2.1 Synthesis and characterization of UMAC-M-Gal

The photoactivatable Pc-ASO was initially designed, as shown in Supporting Information: Figure S1. To examine the UV light-activatable property of Pc-ASO, we designed a Förster resonance energy transfer (FRET) pair as a fluorescent reporter by hybridizing a quencher (BHQ1)-bearing PcDNA strand and a FAM-labeled ASO strand (FAM-ASO). With the formation of a double-stranded structure, energy was transferred from FAM to the BHQ1, from which we could monitor the ultraviolet photoactivation process in real time (Supporting Information: Figure S1a). The fluorescence emission of FAM decreased significantly and almost disappeared, confirming Pc-ASO formation. With UV irradiation, the fluorescence emission of Pc-ASO gradually recovered, indicating the destruction of the FRET pair structure. Then FAM-ASO was released from the double-stranded structure due to the destruction of the PC group (Supporting Information: Figure S1b). As a negative control, we designed a single-strand DNA with the same sequence as PcDNA but without modifying the PC bonds. The fluorescence of the double-strand (nPc-ASO) was quenched by modified FRET, consistent with Pc-ASO results. However, the fluorescence emission of double-stranded structures did not increase with UV irradiation (Supporting Information: Figure S2b). Meanwhile, the specificity of PC bonds during agarose gel electrophoresis analysis (Supporting Information: Figure S2c,d), as in agreement with our previous FRET pair results, suggesting that the UV light-activatable property relies on the PC bonds.

We used an aliovalent Ca²⁺ ion doping strategy to enhance fluorescence intensity and improve the crystallization performance in the UV to visible light region, which could be used to respond to ultraviolet light. First, NaYF₄:Yb/Tm/Ca UCNP with a diameter of ≈30 nm was prepared as described in our previous paper with slight modifications (Figure 1A and Supporting Information: Figure S3a).²⁶ The hexagonal-phase core–shell NaYF₄:Yb/Tm/Ca@NaYF₄:Yb/Nd UCNPs were prepared with a mean diameter of 42 nm using the same method, in which the NaYF₄:Yb/Tm/Ca core was encapsulated with an NaYF₄:Yb/Nd shell through epitaxial growth (Figure 1B  and Supporting Information: Figure S3b). The doping of Nd³⁺ as sensitizers reduced the strong water absorption of biological tissue at 980 nm and enabled the system to be excited at 808 nm to maximize the penetration depth in the tissue compared with the use of Yb³⁺ alone as the sensitizer (Supporting Information: Figure S4). Afterward, to prepare UCNP@MSN, oleic-acid-capped UCNP was modified and coated with mesoporous silica by random etching. A transmission electron microscope (TEM) (Figure 1C) showed that the particle size was approximately 57 nm. Meanwhile, we observed the mesoporous characteristics of UCNP@MSN (Figure 1C) and found that the mesoporous silicon layer was around 10 nm. Energy dispersive spectrome elemental analysis map and spectrum of UCNP@MSN (Figure 1D and Supporting Information: Figure S5) also confirmed successful preparation. High-resolution TEM (HRTEM) and X-ray diffraction revealed that the lattice spacing of UCNPs was 0.52 nm, and UCNPs had a hexagonal phase since their diffraction peaks corresponded to the hexagonal β-NaYF₄ (JCPDS Card No. 28-1192) (Figure 1E and Supporting Information: Figure S6). Besides, an additional diffraction wide peak at 22° was observed in UCNP@MSN, indicating successful preparation of UCNP@MSN, attributed to the amorphous MSN of the sample. The pore characteristics of mesoporous-silica-coated UCNP were also measured by an automatic surface area and porosity analyzer, which showed that the specific surface area, pore size, and the total pore volume of UCNP@MSN were≈2.74 nm, 373.89 m²/g, and 0.352 cm³/g, respectively (Figure 1F and Supporting Information: Figure S7). The size of drug molecule C39 was estimated to be ≈1.49 nm (was analyzed by Chem3D software); therefore, the drug molecule C39 could be loaded into UCNP@MSN due to its appropriate pore size. Furthermore, the upconverting emission spectra of UCNP and UCNP@MSN (Figure 1H and Supporting Information: Figure S8) showed that UCNP@MSN could efficiently convert 808 nm NIR into light with emission peaks at 345 and 361 nm. Compared with UCNP, a moderate decrease in the luminescence emission was observed after being modified with the mesoporous silica layer, which could be due to the quenching effect of the surface of the silica shell.

Next, FT-IR spectrometry was used to verify the functional modification of the nanoparticles. Fourier transform infrared spectra (FTIR) (Figure 1G) showed characteristic peaks at 3428, 1089, and 800 cm⁻¹, which confirmed the successful surface modification of MSN. Compared with UCNP@MSN's FTIR results, an additional characteristic peak (NH₂ ≈ 874 cm⁻¹) was observed, indicating the successful graft of NH₂. A COOH-modified single-stranded DNA was attached to UCNP@MSN via EDC/NHS condensation reaction between PcDNA and −NH₂ in UCNP@MSN. After the ASO strand capping as a gatekeeper and C39 was loaded, the UV-Vis absorption spectrum suggested successful preparation (Supporting Information: Figure S9). Furthermore, C39 and ASO were successfully loaded on UCNP@MSN with an estimated loading content of 7.3 and 101.5 μmol/g UCNP@MSN, confirmed by the difference method (Supporting Information: Figure S10). The zeta potential of nanoparticles showed that the charge of UCNP@MSN-PcASO (UMA) decreased from −28 to −35 mV, which illustrated the successful modification of PcASO on the surface of UCNP@MSN (Figure 2E).

To probe the feasibility of controlled release of the drug delivery system UMAC, the C39 controlled releasing assay was performed in vitro with or without 808 nm NIR in phosphate-buffered saline (PBS), including detection of the time-dependent C39 drug-releasing curve (Figure 1I). In the UMAC without 808 nm NIR, only ≈8.1% of C39 was released at 48 h, indicating that the as-designed photoactivatable Pc-ASO structure blocked the mesoporous channels of MSN. Therefore, the drugs were contained in the nanoparticles. In sharp contrast, in the group treated with 808 nm NIR, the release of C39 significantly increased, with ≈55.7% of C39 released at 24 h (≈50.5% of C39 was released without DNA capping), showing that NIR significantly triggered the release of C39 (Figure 1J). The results showed that UMAC reached 56.8% drug release after continuous irradiation. Subsequently, the drug retention capacity of UMAC was evaluated by the drug leak rate in different solutions. After 24 h of incubation with no irradiation, minimal accidental drug release was observed, indicating that the as-prepared nanocomposite possessed good drug retention ability (Supporting Information: Figure S11).

The membrane vesicles were obtained by sonication and then mechanical extrusion through a 200-nm membrane on an Avanti mini extruder, which was further used to encapsulate the USNP@MSN after post-modification with 4-aminophenyl β-d-galactopyranoside (Gal)-modified polyethylene glycol phospholipid (Gal-PEG-DSPE, Supporting Information: Figure S12) to form a Gal-PEG-RM wrapped UMAC (UMAC-M-Gal) (Figure 2A,B). SDS-PAGE protein analysis followed by coomassie brilliant blue staining demonstrated that RBC membrane proteins were widely retained by the nanocomposite (Figure 2C), showing a successful cell membrane coating on the UMAC. Dynamic light scattering (DLS) was further used to monitor nanocomposite size changes before and after the RBC membrane coating, which indicated that the size of UMAC-M-Gal was around 90 nm (Figure 2D) consistent with the results of the scanning electron microscope and transmission electron microscope. After the UMAC-vesicle fusion process, the hydrodynamic diameter of the resulting biomimetic nanoparticles increased by 30 nm (Figure 2D), and the ζ potential increased to the levels similar to RBC vesicles (Figure 2E). Moreover, we found that the obtained UMAC-M-Gal was stable in water, PBS, 100% fetal bovine serum (FBS), and 10% FBS-containing medium (Supporting Information: Figure S13a) and even possessed excellent stability for up to 14 days in FBS (Supporting Information: Figure S13b,c), assuring the feasibility of the biological experiments followed.

2.2 The evaluation of immune evasion in macrophages and pharmacokinetic study

After confirming the biomimetic membrane coating on the nanoparticles, we assessed the cellular uptake of the UMAC-M-Gal (ASO strand was labeled with FAM) in the liver-derived cells. Galactoside (Gal), modified on the biomimetic nanoparticles, could bind to the ASGPR protein (an endocytosis receptor) with high affinity and be specifically expressed in liver parenchyma cells. Before the experiments, the specific expression of ASGPR in liver cells was determined by western blot analysis, indicating that the receptor was almost exclusively expressed in hepatocytes compared with other tissue-derived cells (Figure 2F). Consequently, the follow-up experiments were performed using HepG2.2.15 cells or L02cells, which expressed more ASGPR protein on the cell membranes. Next, we tested HepG2.2.15 cells by incubating them with UMAC-M-Gal (RM-Gal labeled with Dil and ASO labeled with FAM). we observed the majority of signal for Dil (red) overlapped with FAM signal (green) surrounding the nuclei, which confirmed the successful coating of RBC membrane on the UMAC surface (Figure 2G).

To validate the ability of immune escape, RAW 264.7 murine macrophage-like cells were used to examine the antiphagocytosis capability of UMAC-M-Gal. The cells were incubated with various nanoparticles and then imaged using a confocal laser scanning microscope. We found that the group incubated with UMAC showed brighter green fluorescence than the group incubated with UMAC-M and UMAC-M-Gal (Figure 3A), reflecting the idea that the encapsulation of erythrocyte membrane could effectively reduce the immune clearance. Further, a macrophage uptake test was performed using inductively coupled plasma mass spectrometry (ICP-MS) to detect the content of Si within the cytoplasm, showing that RBC membrane-coated biomimetic nanoparticles (UMAC-M and UMAC-M-Gal) exhibited much lower uptake compared with the uncoated nanoparticles (Figure 3C). As mentioned above, the protection of cell membrane could endow the nanoparticles with the ability to resist immune clearance. During in vivo studies on blood samples at different time points after intravenous injection, we detected the silicon content by ICP-MS, which indicated that the blood retention of the particles coated in the cell membrane was significantly enhanced (Figure 3D). It has been established that the reticuloendothelial system (RES) contains phagocytes mainly distributed in the liver and spleen, which resulted in the accumulation of substantial nanoparticles during treatment. We found that significant Si accumulation of UMAC-M in both organs decreased significantly, which indicated that the cell membrane wrapping could reduce the uptake of RES to some extent (Figure 3E).

2.3 NIR light triggered release of ASO in vitro

Next, to verify that the drug delivery system possessed the potential for temporal and spatial controlled-release capacity, HepG2.2.15 cells were incubated with a medium containing FRET pair-labeled UMAC-M-Gal (PcASO) or UMAC-M-Gal (nPcASO). After coculturing for 3 h, an 808 nm CW (1.5 W/cm²) laser was used to treat the cells. We found almost no fluorescent signal in cells without irradiation, whereas significantly stronger green fluorescence was detected with NIR irradiation, which indicated that NIR triggered liberation of the ASO-FAM strand (Supporting Information: Figure S14a). Flow cytometry quantitation also showed consistent results (Supporting Information: Figure S14b). In contrast, the green fluorescence intensity in HepG2.2.15 cells incubated with FRET pair-labeled UMAC-M-Gal (nPcASO) was barely changed under irradiation (Supporting Information: Figure S15a,b), suggesting that the PC bond was crucial for the biomimetic drug delivery system.

2.4 The evaluation of cellular uptake and liver targeting

A weak green fluorescent signal was observed upon incubation with a free ASO strand as naked DNA could not penetrate the cell membrane. In stark contrast, the cells incubated with UMAC displayed a stronger fluorescence signal, indicating that the presence of nanoparticles could promote the entry of ASO strands into cells. Concurrently, both confocal laser scanning microscopy (CLSM) analysis and flow cytometry demonstrated that UMAC-M-Gal possessed the highest internalization efficiency in the whole group, which indicated that the biomimetic membrane vesicles modified with Gal molecules were crucial for improving the internalization of the UMAC drug delivery system (Figure 4A and Supporting Information: Figure S16). Almost all fluorescence signals were located within the intracellular region, meaning that the UMAC-M-Gal was taken up via the endocytic pathway rather than physical adsorption. Subsequently, we evaluated the internalization of UMAC-M-Gal in L02 cells by co-localization study and found that after 3 h of incubation, UMAC-M-Gal showed good co-localization with lysosomes and endosomes stained by Lysotracker Red, indicating that UMAC-M-Gal was more likely endocytosed into the cells (Supporting Information: Figure S17). Additionally, to further study the key role of Gal in hepatocyte targeting, an antibody blocking test was performed on L02 cells by using an anti-ASGPR receptor antibody. We found that UMAC-M-Gal possessed a weaker fluorescence signal in the cytoplasm after blocking (Supporting Information: Figure S18).

Ultimately, to evaluate the targeting capability of biomimetic nanoparticles, a follow-up study was performed using CLSM, flow cytometry, and ICP-MS. A confocal co-localization assay was performed in various non-liver cell lines (A549, HeLa, and MCF7 cells), indicating significantly weaker fluorescent accumulation of UMAC-M-Gal, compared with the intensive fluorescent signal of the L02 and HepG2.2.15 cells (Supporting Information: Figure S19a). Flow cytometry analysis showed that the amount of UMAC-M-Gal bound to the liver cells was significantly higher than non-liver cell lines after incubation (Figure 4B and Supporting Information: Figure S19b). Moreover, the ICP-MS results confirmed that the biomimetic nanoparticles modified with Gal demonstrated significantly more silicon accumulation (Supporting Information: Figure S19c). These results indicate that UMAC-M-Gal possessed a high hepatocyte targeting specificity. Afterward, the real-time biodistribution of UMAC-M-Gal in vivo was tracked using a modified in vivo fluorescence imaging system. At 24 h after the injection, the biomimetic drug delivery system UMAC-M-Gal, labeled with the lipophilic fluorescent dye DiR, detected massive accumulation in the liver (Figure 4C,D). To study the biodistribution of different nanoparticles in vivo, the number of nanoparticles in the major organs of the killed mice and blood samples were accurately analyzed by detecting the Si content. Consistent with the fluorescence imaging in vivo, nanoparticles modified with Gal possessed a good targeting ability to liver. Concurrently, the low accumulation of UMAC-M and UMAC-M-Gal in the spleen corroborated the encapsulation of erythrocyte membrane could endow nanoparticles with immune escape ability. Moreover, the high accumulation of UMAC-M-Gal in the liver was attributed to the targeting ability of Gal.

2.5 NIR light-triggered anti-HBV effects in vitro

To investigate the molecular biological effects of UMAC-M-Gal nanoparticles in a complex physiological environment, we investigated the capacity of the biomimetic system to suppress intracellular hepatitis B virus replication. Before that, considering the thermal efficiency of the 808 nm NIR and the NIR-responsive photothermal property of nanoparticles, we assessed temperature changes by irradiating the DMEM containing 10% FBS and the UMAC-M-Gal solution, which evaluated the risk of NIR laser-induced thermal effect. A high thermal effect was barely detected even at a concentration of 800 μg/ml and 5 W/cm² NIR irradiation up to 10 min (Supporting Information: Figure S20). L02 and HepG2.2.15 cells, a normal liver cell model and a cell model undergoing the integration of the HBV genome into the chromosomal DNA, were used to assess the in vitro cell viability of UMAC-M-Gal, suggesting negligible potential cytotoxic effects of nanoparticles. First, the cells were divided into four groups: the first group determined the effects of different concentrations of nanoparticles on cell viability, the second group was treated with 808 nm irradiation for 30 min, the third group was treated with 808 nm irradiation with different power density, the fourth group assessed the toxic effect of various concentrations of UMAC-M-Gal on cells under a fixed laser power and irradiation time. Interestingly, even when the laser power reached 5 W/cm² for 5 min or 1 W/cm² for 30 min, the cells in the irradiation group did not exhibit a significant cytotoxic effect, suggesting that near-infrared laser did not affect the survival rate of liver cells (Supporting Information: Figure S21b,c). Furthermore, the results suggested that over 90% of HepG2.2.15 and L02 cells survived after incubating with 800 μg/ml UMAC-M-Gal and 1.0 W/cm² irradiation for 10 min, showing the high biocompatibility and their potential value in further biomedical applications (Supporting Information: Figure S21d).

The following experiments were performed to evaluate the inhibitory effect of UMAC-M-Gal on HBV replication in vitro. First, we determined the difference in therapeutic efficacy between drug treatment and biomimetic drug delivery systems. We found that the combination of ASO and C39 synergistically potentiated the inhibitory effect against HBV (Figure 5D–G). Meanwhile, we observed that the bionic drug delivery system significantly enhanced the antiviral ability. Subsequently, we verified the manipulative role of near-infrared light and PC bonds during the inhibition of HBV replication. We measured HBV 3.5 kb RNA and intracellular HBV DNA by real-time quantitative reverse transcription polymerase chain reaction and real-time quantitative polymerase chain reaction. At the same time, the level of HBV virus antigens (such as HBsAg and HBeAg) in supernatants was detected with an enzyme-linked immunosorbent assay. Due to the absence of PC bonds, UMAC-M-Gal had no obvious effect on HBV replication with or without NIR irradiation, which could be induced by the inhibition of drug release by hybridization with nPcDNA (Figure 5H,I). In contrast, the HBV 3.5 kb RNA, intracellular HBV DNA, and HBV viral antigens of the group incubated with nanocomposites modified with PC bonds decreased significantly under NIR irradiation, confirming that the observed HBV inhibitory activity of UMAC-M-Gal was indeed induced by NIR-mediated activation (Figure 5J,K).

Finally, we determined the effect of enhancement of the biofilm and Gal on the ability of the biomimetic nano-delivery system to inhibit HBV replication. We found that the encapsulation of RBC membrane vesicles and the modification of Gal could promote the delivery of the biomimetic nanoparticles. Nevertheless, the effect was not significant (Supporting Information: Figure S22) and should be further verified in vivo experiments.

2.6 NIR light-triggered Anti-HBV Effects in vivo

HBV-transgenic mice (HBV-Tg C57 BL/6), encoding a 1.2-overlength copy of the HBV genome (serotype awy), were used as the animal model to evaluate the ability of the UMAC-M-Gal drug delivery system to suppress HBV replication in vivo. These transgenic mice synthesize high concentrations of viral particles and HBsAg, pre-S1, and HBeAg, which should be used for detailed studies of the replication and expression of HBV and for pathological studies of hepatitis. Before this, to assess the further biomedical application of the as-obtained nano-system, a hemolysis assay was performed, suggesting negligible hemolysis capacity of the nanoparticles (Supporting Information: Figure S23).

As shown in Figure 6A, the parameters related to HBV replication in the HBV-Tg mice serum declined strikingly on Day 15 after treatment with biomimetic nanoparticles and illumination. Among them, the serum HBV DNA in the mice treated with NPs loaded with ASO only decreased to 15% under near-infrared irradiation, and the mice treated with NPs loaded with ASO and C39 even decreased to 1.3% (Figure 6B), indicating the potential for a functional cure. Analogously, the detected concentrations of HBsAg and HBeAg were predominantly consistent with the above after treatment with the biomimetic nanoparticles and NIR but was not observed decline in the transgenic mice exposed to NIR irradiation only (Figure 6C,D). Furthermore, immunohistochemical (IHC) staining was used to investigate the expression of viral antigen (HBcAg and HBsAg) in the HBV transgenic mice liver (Figure 6F), showing a barely detectable signal on Day 15 after treatment with both UMAC-M-Gal nanoparticles and NIR irradiation. Moreover, histopathological analysis (H&E staining) of the organs suggested no organ injury, while a definite therapeutic effect in IHC staining was detected in the liver after treatment (Supporting Information: Figure S24). Meanwhile, there were no significant changes in body weight during treatment (Supporting Information: Figure S25a) and blood biochemical parameters after treatment (Supporting Information: Figure S25b–d). These experimental data confirmed that treatment with UMAC-M-Gal plus NIR irradiation possessed good biocompatibility and properties without systemic toxicity in vivo, exhibiting tremendous potential utility to achieve a functional cure for chronic HBV infection.