research conducted by Yang et al. [14] and Jia et al. [15], ssDNA nanostructures have the potential to facilitate the sorting of liposomes by size through isopycnic centrifugation. We initiated the mTriangle-AP613-1-assisted sorting of liposomes by labeling the complementary oligonucleotides with a Cy5 dye at the 5′ end for tracking purpose and a cholesterol moiety at the 3′ end for attachment to the lipid bilayer (Figs. S5A and B). Liposomes were incubated with mTriangle-AP613-1 at 37 °C for 2 h to achieve coating. The mTriangle-AP613-1-liposomes were then mixed with iodixanol to create a 22.5% iodixanol solution, which was placed at the bottom of a 4-mL iodixanol density gradient centrifuge tube (3%, 6%, 9%, 12%, 15%, 18%, and 20%) (Fig. S5C). Following isopycnic ultracentrifugation, the mTriangle-AP613-1-liposomes were theoretically separated by size, with larger liposomes (low density) positioned at the top and smaller ones (high density) at the bottom. The mixture was subsequently fractionated into 200 μL portions for storage.

The analysis of the mixture using 3.5% SDS-agarose gel electrophoresis revealed the coexistence of mTriangle-AP613-1 and liposomes in most fractions, confirming the successful coating of mTriangle-AP613-1. Additionally, the presence of bright Cy5 signals and barely detectable rhodamine signals indicated the saturation coating of mTriangle-AP613-1 (Fig. S5C). Further examination using negative-stain TEM provided images that confirmed size homogeneity of liposomes within each fraction, with mean diameters estimated at 111.9 nm in the 8th fraction and 47.2 nm in the 16th fraction (Fig. S5D). Overall, the findings demonstrated that mTriangle-AP613-1 is effective in sorting liposomes with distinct sizes.

3.3. Affinity analysis of mTriangle-AP613-1

Aptamers are known for their ability to spontaneously fold into diverse tertiary structures, forming critical domains that interact with ligands with high affinity. In this study, we employed a computational method to predict the three-dimensional (3D) spatial structure of AP613-1 and its potential binding sites with GPC3 proteins. Models 1–5 represent the top five 3D structures of AP613-1 predicted by Zhang et al. [16]. In this context, a lower score indicates a higher possibility (Fig. 5A). Furthermore, ligand positions were modeled using the ZDOCK sever [17]. dA14, dG15, dT17, dG23, dT26, dT27, dC29, and dA30 of AP613-1 are probably in close contact with GPC3 (Fig. 5B). To evaluate the affinity of AP613-1, we utilized SPR technology to monitor the binding dynamics of AP613-1 to GPC3. Biotin-modified AP613-1 (1 μM) was immobilized onto eight SA biosensors for duration of 10 min. Subsequently, these biosensors were incubated with a series of two-fold dilutions of human recombinant GPC3 proteins, with concentrations ranging from 1600 to 0 nM (control) (1600, 800, 400, 200, 100, 50, 25, and 0 nM), for additional 10 min. This experimental setup yielded a K_D value of approximately 271 ± 5.76 nM (Fig. 5C). In parallel, the affinity of mTriangle-AP613-1 was also assessed using SPR, revealing an improved K_D value of 15.2 ± 0.21 nM (Fig. 5D). This time, polyhistidine (poly-His) modified GPC3 (400 nM) was immobilized onto eight NTA biosensors for the same duration of 10 min, and then incubated with a series of two-fold dilutions of mTriangle-AP613-1 at concentrations of 62.5, 31.25, 15.63, 7.81, 3.91, 1.95, 0.98 and 0 nM (control). Notably, the mTriangle-AP613-1 nanostructure exhibited remarkable stability in solutions with various pH values (Fig. S6). The increased affinity of mTriangle-AP613-1 for GPC3 likely results from the enrichment of AP613-1 in ssDNA nanostructures. Additionally, we tested the affinity of mTriangle for GPC3 following the same procedure. However, the results indicated nonsignificant interaction between these two substances (Fig. S7). In summary, our findings demonstrated that mTriangle-AP613-1 exhibits an enhanced K_D value compared to AP613-1.

Fig. 5.

In vitro analysis of mTriangle-AP613-1. (A) Three-dimensional (3D) structure of AP613-1. The numbers 1−5 indicate the five most probable secondary structures of AP613-1 predicted by the software, respectively. A smaller number indicates a more stable structure. (B) ZDOCK is used to predict glypican-3 (GPC3)-AP613-1 docking (dA14, dG15, dT17, dG23, dT26, dT27, dC29, and dA30 are in close contact with GPC3. C, T, G, and A represent the bases and the number represents the location of the base in the single-stranded DNA (ssDNA) sequence). (C, D) Surface plasmon resonance (SPR) results of AP613-1 (C) and mTriangle-AP613-1 (D). All the buffer used is 1 × 4-(2-Hydroxyethyl) piperazine-1-ethanesulfonic acid (HEPES) (10 mM HEPES, 150 mM NaCl, 3 mM ethylenediaminetetraacetic acid (EDTA), and 0.005% Tween-20; pH 7.4). SA: streptavidin; Poly-His: polyhistidine; K_D: dissociation constant; NTA: nitrilotriacetic acid.

Notably, nucleic acids are typically immobilized on biosensors. However, only when employed as ligands do mTriangle and mTriangle-AP613-1 display favorable kinetic characteristics, most likely as a result of their complex spatial nanostructures or greater molecular weight.

3.4. In vitro targeting analysis of the mTriangle-AP613-1-liposomes

To explore the potential application of mTriangle-AP613-1-liposomes in targeting GPC3-positive cells, we conducted in vitro flow cytometry and cellular immunofluorescence techniques. Initially, we validated the expression of GPC3 in three cells lines using an anti-GPC3 antibody. HepG2 and Huh-7 cells, both HCC cell lines, are known for their high expression of GPC3, whereas MGC-803 cells, gastric adenocarcinoma cell line, exhibit low expression of GPC3. Flow cytometry results revealed intensive fluorescent signals in HepG2 and Huh-7 cells, with particularly strong signals in Huh-7 cells (Fig. S8A). This finding was corroborated by cellular immunofluorescence, which showed bright fluorescent signals on the membranes of HepG2 and Huh-7 cells, in contrast to the sporadic fluorescent brightness observed on MGC-803 cell membranes (Fig. S8B). Furthermore, Western blot analysis confirmed elevated expression of GPC3 proteins in HepG2 and Huh-7 cells compared to MGC-803 cells (Fig. S8C). Consistently with these findings, the mTriangle-AP613-1-liposomes exhibited strong staining in HepG2 and Huh-7 cells during both flow cytometry and cellular immunofluorescence assays compared to MGC-803 cells (Fig. 6). In contrast, no obvious signals were detected in any of the three cell lines when tested with the mTriangle-liposomes (Fig. S9). Additionally, competitive binding assays and time-dependent flow cytometry experiments were included to further validate the targeting ability of mTriangle-AP613-1-liposomes (Fig. S10). The cells were pre-cultured with AP613-1 (100 pmol) for 1 h, and then incubated with mTriangle-AP613-1-liposomes under the same conditions. The fluorescence intensity of both HepG2 and Huh-7 cells decreased significantly after co-incubation with AP613-1. Consistently, the fluorescence intensity increased with the elevation of the mTriangle-AP613-1-liposomes concentration in HepG2 and Huh-7 cells. These results collectively demonstrated that mTriangle-AP613-1-liposomes can specifically target GPC3-positive cells in vitro.

Fig. 6.

In vitro analysis of mTriangle-AP613-1-liposomes. (A) Flow cytometric profiles of mTriangle-AP613-1-liposomes binding to MGC-803, HepG2 and Huh-7 cells, respectively. Data are presented as mean ± standard deviation (SD) (n = 3). ^∗∗∗∗P < 0.0001. (B) Fluorescence imaging of MGC-803, HepG2, and Huh-7 cells stained by cyanine 5.5 (Cy5.5)-labeled mTriangle-AP613-1-liposomes for 2 h. Mild and intensive fluorescent signals could be observed on HepG2 and Huh-7 cells, respectively, while scarcely on MCG-803 cells.

3.5. Encapsulation of DOX

DOX is widely recognized as one of the most commonly used anti-cancer drugs, primarily due to its ability to directly interact with DNA, which underpins its antitumor efficacy. However, the free form of DOX is associated with significant cardiotoxicity, posing a challenge to its clinical use. Early research has demonstrated that encapsulating DOX within liposomes can substantially mitigate this cardiotoxicity. Importantly, this encapsulation does not compromise the drug's anticancer potency, thereby offering a more favorable therapeutic profile [18]. Here, we prepared DOX-encapsulated liposomes by (NH₄)₂SO₄ gradient method.

The ammonium sulfate gradient method utilizes an aqueous solution of (NH₄)₂SO₄ as the aqueous phase. Within the bilayer of the liposomes, (NH₄)₂SO₄ dissociates into ammonium ions (NH₄⁺), which can further dissociate into ammonia molecules (NH₃) and hydrogen ions (H⁺). The transmembrane rate and permeability coefficient of NH₃ are significantly higher than those of sulfate ions (SO₄²⁻). Consequently, by removing (NH₄)₂SO₄ from the aqueous phase outside the liposomes with 9.4% sucrose (pH 6.3), a concentration gradient of (NH₄)₂SO₄ is established between the internal and external aqueous phase of the liposomes. This concentration gradient acts as a driving force, allowing NH₃ to diffuse from the internal aqueous phase to the external aqueous phase. Concurrently, as NH₃ diffuses outward, H⁺ accumulates in the internal aqueous phase, leading to the establishment of pH gradient across the liposome membranes. According to the principle of ion equilibrium across membranes, these two gradients facilitate the preferential loading of weakly alkaline drugs from the outer aqueous phase into the lipophilic liposomes, against the concentration gradient (Fig. 7A).

Fig. 7.

Encapsulation of doxorubicin (DOX). (A) The mechanism of encapsulation of DOX. (B) Encapsulation of DOX. Ultrafiltration devices were used to separate the unencapsulated DOX. (C) Collection of the filtered solution. After 3−5 times of centrifugation, the discarded liquid became clear. The numbers 1−5 indicate the filtrates collected after the 1st to 5th centrifugation, respectively. (D) Evaluation of the stability of liposomes in liquid phase with different pH values (5.0, 6.5, and 7.4).

DOX was incubated with liposomes (m_DOX = 10%), which was dissolved in (NH₄)₂SO₄, at 65 °C for 15 min. Then the mixture was centrifuged at 12,000 rpm for 3 min, repeated 4–5 times with an ultrafiltration device (Fig. 7B) until the filtered waste became completely clear (Fig. 7C). The discarded solution was collected together to calculate the encapsulation efficiency (typically over 90%) according to the OD (λ = 480 nm) (Fig. S11). Furthermore, the established liposomes (DOX) were incubated with PBS at varying pH levels (5.0, 6.5, and 7.4) for at least 6 h in vitro. No DOX was found permeating from the liposomes, confirming their high stability (Fig. 7D).

3.6. Cytotoxicity analysis in vitro

Liposomes are typically regarded as nontoxic to cells, but high concentration can still cause cellular damage. To assess this, various concentrations of liposomes (0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4, 0.45, and 0.5 mM) were incubated with three cell lines for 24 h respectively, and cell viability was measured using CCK-8 reagents. Our data showed that 0.15 mM liposomes were relatively safe for cells, whereas concentrations above 0.2 mM caused significant cellular damage (Fig. 8A). Meanwhile, we employed IC₅₀ to evaluate the efficacy of both free-form of DOX and liposomes (DOX) in vitro. All of three cell lines were incubated with a series of two-fold dilutions of DOX at concentrations of 100, 50, 25, 12.5, 6.25, 3.125, 1.56, 0.78, and 0 μg/mL (control) for 24 h. The results indicated that the IC₅₀ value of liposomes (DOX) was lower than that of free DOX (except MGC-803). This can be attributed to liposomes' ability to improve cellular uptake efficiency, primarily through mechanisms such as membrane fusion, which facilitate more effective drug delivery into cells compared to free drugs (Figs. 8B and S12A). In order to further investigate the cell-killing effect of liposomes (DOX) in vitro, a cytotoxicity analysis assay was conducted using mTriangle-AP613-1, liposomes, and liposomes (DOX). The concentration of liposomes was adjusted to match that of liposomes (DOX), ensuring that both concentrations were equivalent to the concentration at IC₅₀ and fell within a safe range (MGC-803: approximately 0.03 mM; HepG2: approximately 0.18 mM; and Huh-7: approximately 0.06 mM). Additionally, an excess concentration of mTriangle-AP613-1 (>1 μM) was included. The results indicated that mTriangle-AP613-1 and liposomes were nontoxic to cells, whereas liposomes (DOX) induced cellular damage by releasing DOX into the medium (Fig. 8C). Additionally, cytotoxicity of free DOX, liposomes (DOX), and mTriangle-AP613-1-liposomes (DOX) (50 and 100 nM) against MGC-803, HepG2, and Huh-7 cells was performed in vitro. The results showed that under the same DOX concentration, the mTriangle-AP613-1-liposomes (DOX) (100 nM) exhibited a better cell-killing effect compared to the free DOX and the liposomes (DOX) in Huh-7 group (Fig. S12B).

Fig. 8.

In vitro cytotoxicity assay. (A) Cytotoxicity of different concentration of liposomes against MGC-803, HepG2, and Huh-7 cells, respectively. Data are presented as mean ± standard deviation (SD) (n = 6). (B) Half maximal inhibitory concentration (IC₅₀) of free doxorubicin (DOX) and liposomes (DOX) in MGC-803, HepG2 and Huh-7 cells, respectively. Data are presented as mean ± SD (n = 6). (C). Cytotoxicity of mTriangle-AP613-1, liposomes, and liposomes (DOX) against MGC-803, HepG2, and Huh-7 cells after 24, 48, and 72 h incubation, as analyzed by cell viability assay. Data are presented as mean ± SD (n = 4). ^∗P < 0.05, ^∗∗∗P < 0.001, and ^∗∗∗∗P < 0.0001. ns: not significant.

3.7. Safety assessment and antitumor efficacy in vivo

The antitumor efficacy of mTriangle-AP613-1-liposomes (DOX) was evaluated in vivo using Huh-7-tumor-bearing nude mice. Treatment commenced eight days post-xenograft, once the Huh-7 tumor volumes reached approximately 100 mm³. The tumor-bearing nude mice were divided into three groups and administered saline, mTriangle-liposomes (DOX), or mTriangle-AP613-1-liposomes (DOX) via tail vein injections every two days for a total of six doses (Fig. 9A). Each injection delivered a DOX dose of approximately 5 mg/kg and ssDNA nanostructures (Cy5.5 conjugated) at a concentration of about 250 pmol. Before the in vivo injection, we first conducted a nuclease degradation assay of AP613-1 and mTriangel-AP613-1 respectively (Fig. 9B). mTriangel-AP613-1 was substantially digested after 5 min while AP613-1 was thoroughly stable during the whole degradation process. We hypothesized that this might be because the hairpin structure formed by AP613-1 itself hinders the binding of DNase I to ssDNA. However, after 30 min of enzymatic digestion, the band of mTriangle-AP613-1, although weakened and diffused, still exhibited a distinct fluorescent signal and was not completely degraded into small fragments, indicating that it possesses strong nuclease resistance. During the treatment period, all groups experienced slight body weight loss (Fig. 9C). In the saline-treated group, tumor exhibited rapid growth, with average volumes reaching approximately 600 mm³ within 10 days (Fig. 9D). Notably, DOX treatment significantly inhibited tumor growth, with mTriangle-AP613-1-liposomes (DOX) demonstrating superior antitumor effects compared to mTriangle-liposomes (DOX) (average tumor volume: 126.2 mm³ vs. 300.95 mm³, ^∗∗P < 0.01; average tumor weight: 0.097 g vs. 0.2345 g, ^∗P < 0.05) (Figs. 9D and E). Consistently with these results, mTriangle-AP613-1-liposomes (DOX) also exhibited superior antitumor effects compared to free DOX (average tumor volume: 244.76 mm³ vs. 503.35 mm³, ^∗P < 0.05; average tumor weight: 0.1893 g vs. 0.3395 g, ^∗P < 0.05) (Fig. S13). Hematoxylin and eosin (H&E) staining experiments showed that there were different degrees of pathological changes in tumor tissues in the saline group and the liposomes group due to the rapid proliferation of tumor cells (Fig. S14). In the saline group, the tumor tissues showed extensive moderate hemorrhage, indicating that it has more aggressive biological behavior. Meanwhile, when using free DOX or mTriangle-AP613-1-liposomes (DOX), the tumor tissues exhibited notable necrosis, especially in the mTriangle-AP613-1-liposomes (DOX) group (Fig. S14). Different degrees of hemorrhage also indicate vascular damage following treatment. To further confirm whether the heart, the liver and the kidneys were not affected by DOX, serological tests on the saline group and the mTriangle-AP613-1-liposomes (DOX) group, which included three liver function indices (alanine transaminase (ALT), aspartate transaminase (AST), and albumin), three renal function indices (blood urea nitrogen (BUN), creatinine (Cre), and uric acid (UA)), and two myocardial enzyme spectrum indices (creatine kinase (CK) and lactate dehydrogenase (LDH)) were conducted respectively (Fig. S15). Compared with the saline group, the mTriangle-AP613-1-liposomes (DOX) group showed abnormal values in ALT and AST, suggesting that slight liver lesions may occur in this group due to drug metabolism. Meanwhile, all BUN values in the mTriangle-AP613-1-liposomes (DOX) group were abnormal; however, BUN cannot be used alone as an indicator for evaluating renal function impairment. In the LDH test, both the saline group and the mTriangle-AP613-1-liposomes (DOX) group had abnormal values. Since we performed cardiac blood collection on anesthetized nude mice, the increase in LDH might be related to myocardial cell damage during the blood collection process. Based on a comprehensive assessment, the mTriangle-AP613-1-liposomes (DOX) might indeed have a slight impact on the liver, but little impact on the kidneys and heart. Additionally, given the propensity of liposomes to accumulate in the liver, a biosafety assessment was conducted. This involved injecting the mTriangle-AP613-1-liposomes into healthy nude mice every other day, followed by testing the serum of these mice to evaluate liver function. The results indicated that there were no obvious abnormalities in liver function indicators when compared to the saline group (Fig. S16). H&E staining experiments and serological tests were performed by Servicebio.

Fig. 9.

In vivo antitumor efficacy of mTriangle-liposomes doxorubicin (DOX) and mTriangle-AP613-1-liposomes (DOX) in glypican-3 (GPC3)-positive tumor model. (A) Schematic illustration of the timeline for the therapeutic experiments. After Huh-7-tumor-bearing mice treated with saline, mTriangle-liposomes (DOX) and mTriangle-AP613-1-liposomes (DOX), the changes of body weight and tumor volumes were recorded in the experiment for the analysis of antitumor efficacy. (B) Nuclease degradation assay of AP613-1 and mTriangle-AP613-1. (C, D) Diagram of body weight and tumor volume. Data are presented as mean ± standard deviation (SD) (n = 4). (E) Representative ex vivo images of tumors and diagram of tumor weight. The numbers 1−4 indicate biologically independent tissues. Data are presented as mean ± SD (n = 4). ^∗P < 0.05, ^∗∗P < 0.01, ^∗∗∗P < 0.001, and ^∗∗∗∗P < 0.0001.

After tail vein injection of mTriangle/mTriangle-AP613-1-liposomes (DOX), tumor tissues were harvested for immunofluorescence analysis at 2 and 24 h post-injection. The immunofluorescence imaging of the tissues revealed that within 24 h, the mTriangle-AP613-1-liposomes (DOX) exhibited specifical accumulation in the tumor tissues, while the mTriangle-liposomes (DOX) did not show significant aggregation. Notably, no discernible signals were observed in the tumor tissues at the 2-h time point, possibly due to the challenging penetration of nanomaterials into the deep solid tissues (Fig. 10). These findings suggest that the mTriangle-AP613-1-liposomes (DOX) effectively recognizes GPC3-positive cells and exhibits efficient antitumor efficacy. Particularly, to further validate the targeting specificity of the aptamer-modified delivery system at the organism level, the immunofluorescence imaging data of other major organs and tumor following the administration of mTriangle-liposomes (DOX) (control) and mTriangle-AP613-1-liposomes (DOX) were performed at 6 h by the IVIS Lumina XRMS Series III fluorescence imaging system (PerkinElmer, Waltham, MA, USA) (Fig. S17). In the mTriangle-AP613-1-liposomes (DOX) group, although nanostructures inevitably accumulated in the liver and kidneys, the tumor tissues still exhibited a relatively strong fluorescence signal. However, in the mTriangle-liposomes (DOX) group, the fluorescence signal in the tumor tissues was very weak. These results indicate that the scaffold structure itself has no targeting ability to GPC3-positive tumors, and it is only after loading AP613-1 that targeting capability is acquired.

Fig. 10.

Frozen sections of tumor tissue. Tumor tissue imaging with mTriangle-liposomes (doxorubicin (DOX)) and mTriangle-AP613-1-liposomes (DOX) at (A) 2 h and (B) 24 h post-injections. Cy5.5: cyanine 5.5; DAPI: 4′,6-diamidino-2-phenylindole.

4. Discussion

HCC is a highly prevalent malignancy and a leading cause of cancer-related mortality. Its asymptomatic nature in the early stages often leads to a late-stage diagnosis for most patients. This late detection results in a dismal prognosis and poses significant challenges for treatment [1]. Consequently, there is an urgent imperative to develop innovative and effective therapeutic strategies for HCC. Among the emerging treatment modalities, targeted therapy, alongside immune checkpoint therapy, has shown great promise and achieved notable success in the management of HCC.

In this study, we engineered a liposome-based drug delivery system tailored specifically for GPC3-positive tumors. This system harnesses the GPC3-binding aptamer AP613-1 to recognize GPC3 proteins, enabling the direct delivery of therapeutic agents into tumor cells. In our prior work [12], we implemented a six-round CE-SELEX strategy. We utilized human recombinant GPC3 proteins for screening and subsequent flow cytometry to identify and eliminate aptamer candidates that bound to GPC3-negative cells. Afterward, we synthesized truncated forms of the aptamer candidates, ensuring they retained the key characteristic sequences of the full-length versions. Further analysis using flow cytometry with both GPC3-positive and GPC3-negative cells revealed that the truncated aptamer AP613-1, featuring a two-small-arm structure, exhibited a superior GPC3-binding affinity compared to other candidates. This 37-nucleotide aptamer was thus identified as a promising probe for GPC3-positive HCC.

However, in vivo positron emission tomography-computed tomography (PET-CT) imaging demonstrated that ^99mTc-labeled AP613-1 accumulated nonspecifically in the intestines, kidneys, and bladder. This nonspecific accumulation obscured the signals from the tumor sites. We hypothesized that the small molecular weight of AP613-1 contributed to its short-term systemic circulation and subsequent accumulation in the digestive and urinary tracts. To address this issue, we employed the DNA origami technique. By incorporating AP613-1 into a long ssDNA nanostructure (approximately 600 nucleotides) named mTriangle-AP613-1, we increased its molecular weight. This triangular nanostructure not only displayed enhanced biostability and biosafety in vivo but also exhibited an increased affinity for GPC3, with a K_D of 15.2 nM. Notably, we utilized a phagemid-helper-phage system for the large-scale production of mTriangle-AP613-1. This biotechnological approach offers significant advantages over traditional synthesis methods [8]. It is highly cost-effective and can yield substantially higher quantities of long ssDNA, facilitating large-scale applications.

Another significant innovation in this research was the integration of DNA-cleaving deoxyribozymes, specifically I-R1a, into the templates. This replaced the traditional protein enzymes used in the final processing step of the target ssDNA. In this process, the phagemid-helper-phage system was first used to extract phagemid ssDNA. Subsequently, the self-cleaving catalytic DNAs efficiently cleaved the target ssDNA from the phagemid ssDNA, eliminating the need for costly protein enzymes. This provided a convenient and economical method for large-scale production, streamlining the manufacturing process.

The ssDNA nanostructure-assisted liposome-sorting technique has been shown to generate liposome libraries with distinct sizes. Previous studies have confirmed that liposome size influences their uptake efficiency [15]. However, our study did not explore whether variations in liposome size impact their antitumor effects in vivo. Despite this limitation, this technology holds great potential for enabling a more comprehensive and systematic understanding of liposome applications in future research. It may open new avenues for optimizing liposome-based drug delivery systems.

In vitro, the mTriangle-AP613-1-liposomes demonstrated outstanding performance in binding to GPC3-positive cells, as evidenced by flow cytometry and cellular immunofluorescence analysis. To achieve potent tumor-killing efficacy, we encapsulated the broad-spectrum anticancer drug DOX within the liposomes using the ammonium sulfate gradient method. DOX was selected to overcome the common resistance mechanisms in tumors. In vivo, the mTriangle-AP613-1-liposomes (DOX) exhibited superior antitumor efficacy compared to the control group, without causing significant side effects. Tumor tissue imaging revealed the specific accumulation of ssDNA nanostructures and liposomes within 24 h. The lack of significant accumulation within 2 h was likely due to the challenges associated with the penetration of nanomaterials into deep solid tissues.

In conclusion, the ssDNA nanostructure-based medicine mTriangle-AP613-1-liposomes (DOX) developed in this study can precisely target and deliver drugs to GPC3-positive liver cancer cells. It showcases promising biosafety profiles and remarkable antitumor effects. The construction strategies employed in this research represent a significant step forward in the development of more effective DNA-based materials for cancer treatment. Moreover, these strategies hold potential for application in other biological fields in the future, offering new possibilities for targeted therapies and nanomedicine development.

5. Conclusion

This study successfully engineered a targeted nanomedicine platform, mTriangle-AP613-1-liposomes loaded with DOX, for the precise treatment of GPC3-positive HCC. By integrating the GPC3-specific aptamer AP613-1 into a self-assembled triangular ssDNA nanostructure–mTriangle-AP613-1 via a long ssDNA scaffold, we achieved significant improvements in binding affinity (K_D = 15.2 ± 0.21 nM) and biostability compared to the free aptamer, addressing key limitations of unmodified aptamers in targeted delivery. The functionalization of liposomes with mTriangle-AP613-1, enabled by cholesterol-mediated anchoring and size-sorting through isopycnic ultracentrifugation, resulted in a delivery system that selectively targeted GPC3-positive cells (HepG2 and Huh-7) in vitro, as confirmed by flow cytometry and fluorescence imaging. Encapsulation of DOX using the ammonium sulfate gradient method ensured high loading efficiency and stability, with the liposomal formulation exhibiting enhanced cytotoxicity against GPC3-positive cells (lower IC₅₀) relative to free DOX, without inherent toxicity from the nanostructural component itself. In vivo studies using Huh-7 tumor-bearing nude mice validated the therapeutic superiority of mTriangle-AP613-1-liposomes (DOX). This system demonstrated superior tumor growth inhibition compared to controls, with specific accumulation in tumor tissues within 24 h and minimal adverse effects on liver function, confirming its biosafety. Overall, mTriangle-AP613-1-liposomes (DOX) represents a robust and safe targeted delivery system for GPC3-positive HCC. This work establishes a scalable, aptamer-nanostructure-based strategy to enhance drug specificity and efficacy, laying a foundation for advancing precision oncology and potentially extending to other GPC3-overexpressing cancers.

CRediT authorship contribution statement

Hongxin Zhou: Writing – review & editing, Writing – original draft, Project administration, Methodology, Formal analysis, Data curation, Conceptualization. Shuangshuang Dong: Validation, Methodology, Data curation. Zhan Si: Software, Methodology, Data curation. Hao Niu: Methodology. Jianwen Mao: Methodology. Yiran Chen: Methodology. Yuhuan Li: Writing – review & editing, Validation, Methodology, Conceptualization. Weizhong Wu: Writing – review & editing, Supervision, Funding acquisition, Conceptualization.

Declaration of interest statement

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

Appendix A. Supplementary data

The following are the Supplementary data to this article.