pH-responsive virus-drug conjugates for targeted PD-L1 silencing and enhanced cancer immunotherapy

Abstract

Clinical efficacy of virotherapy is often limited by poor transduction and off-target effects. To overcome these hurdles, we engineered a multi-stage, pH-responsive virus-drug conjugate (VDC) for targeted, cancer immunotherapy. Our VDC comprises a recombinant adeno-associated virus serotype 2 (AAV2) encoding a programmed death-ligand 1 (PD-L1) short hairpin RNA (shRNA), which is chemically conjugated to the transduction-enhancing doxorubicin (Dox) with minimally cytotoxic dosage. This AAV2-Dox is shielded by a polyethylene glycol (PEG) layer attached via a pH-sensitive benzoic-imine linker, designed to be stable at physiological pH but cleavable within the acidic tumor microenvironment. AAV2-Dox produced an equivalent enhancement of transgene expression at a much lower concentration (0.005 µg mL^− 1) than Dox alone (0.02 µg mL^− 1). The pH-sensitive PEG shield successfully renders the VDC inert at pH 7.4 while restoring full transduction and therapeutic efficacy upon acid-triggered unshielding. Mice treatment with the PEGylated AAV2-Dox leads to efficient PD-L1 knockdown, which enhanced T-cell infiltration and polarization toward M1 macrophages. This study establishes a smart VDC platform that integrates a transduction enhancer with a gene therapy payload under the control of an acidic tumor microenvironment-activated stealth shield. This strategy provides a new paradigm for developing safer and more potent nanomedicines for combinatorial cancer immunotherapy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04347-9.

Introduction

The inherent tumor selectivity and cell-killing capabilities of oncolytic viruses have generated considerable interest in their development as innovative cancer treatments [1]. While the U.S. Food and Drug Administration (FDA) have approved virotherapy as an intratumoral immunotherapy [2], oncolytic viruses alone have yielded only modest clinical benefits due to poor biodistribution [3]. This limitation has spurred interest in combination strategies, drawing inspiration from targeted therapies like antibody-drug conjugates (ADCs), which deliver tumor-specific cytotoxicity while minimizing chemotherapy dosage [4]. Extending this concept, virus-drug conjugates (VDCs) could potentially boost therapeutic effectiveness [5]. For example, by linking a transduction-enhancing agent to viruses with a natural specificity for target cells, thereby facilitating their co-delivery and potentially increasing the overall impact of the treatment. While doxorubicin (Dox) typically acts as a chemotherapeutic by causing DNA strand breaks to inhibit cell proliferation, using a minimally cytotoxic Dox dosage can paradoxically amplify the rate and efficiency of viral transduction [6].

The interaction between programmed cell death protein 1 (PD-1) on activated T cells and its ligand, programmed death-ligand 1 (PD-L1), forms a crucial inhibitory checkpoint that regulates T-cell activity [7]. Many cancers exploit this pathway by overexpressing PD-L1, allowing them to evade immune detection and suppress anti-tumor immune responses [7, 8]. Targeting PD-L1/PD-1 signaling has become a major focus in cancer immunotherapy. One promising approach for silencing cancer-related genes is RNA interference (RNAi), where small-interfering RNAs (siRNAs) are used to suppress gene expression [9]. While siRNAs have shown potential in cancer therapy, delivering them effectively remains a challenge [10]. Synthetic siRNA duplexes and plasmid-mediated siRNAs suffer from limitations such as short half-life, low transfection efficiency, and cytotoxicity associated with the delivery process [10]. These barriers have hindered their clinical application. Notably, an efficient method for generating small RNA triggers involves using DNA-encoded short hairpin RNAs (shRNAs) [11].

Recombinant adeno-associated virus (AAV) vectors, measuring approximately 20 nm, are a leading nanoplatform for gene delivery in the treatment of various human diseases [12]. AAV products have gained authorization for clinical use, but they continue to struggle with poor biodistribution [12, 13]. Leveraging AAVs as delivery vectors for shRNAs offers a promising alternative to siRNA-based therapies [14]. Taken together, we developed a therapeutic strategy conjugating AAV serotype 2 (AAV2) encoding PD-L1 shRNA with Dox treatment (AAV2-Dox). Unlike combination therapies, this ensured that the drug and virus utilize the same receptor-mediated endocytosis pathway [12, 15]. This approach enhanced the expression of PD-L1 shRNA in Dox-treated cancer cells, resulting in the knockdown of PD-L1 on their surface. The reduction in PD-L1 levels reactivated T cells, allowing them to initiate a tumor-specific immune response. Additionally, the distinct physicochemical characteristics of the tumor microenvironment—such as hypoxia and mildly acidic pH (6.7–7.0)—not only promote cancer progression but also offer key targets for enhancing the specificity and efficacy of virotherapy [3, 16, 17]. To create a pH delivery system responsive to the tumor microenvironment, we attached polyethene glycol (PEG) with stealth properties to the AAV2-Dox conjugate strategically utilizing a pH-sensitive benzoic imine linker (Fig. 1A). This linker acted as an environmental sensor: the benzoic imine’s dynamic covalent bond is stable under neutral pH conditions typical of healthy tissues [18], allowing the PEG layer to shield the VDC systemically. However, the linker is labile in the acidic environment often found in solid tumors. Exposure to this lower pH triggers hydrolytic cleavage of the linker and subsequent detachment of the PEG shield. This tumor microenvironment-specific activation unmasks the AAV2-Dox conjugate precisely at the target site, promoting enhanced uptake by cancer cells and localized release of its therapeutic components. Furthermore, AAV transduction involves viral entry via endocytosis, subsequent movement to the nucleus, capsid uncoating in the nucleolus, and final release of the viral genome to express the transgene [19]. Concurrently, the transduction-enhancing agent Dox with minimal cytotoxicity has been released from the AAV2-Dox conjugate within the nucleus. This integrated, pH-activated VDC approach could be engineered to overcome limitations of systemic delivery and holds significant promise for improving the targeted efficacy of cancer virotherapy.

Fig. 1.

Design and characterization of a virus-drug conjugate. (A) Concept for a pH-activated virus-drug conjugate (VDC) for Targeted virotherapy. This figure illustrates a strategy using an engineered AAV2 conjugate designed to overcome delivery challenges and combine chemotherapy with immunotherapy specifically within the tumor microenvironment. (B) The sequence map of pAAV-PD-L1 shRNA construct. (C) Schematic of AAV2-PD-L1 shRNA chemical conjugation to Dox via sulfo-SMCC chemistry targeting SH-containing AAV2. Inset photographs visually confirm the conjugation, displaying the characteristic red color of Dox in the AAV2-Dox solution. The conjugation is confirmed by size-exclusion chromatography, where the AAV2-Dox (green line) co-elutes with the AAV2 (blue line), distinct from the retention time of Dox-SMCC (red line). (D) Experimental protocol evaluating transgene transduction in MCA205 cells at 72–144 h following single or triple administration rounds (one dose per day for 3 consecutive days) of AAV2, Dox^pre-AAV2, or AAV2-Dox. (E) Cytotoxicity of AAV2, plasmid DNA, AAV2-Dox, and Dox^pre-AAV2 assessed by MTS assay in MCA205 cells at various Dox concentrations. Data are presented as the mean ± standard deviation (n = 6)

Results

Optimized sequence of PD-L1 siRNA and dosage of Dox for VDC

To identify the optimal sequence for AAV2-delivered PD-L1 inhibition, we screened four sequences of PD-L1 siRNAs. The sequence demonstrating the highest protein inhibition efficiency was selected for AAV2 construction. We first confirmed stable, non-toxic complex formation between PD-L1 siRNA and Maestrofectin (1/1 weight ratio) using gel electrophoresis (Fig. S1A) and MTS assays in MCA205 cells (Fig. S1B). Subsequently, screening four siRNAs via ELISA showed significant PD-L1 knockdown by all, with siRNA3 being the most effective (26.1 ± 3.3% inhibition; Fig. S2A), corroborated by fluorescence imaging (Fig. S2B). Based on its superior efficiency, the siRNA3 sequence was selected for the AAV2-PD-L1 shRNA construct (Fig. 1B and Fig. S3).

We validated a dual-action approach by chemically conjugating Dox to AAV2-PD-L1 shRNA surface thiols via sulfo-SMCC (Fig. 1C). Inset photographs visually confirmed the conjugation by showing the distinct red color of Dox in the AAV2-Dox solution. Successful conjugation was confirmed by size-exclusion chromatography (right panel, Fig. 1C), which showed that the AAV2-Dox (green line) co-elutes with AAV2 (blue line) in the void volume, clearly separated from the Dox-SMCC peak (red line). No peak was detected at the retention time of Dox-SMCC in the AAV2-Dox profile, indicating complete conjugation to the AAV2. Additionally, in vitro assays confirmed that Dox-SMCC retains the cytotoxic potency of Dox, yielding comparable dose-dependent viability profiles (Fig. S4A). To find the optimal Dox dosage for maximizing AAV2 transduction with minimal cytotoxicity, cytotoxicity assessments in MCA205 cells compared the AAV2-Dox conjugate, sequential Dox treatment/AAV2 infection (Dox^pre-AAV2), and unmodified AAV2 (Fig. 1D). Both AAV2-Dox (up to 0.005 µg mL^{− 1} Dox equiv.; 87.40 ± 2.6% cell viability) and Dox^pre-AAV2 (up to 0.02 µg mL^{− 1} Dox; 85.19 ± 2.5%% cell viability) exhibited minimal cytotoxicity compared to unmodified AAV2 (93.08 ± 5.2% cell viability; Fig. 1E). The superior efficiency of AAV2-Dox over Dox^pre-AAV2 was attributed to divergent cellular entry and trafficking mechanisms. While free Dox entered through passive diffusion and relies on stochastic intracellular transport, the AAV2-Dox conjugate was internalized via receptor-mediated endocytosis [12, 15]. Upon endocytosis, AAV traffics to the nucleus and uncoats in the nucleolus to release its genome [19]. AAV2-Dox showed minimal leakage under physiological conditions but released nearly 100% of its payload following thermal positive control (75 °C) (Fig. S4B). Notably, this design ensures that Dox release is mediated by the AAV’s natural nuclear uncoating pathway. The results implied that conjugating Dox to AAV2 makes its cytotoxicity more efficient by enhancing both cellular uptake and subsequent nuclear accumulation of Dox. To prioritize the role of doxorubicin as a transduction enhancer rather than a primary chemotherapeutic agent, these optimized, minimally cytotoxic conditions (0.005 µg mL^{− 1} for AAV2-Dox, 0.02 µg mL^{− 1} for Dox^pre-AAV2) were deemed suitable for efficient infection and subsequent PD-L1 silencing studies, leading the superior efficiency of our targeted catalyst approach.

VDC-enhanced viral transduction

To overcome the inherent barriers of AAV-mediated gene delivery, Dox leverages the host’s DNA repair machinery to bypass the critical rate-limiting step of second-strand synthesis [6, 20]. Leveraging Dox’s ability to enhance AAV2 transduction [6], we compared AAV2, Dox^pre-AAV2 (DOX pretreatment), and AAV2-Dox in MCA205 cells using single vs. triple treatments. Furthermore, following the triple-round conditions, these treatments of Dox^pre-AAV2 and AAV2-Dox resulted in cell viability of 80.00 ± 6.6% and 82.51 ± 6.4% (Fig. S5). Transduction efficiency from triple Dox^pre-AAV2 (91.36%) and AAV2-Dox (95.77%) treatments were greater than triple AAV2 alone (83.19 ± 0.005%), measured by flow cytometry (Fig. 2A). A triple treatment with AAV2-Dox showed ~ 1.2-fold higher EGFP expression compared to a single rounds (p < 0.05), and was ultimately superior to the gain from the single AAV2 treatment (64.57 ± 0.004%) to the triple treatment (83.19 ± 0.005%). Imaging confirmed these results (Fig. 2B). Notably, AAV2-Dox achieved comparable enhancement at a lower Dox concentration (0.005 µg mL^{− 1}) than Dox^pre-AAV2 (0.02 µg mL^{− 1}), suggesting a benefit of VDC delivery. These findings demonstrate that the low-dose Dox conjugation regimen not only preserves AAV2 function but actively enhances transduction efficiency, indicating a synergistic, non-compromising effect between the conjugate components.

Fig. 2.

VDC-mediated enhancement of aav2 transduction and pd-l1 knockdown drives in vitro antitumor immunity. (A) Transduction efficiency of AAV2 analyzed by flow cytometry, showing enhanced EGFP⁺ cell percentages following triple rounds of Dox^pre-AAV2 and AAV2-Dox compared to AAV2 alone (*, p < 0.05, two-tailed unpaired Student’s t test). Data are presented as the mean ± standard deviation (n = 6). (B) Fluorescence microscopy images of EGFP expression in MCA205 cells treated with AAV2, Dox^pre-AAV2, or AAV2-Dox. Scale bar = 100 μm. (C) Levels of PD-L1 protein measured by ELISA following single or triple rounds of treatment, with AAV2, Dox^pre-AAV2, and AAV2-Dox (*, p < 0.05, two-tailed unpaired Student’s t test). Data are presented as the mean ± standard deviation (n = 6). (D) Fluorescence microscopy images of PD-L1 silencing in MCA205 cells under different treatment conditions. Scale bar = 20 μm. (E) Characterization of CTLL-2 cells showing expression of surface receptors. (F) Protocol for the analysis of MCA205 cells previously subjected to various treatments and co-cultured with CTLL-2 cells. (G) Quantification of MCA205 cell viability after 24, 48, 72, and 96 h of co-culture with CTLL-2 cells. MCA205 cells were pre-treated with AAV2 encoding PD-L1 shRNA (AAV2 only), Dox^pre-AAV2, or AAV2-Dox for either single (left panel) or triple rounds (right panel). Data are presented as the mean ± standard deviation (n = 8). Representative fluorescence microscopy images of RFP-expressing MCA205 from the co-culture experiment with CTLL-2 cells at 48 h incubation, visually confirming cancer cell numbers following AAV2, Dox^pre-AAV2, and AAV2-Dox treatments. Scale bar = 100 μm

We next evaluated PD-L1 silencing in MCA205 cells treated with AAV2, Dox^pre-AAV2, or AAV2-Dox. While single treatments substantially reduced PD-L1 levels, triple-round treatments offered significantly greater benefit in enhanced silencing (Fig. 2C). Notably, triple Dox^pre-AAV2 and AAV2-Dox treatments resulted in greater PD-L1 reduction (67.99% and 73.36%, respectively) compared to AAV2 alone (54.51%), corroborated by fluorescence imaging (Fig. 2D). To assess the functional consequence of this enhanced silencing on T-cell interactions, treated MCA205 cells were co-cultured with murine CTLL-2 T cells expressing CD8⁺ and PD-1 (Fig. 2E). The protocol used for MCA205 cell exposure to various treatment conditions while in co-culture with CTLL-2 cells is summarized in Fig. 2F. AAV2-Dox treatment significantly increased CTLL-2-mediated killing of MCA205 cells, reducing viability to 33.06 ± 1.6% (single round) and 21.7 ± 1.1% (triple round) at 48 h, representing over 1.5-fold enhancement over AAV2 alone (Fig. 2G). Dox^pre-AAV2 treatment yielded comparable improvements in T-cell killing. Fluorescence imaging of RFP-expressing MCA205 cells in co-culture confirmed increased elimination following Dox^pre-AAV2 or AAV2-Dox treatments (central panel, Fig. 2G). These results demonstrate that the unique strategy of AAV2-Dox enhance PD-L1 silencing and subsequent T-cell mediated cancer cell elimination, achieving potent effects at a lower Dox concentration compared to the pretreatment approach.

pH-activated VDC

With the aim of improving in vivo biodistribution and tumor specific transduction, AAV2-Dox was chemically conjugated with amine-terminated PEG using terephthalaldehyde as a linker, targeting the amino groups on AAV surface proteins (Fig. 3A). This PEGylation was performed at various molar ratios of PEG to terephthalaldehyde. To assess the impact of a stealth effect of PEGylation on AAV2 transduction efficiency, we measured EGFP expression in MCA205 cells via flow cytometry at 3 days post-transduction (upper panel, Fig. 3B and Fig. S6A). Compared to non-PEGylated AAV2-Dox, PEGylated AAV2-Dox under analogous conditions, maintained similar EGFP expression levels at a PEG/terephthalaldehyde molar ratio of 3 (p = 0.07). However, transduction efficiency significantly decreased to 50.5% at a 0.75 ratio and to 14.7% at a 0.5 ratio. We also evaluated AAV2-Dox capsid integrity using a dot blot assay across various PEG/terephthalaldehyde ratios (Fig. S6B). The sample prepared at a ratio of 3 or 1.5 exhibited the better signal intensity, indicating optimal preservation of capsid structure and antigenicity. In contrast, lower ratios showed significantly reduced immunoreactivity, suggesting steric hindrance of PEGylation. Furthermore, neutralizing antibodies (NAbs) neutralize infection by binding conformational or linear epitopes, notably the AAV2’s VP1/2 N-termini, thereby blocking essential intracellular processes [21]. In the presence of anti-VP1 neutralizing antibody, naked AAV2 susceptibility was evident, with transduction efficiency dropping progressively across the 1st, 2nd, and 3rd incubations (from 34.6 ± 6.7% to < 5%) (Fig. S7). In contrast, PEGylated AAV2-Dox maintained a consistently low transduction profile across all antibody incubation steps, consistent with its shielded state at pH 7.4, suggesting that the PEG layer effectively prevents antibody interaction or capsid internalization under neutral conditions. Subsequent pH 6.7 exposure reactivated PEGylated AAV2-Dox (65.4 ± 5.3%) while naked AAV2 remained neutralized. Taken together, these findings indicate that the degree of PEGylation, which correlates with the virus-cell interaction of the PEGylated AAV2-Dox, critically influences viral transduction efficiency. Importantly, no cytotoxicity was observed in MCA205 cells incubated with PEGylated AAV2-Dox at any of the tested molar ratios (lower panel, Fig. 3B). Taken together, we found that PEGylated AAV2-Dox, especially when formulated at an optimized molar ratio of 0.5, is well-suited for preventing transduction in cells under normal conditions.

Fig. 3.

pH-activated VDC: characterization, and pH-responsive transduction. (A) Schematic illustration of the chemical conjugation of AAV2-Dox with amine-terminated PEG using terephthalaldehyde as a linker, targeting AAV surface amino groups. (B) Impact of PEGylation on AAV2 transduction and cytotoxicity in MCA205 cells. EGFP expression measured by flow cytometry 3 days post-transduction in MCA205 cells. Cells were treated with PEGylated AAV2-Dox at various PEG/terephthalaldehyde molar ratios alone (*, p < 0.05; #, p > 0.05, two-tailed unpaired Student’s t test) (upper panel). Cytotoxicity assessment of the same PEGylated AAV2-Dox ratios in MCA205 cells after 24 h of incubation at pH 7.4 (lower panel). Data are presented as the mean ± standard deviation (n = 3). (C) pH-sensitive transduction. EGFP expression in MCA205 cells cultured at pH 7.4 (neutral) or pH 6.7 (acidic) and incubated with AAV2, AAV2-Dox, PEGylated AAV2-Dox, or non-pH-responsive PEG-AAV2-Dox (*, p < 0.05; #, p > 0.05, two-tailed unpaired Student’s t test). Data are presented as the mean ± standard deviation (n = 3). (D) pH-dependent therapeutic effect. Reduction in PD-L1-expressing MCA205 cells following treatment with AAV2, AAV2-Dox, PEGylated AAV2-Dox, or non-pH-responsive PEG-AAV2-Dox at acidic pH compared to neutral pH alone (*, p < 0.05; #, p > 0.05, two-tailed unpaired Student’s t test). Data are presented as the mean ± standard deviation (n = 3). (E) Transmission electron micrographs (TEM) comparing PEGylated AAV2-Dox at pH 7.4 versus pH 6.7 (arrow indicates PEG layer). Scale bar: 50 nm. (F) MD simulation illustrating the pH-triggered unshielding mechanism, showing PEG chain dissociation from AAV2-Dox at acidic pH 6.7

A non-pH-responsive PEG-AAV2-Dox control was used to verify that pH-mediated de-PEGylation is responsible for restoring infectivity (Fig. S8). To investigate the influence of pH on viral transduction, MCA205 cells were cultured in complete medium adjusted to either pH 7.4 or pH 6.7 and subsequently incubated with AAV2, AAV2-Dox, or PEGylated AAV2-Dox, or non-pH-sensitive PEGylated AAV2-Dox. At pH 7.4, the PEGylated AAV2-Dox and non-pH-sensitive PEGylated AAV2-Dox exhibited a stealth effect, resulting in lower transduction efficiency compared to AAV2 or AAV2-Dox (p = 3.12 × 10^− 6 or 1.8 × 10^− 6). At the acidic pH of 6.7, PEGylated AAV2-Dox significantly enhanced AAV2-mediated EGFP expression in MCA205 cells compared to PEGylated AAV2-Dox at pH 7.4 (p = 3.7 × 10^− 6, Fig. 3C and Fig. S9). pH-driven hydrolysis leads to the detachment of the PEG shield, exposing the naked AAV2-Dox and thereby improving its viral infectivity. As seen with EGFP, the loss of the PEG shield at this lower pH led to a clear reduction in PD-L1-expressing cells (Fig. 3D). Notably, the non-pH-sensitive PEGylated AAV2-Dox maintained its low transduction levels and high PD-L1 expression even at pH 7.4 or 6.7. Furthermore, we confirmed that acidic conditions also impact the delivery of a therapeutic PD-L1 shRNA to MCA205 cells. Overall, these findings indicate that an acidic pH promotes the hydrolysis of the benzoic-imine bond within the PEGylated AAV2-Dox, allowing the AAV2-Dox to unmask only at the target site.

Compared to the stable morphology of AAV2 alone (Fig. S10), PEGylated AAV2-Dox demonstrated a loss of the outer layer PEG structure when the pH was lowered from 7.4 to 6.7, as observed through transmission electron micrographs (TEM) analysis (Fig. 3E). Furthermore, the PEGylated AAV2-Dox had a hydrodynamic diameter of 21.7 ± 2.8 nm at pH 6.7 and 32.5 ± 3.3 nm at pH 7.4 (Fig. S11A). In contrast, AAV2 alone showed no significant size change, with diameters of 22.4 ± 3.9 nm and 22.7 ± 3.6 nm at pH 7.4 and pH 6.7, respectively. The PEGylated AAV2-Dox maintained comparable hydrodynamic diameters in complex biological fluid (culture medium: 39.9 ± 4.9 nm) and simple buffer (PBS: 36.1 ± 3.5 nm). This similarity indicates that the PEGylated AAV2-Dox resist aggregation or destabilization when introduced into cell culture medium (Fig. S11A). While AAV2 remained negatively charged regardless of pH, PEGylated AAV2-Dox exhibited effective charge shielding at pH 7.4 (-2.3 ± 1.1 mV) (Fig. S11B). However, at pH 6.7, its potential shifted to -13.1 mV, effectively restoring the native capsid’s negative charge. The PEGylated AAV2-Dox exhibited colloidal stability, maintaining a consistent hydrodynamic diameter without significant aggregation in either PBS or culture medium over the 72-hour incubation period (Fig. S12). The PEGylation is thus critical for enabling this disintegration, which is expected to modulate AAV2 delivery and therapeutic performance during systemic circulation.

This morphological change of PEGylated AAV2-Dox in the TEM images aligned with results from molecular dynamics (MD) simulations (Fig. 3F). MD simulations demonstrate that the observed pH-responsive behavior does not originate from the PEG chain itself, which remains chemically inert to pH variations. Instead, the pH sensitivity is attributed to the benzoic imine linkage and the protonation state of histidine residues of AAV2, which undergo protonation near this pH range. At pH 7.4, deprotonation stabilizes the interaction between the AAV2 surface and the conjugated PEG molecule. In contrast, at pH 6.7, protonation of histidine and the benzoic imine bond disrupts this interaction, leading to PEG detachment from the viral surface. This is reflected in a significant shift in activation energy, from − 20,074.23 ± 0.89 kcal·mol^− 1 at pH 7.4 to 213.34 ± 2.09 kcal·mol^− 1 at pH 6.7 (Fig. 3F), indicating pH-induced structural rearrangements that promote AAV2-cell interactions. This unshielding mechanism markedly enhances AAV2-Dox infectivity, underscoring the system’s responsiveness to acidic tumor microenvironments.

Preclinical treatment of VDC

The MCA205 mouse fibrosarcoma model is a well-established benchmark in cancer immunotherapy research. We selected this cell line specifically for its high constitutive or inducible PD-L1 expression, which renders it an ideal platform for evaluating PD-L1-targeted therapies [22]. Once subcutaneous MCA205 tumors reached approximately 200 mm³ in volume in C57BL/6 mice, experimental groups then received either PEGylated AAV2-Dox, PEGylated AAV2, AAV2-Dox, Dox^pre-AAV2, AAV2 alone, or PBS via tail-vein injection (Fig. 4A). The in vivo therapeutic efficacy of AAV2-Dox was significantly enhanced through PEGylation, as determined by tumor weight and volume on Day 15 (Fig. 4B and D). PEGylated AAV2-Dox treatment achieved a notable, > 2-fold reduction in tumor weight (0.717 ± 0.172 g) and volume (753.24 ± 130.60 mm³) compared to the PBS group (tumor weight: 1.783 ± 0.279 g, p = 1.21 × 10⁻⁵; tumor volume: 1699.92 ± 171.17 mm³, p = 8.02 × 10⁻⁷). Interestingly, the Dox^pre-AAV2 demonstrated slight tumor inhibition (1513.20 ± 75.11 mm³) potentially due to systemic Dox, while the PEGylated AAV2 vector alone exhibited moderate efficacy (1086.57 ± 60.74 mm³) relative to the PBS control. However, the therapeutic outcomes of these groups were significantly inferior to the PEGylated AAV2-Dox. Conversely, and as anticipated, unmodified AAV2, Dox^pre-AAV2, and AAV2-Dox exhibited tumor growth profiles comparable to the PBS control group. While AAV vectors are established clinical tools capable of high-dosage administration [3, 5], their systemic application is fundamentally constrained by poor biodistribution [16, 17, 23].

Fig. 4.

Preclinical Therapeutics for PEGylated AAV2-Dox. (A) Schematic of the in vivo experimental design for preclinical virotherapy, showing treatment administration via tail-vein injection in tumor-bearing mice. (B) Tumor growth curves for individual mice bearing MCA205 tumors under various treatments. Tumor volume (C) and tumor weight (D) in mice treated with PBS, AAV2 alone, Dox^pre-AAV2, AAV2-Dox, PEGylated AAV2, or PEGylated AAV2-Dox (*, p < 0.05, two-tailed unpaired Student’s t test). Data represent the mean ± standard deviation (n = 6). Representative images of MCA205 xenograft tumors after various treatments are shown. Bar = 1 cm. (E) Relative mRNA levels of PD-L1 and GAPDH analyzed following treatment with PBS, AAV2, Dox^pre-AAV2, AAV2-Dox, PEGylated AAV2, or PEGylated AAV2-Dox (*, p < 0.05, two-tailed unpaired Student’s t test). GAPDH expression served as an internal housekeeping control to confirm the consistency. Data represent the mean ± standard deviation (n = 6). (F) Quantitative analysis of IFN-γ and Granzyme B levels in tumor tissue from mice treated with PBS, AAV2, Dox^pre-AAV2, AAV2-Dox, PEGylated AAV2, or PEGylated AAV2-Dox (*, p < 0.05, two-tailed unpaired Student’s t test). Data represent the mean ± standard deviation (n = 6). (G) Mouse body weight changes over time across different treatment groups. Data represent the mean ± standard deviation (n = 6). (H) In vivo biocompatibility of various materials. Biochemical parameters in treated mice. Data represent the mean ± standard deviation (n = 6). (I) Histological sections of major organs (liver, lung, kidney, heart, and spleen) from mice after various treatments. Representative images are shown. Scale bar = 200 μm

Treatment with PEGylated AAV2-Dox resulted in a specific and significant downregulation of PD-L1 mRNA expression compared to the PBS control (0.64 ± 0.47 vs. 1.10 ± 0.12; p = 0.0334) (Fig. 4E). Notably, neither AAV2, Dox^pre-AAV2, nor AAV2-Dox achieved this effect, suggesting that pH-sensitive PEGylation is crucial for the delivery efficacy of AAV2 in this model. The GAPDH expression levels confirmed the specificity of these results. To evaluate the potential for immune activation, the levels of the cytotoxic cytokines interferon gamma (IFN-γ) and Granzyme B were quantified across the treatment groups (Fig. 4F). The groups treated with PBS, AAV2, Dox^pre-AAV2, and AAV2-Dox exhibited low, basal levels of both markers. In contrast, the PEGylated AAV2-Dox or PEGylated AAV2 treatment resulted in a robust immune response, showing the highest concentrations of both IFN-γ and Granzyme B. Statistical analysis confirmed that the levels in the PEGylated AAV2-Dox group were significantly higher than those in the PBS (IFN-γ, p = 3.15 × 10^{− 7}; Granzyme B, p = 1.1 × 10^{− 9}) and non-PEGylated control groups (p < 1 × 10^{− 5}), indicating that this formulation effectively stimulates the release of key cytotoxic effectors from immune cells. This substantial improvement in delivery and treatment efficacy with PEGylated AAV2-Dox is likely attributed, at least partially, to the tumor microenvironment pH-triggered release of AAV2-Dox, corroborating previous pH sensitivity findings [16, 17].

To assess biocompatibility, we measured body weight and biochemical parameters including GPT, GOT, TBIL, BUN, and CRE in mice treated with PEGylated AAV2-Dox, PEGylated AAV2, AAV2-Dox, Dox^pre-AAV2, AAV2, or PBS. In the PEGylated AAV2-Dox group, no significant changes were detected in these markers on Days 21 and 28 (Fig. S13). Across all groups, mouse body weights tended to increase over time, likely a result of tumor growth (Fig. 4G). No obvious abnormalities were detected in these markers (Fig. 4H). Subsequently, histopathological examination of hematoxylin and eosin (H&E) stained major organs was conducted to identify any further side effects not indicated by the biochemical data (Fig. 4I). This analysis revealed almost no injuries or pathological changes. Notably, even with substantial AAV2, Dox^pre-AAV2, AAV2-Dox, PEGylated AAV2, or PEGylated AAV2-Dox accumulation in the liver [16, 17, 23], no associated abnormalities were detected in this organ. To evaluate the pH-triggered strategy re-directed AAV2 distribution, we quantified AAV genome copy numbers in tumor and non-target organs by quantitative polymerase chain reaction (qPCR). The treatments of AAV2 and AAV2-Dox resulted in primary accumulation in the liver [16, 17, 23], with genome copies exceeding the tumor load by over 40-fold. In contrast, the PEGylated AAV2-Dox showed an altered biodistribution, with the liver accumulating 3.75-fold more genome copies than the tumor (Fig. S14). By combining the stealth properties of PEG with pH-sensitive activation, we successfully reduced hepatic sequestration, a major clinical barrier for AAV2 [3, 5], from a 40-fold liver-to-tumor ratio to just 3.75-fold. Taken together, these in vivo findings indicate that PEGylated AAV2-Dox is very well tolerated, demonstrating excellent compatibility and no discernible toxicity in mice, positioning it as a promising carrier for nano-modified virotherapy.

Distribution of immune cells in tumor microenvironment

Key functions of the spleen include extramedullary hematopoiesis and the promotion of tumor immunotolerance [23]. Analysis of spleen size confirmed a strong immune response in all tumour conditions, including those treated with PEGylated AAV2-Dox. Compared to healthy controls, mice in the PBS, AAV2, Dox^pre-AAV2, AAV2-Dox, PEGylated AAV2, and PEGylated AAV2-Dox groups all exhibited significantly enlarged spleens (all p < 0.05; Fig. S15), consistent with reported immune responses against tumor cells [24]. This finding demonstrates that PEGylation of AAV2-Dox does not compromise effective immune activation.

PD-1 on cytotoxic T-cells binds its ligand, PD-L1, to suppress beneficial anti-tumor immune responses [7, 8, 24]. By binding to PD-1, anti-PD-1 antibodies disrupt this suppressive interaction, thus overcoming immunotolerance and eliciting an anti-tumor effect [7, 8, 24]. Key mediators of this anti-tumor immunity are tumor-infiltrating leukocytes (TILs), predominantly T cells, B cells, macrophages, and other immune cells [25]. While often scarce, TILs possess prognostic value [13], and notably, anti-PD-1 antibody treatment has been associated with their increased infiltration into tumors [24]. To evaluate the effect of PD-L1 shRNA encoded by AAV2 on the immune system, we measured CD8⁺ and CD4⁺ T cells, B cells, M1 macrophages, and M2 macrophages levels in MCA205 xenograft tumor samples (Fig. 5A). The PEGylated AAV2-Dox group exhibited a significant increase in CD8⁺ T cells compared to the PEGylated AAV2 group (p = 0.0031), Dox^pre-AAV2 group (p = 9.8 × 10^{− 6}), the AAV2-Dox group (p = 8.4 × 10^{− 8}), the AAV2 alone group (p = 7.8 × 10^{− 8}), or the PBS group (p = 8.5 × 10^{− 7}), respectively (Fig. 5B). Treatment with PEGylated AAV2-Dox or PEGylated AAV2 resulted in a more than 1.8-fold increase in CD4⁺ T cells compared to all other groups (Dox^pre-AAV2, AAV2-Dox, and AAV2). B cell recruitment remained statistically unchanged across all treatment groups; however, only the PEGylated AAV2-Dox or PEGylated AAV2 treatment caused a minor, non-significant effect on B cell numbers. These results suggest that PEGylated AAV2-Dox effectively accumulates in MCA205 xenograft tumors, generating an immunomodulatory effect that strongly suppresses tumor growth and supports the prognostic value of T cells in xenograft tumors.

Fig. 5.

Modulation of the tumor immune microenvironment by different therapeutic regimens. (A) Representative immunofluorescence images in xenograft tumors. The densities of CD8⁺ T cells (black), CD4⁺ T cells (green), CD19⁺ B cells (blue), M1 macrophages (F4/80⁺MHC II⁺CD206⁻; red), and M2 macrophages (F4/80⁺MHC II⁻CD206⁺; grey) are presented. (B) Quantification of immune cells in xenograft tumors (*, p < 0.05; #, p > 0.05, two-tailed unpaired Student’s t test). Data represent the mean ± standard error of mean (SEM) (n = 80)

Macrophages, prevalent immune cells in the tumor microenvironment, can be polarized into either tumor-suppressive M1 or tumor-supportive M2 phenotypes [26]. A positive feedback loop that enhances tumor elimination occurs when CD8⁺ T cells promote M1 polarization, and these M1 macrophages in turn recruit more CD8⁺ T cells. Conversely, a high concentration of M2 macrophages fosters tumor growth by suppressing the CD8⁺ and Th1 T cell response, thereby reducing the body’s tumor-killing ability. Standard AAV2, Dox^pre-AAV2, or AAV2-Dox treatment had a negligible effect on M1 macrophage recruitment compared to PBS (Fig. 5B). This suggests that under common therapy, tumors may promote a switch from tumor-suppressive to tumor-supportive macrophages, correlating with induced tumor growth (Fig. 4C and D). Treatment with PEGylated AAV2-Dox or PEGylated AAV2 significantly increased M1 recruitment by more than three-fold compared to other groups. Notably, PEGylated AAV2-Dox significantly increased the M1/M2 macrophage ratio compared to PBS only (p = 7.8 × 10^{− 6}), AAV2 (p = 0.0017), Dox^pre-AAV2 (p = 4.4 × 10^{− 6}), AAV2-Dox alone (p = 9.7 × 10^{− 5}), or PEGylated AAV2 (p = 0.0024) (Fig. 5B). Consistent with the data in Fig. 5B, fluorescent imaging visually confirmed the presence of M1-type macrophages, M2-type macrophages, CD8⁺ T cells, CD4⁺ T cells, and B cells after various treatments (Fig. S16-S21). Similarly, the immune cell composition via flow cytometry clearly demonstrated that the PEGylated AAV2-Dox elicits the better potent immune activation within the tumor microenvironment, characterized by the recruitment of T cells and macrophages (Fig. S22). These findings strongly suggest that PEGylated AAV2-Dox acts as a potent immunomodulator by restoring T cell function and shifting the macrophage population towards an M1 phenotype, which leads to enhanced tumor shrinkage.

Discussion

Recombinant AAVs are established vectors for in vivo gene transfer, distinguished by an excellent safety record in therapeutic settings [12, 13]. Despite this success, their application is limited by insufficient transduction efficiency, which often necessitates high vector doses. To improve AAV transduction, we investigated an ADC-based strategy. The Dox’s ability to enhance AAV transgene expression is likely driven by multiple mechanisms [6, 20]. Specifically, whether administered as a pretreatment or via co-delivery, Dox significantly enhances AAV2 transduction efficiency by overcoming critical rate-limiting barriers within the viral life cycle [6, 20]. Mechanistically, doxorubicin acts as a topoisomerase II inhibitor to trigger a cellular DNA damage response, which effectively mobilizes host DNA repair factors and polymerases. This recruitment bypasses a critical bottleneck by accelerating the second-strand synthesis of the AAV genome, ultimately leading to a significant augmentation of transgene expression. Consistent with the action of other DNA-damaging agents [27, 28], Dox primarily promotes the formation of stable, transcription-ready dsDNA episomes from the viral genome [6, 29]. In this study, we investigated two strategies to enhance AAV2 transduction in cancer cells: pre-treatment with Dox (Dox^pre-AAV2) and direct conjugation of Dox to the AAV2 (AAV2-Dox). Both methods led to a marked enhancement of transgene expression while Dox was maintained at a minimally cytotoxic concentration. (Figures 1E and 2D). Significantly, AAV2-Dox achieved a significant positive effect at a much lower concentration (0.005 µg mL^{− 1} vs. 0.02 µg mL^{− 1}) compared to Dox alone. This level of enhancement compares favorably to other drug-adjuvant strategies for AAV [6, 28], but our conjugation approach offers the added benefit of ensuring co-localization of the drug and virus at the cellular level. Infection with AAV2-Dox caused cancer cells to produce PD-L1 shRNA. The resulting decrease in PD-L1 expression reverses immune suppression, enabling T-cell-mediated attack and inhibiting tumor growth (Fig. 2G).

The tumor microenvironment presents a dual target for virotherapy: its physicochemical properties both drive malignancy and obstruct viral efficacy [3]. This has led to a promising therapeutic approach: actively exploiting these tumor microenvironment characteristics to enhance viral delivery and efficacy [16, 17]. Indeed, strategies that manipulate, or exploit, the tumor microenvironment are now at the forefront of advancing cancer virotherapy. PEGylation provides nanoparticles with stealth properties by preventing their clearance [30]. The dense layer of hydrophilic PEG chains creates a steric barrier that inhibits viral infection (Fig. 3B). In our study, we engineered a pH-responsive AAV2-Dox by attaching a PEG stealth shield via a cleavable benzoic-imine linker. The PEGylated AAV2-Dox showed minimal activity at pH 7.4 but demonstrated robust, restored transduction and therapeutic efficacy upon exposure to acidic conditions (pH 6.7) (Fig. 3C). This acid-triggered unshielding provides a method for confining viral activity specifically to the acidic tumor microenvironment, a finding supported by our MD simulation (Fig. 3E and F). While other pH-responsive nanoparticles have been developed [16, 17], our use of a benzoic-imine linker directly on the viral capsid represents a new approach for virotherapy, offering a more direct control over viral infectivity compared to systems where the virus is merely encapsulated.

Despite advancements in pH-sensitive drug delivery, the clinical adoption of pH-responsive nanomedicines is hindered by biological complexity, technical manufacturing challenges, and safety/regulatory hurdles [31]. Key issues include inconsistent tumor acidity, the narrow pH difference between healthy and tumor tissue leading to leakage or immune clearance, difficulties in scaling complex syntheses while maintaining consistency, poor stability, and concerns about the long-term toxicity of synthetic polymers under regulatory scrutiny. Consistent with these limitations, the non-PEGylated AAV2-Dox group exhibited minimal therapeutic efficacy compared with its PEGylated counterpart (Fig. 4), underscoring the intrinsic challenges of achieving effective systemic AAV delivery [3, 12, 13]. While AAV2-Dox demonstrated potent PD-L1 silencing and T-cell mediated killing in vitro, its in vivo performance was severely compromised by natural physiological barriers. The primary reason for this discrepancy is hepatic sequestration [3]. The qPCR analysis revealed that the AAV2 and AAV2-Dox treatments resulted in a liver accumulation of genome copies that exceeded the tumor load by over 40-fold. This massive off-target distribution effectively depletes the circulating vector concentration, preventing a sufficient therapeutic dose from reaching the tumor microenvironment.

Our pH-gated VDC strategy enabled efficient PD-L1 gene silencing (Fig. 3D). This in vivo knockdown reversed immune suppression, leading to a robust anti-tumor immune response (Fig. 5A and B) and significant tumor inhibition (Fig. 4B and D). Looking forward, this versatile platform can be readily adapted for a wide range of synergistic therapies by simply exchanging the genetic payload. Furthermore, its application could be extended beyond cancer by incorporating targeting moieties or utilizing local delivery for other diseases. While this pH-gated activation was highly effective in our model, we acknowledge that the heterogeneity of the in vivo tumor microenvironment means that some tumor regions may not be sufficiently acidic to trigger complete unshielding. While the MCA205 model is a benchmark for PD-L1 immunotherapies [22], we acknowledge that a single ectopic model cannot fully capture the complexity of orthotopic microenvironments. Future studies will validate this platform in orthotopic and metastatic models to confirm its broad clinical applicability. Nevertheless, our approach represents a significant advance in achieving targeted viral delivery with enhanced specificity and safety.

Conclusions

Our pH-responsive PEGylated AAV2-Dox system, designed to deliver PD-L1 shRNA, pioneers a promising approach to surmount cancer virotherapy obstacles. By diminishing PD-L1 expression on cancer cells, AAV2-Dox invigorates T cell-mediated anti-tumor immunity, effectively reversing immune suppression and curbing cancer cell proliferation. A key advantage is its efficacy at lower concentrations compared to Dox alone, suggesting enhanced therapeutic benefits. This pH-gated VDC strategy provides a potentially more potent cancer treatment by tackling limitations in viral delivery and boosting overall therapeutic success. In conclusion, this study indicates a promising initial safety profile for the established a virus-drug conjugate, presenting a promising strategy for developing advanced nanomedicines for combinatorial cancer immunotherapy.

Materials and methods

Materials

Dulbecco’s Modified Eagle Medium (DMEM), Roswell Park Memorial Institute 1640 (RPMI 1640) medium, phosphate-buffered saline (PBS) powder, fetal bovine serum (FBS), penicillin/streptomycin, non-essential amino acids (NEAA) 100X, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), sodium pyruvate 100X, 0.25% Trypsin-EDTA, Novel Juice nucleic acid safe dye, and Kplus DNA Ladder RTU were obtained from GeneDireX (US). Maestrofectin in vitro transfection reagent and 50X TAE buffer were purchased from Oimcs Bio (Taiwan). (1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride) (EDC), N-hydroxysulfosuccinimide (Sulfo-NHS), and 2-(N-morpholino)ethanesulfonic acid (MES) buffered saline were purchased from Thermo Scientific Inc. (US). Branched polyethylenimine (bPEI 25K, Mw = 25,000), sodium bicarbonate, 4’,6-diamidino-2-phenylindole (DAPI) dihydrochloride, OptiPrep Density Gradient Medium, 10% neutral buffered formalin, agarose, goat serum, and bovine serum albumin were acquired from Sigma (US). Methoxy polyethene glycol (mPEG48, Mw = 2,145.58 or 2,145.6) with amine group (NH₂) or carboxylic acid group (COOH) was purchased from Biopharma PEG (US). Viral capsid formation was conducted using reagents from Cell Biolabs (US). CutSmart Buffer and ScaI-HF were procured from New England Biolabs (US). QuickTiter AAV Quantitation Kit and plasmid DNAs, including pHelper, pAAV-RC2, pAAV-EGFP, and pAAV-MCS, were obtained from Cell Biolabs (US). The CellTiter Aqueous One Solution Cell Proliferation Assay was sourced from Promega (US). PD-L1/CD274 antibody, PD-1/CD279 antibody were purchased from Proteintech (UK). CD8 antibody was purchased from GeneTex (USA). Goat Anti-Mouse IgG H&L Alexa Fluor 555 and Alexa Fluor^® 488 were purchased from Abcam (UK). CD45 and CD11b antibodies were purchased from BioLegend (US). CD4 and F4/80 antibodies were obtained from BD Biosciences (US). CD3 antibody was acquired from Thermo Fisher Scientific (US).

Culture of various cell lines and identification of cell membrane markers

MCA205 mouse fibrosarcoma cell line (SCC173, Millipore, US) were expanded in RPMI-1640 medium supplemented with 2 mM L-glutamine, 1 mM sodium pyruvate, 10% FBS, 1X non-essential amino acids, and 1X β-mercaptoethanol. A clone of murine CTLL-2 T cells (TIB-214, ATCC) were cultured in RPMI 1640 medium supplemented with 10% FBS, 100 U mL^− 1 penicillin, 100 µg mL^− 1 streptomycin, 1% sodium pyruvate, 1% non-essential amino acids (Invitrogen Life Science), and 10% recombinant human interleukin 2 (IL-2, PeproTech) with rat T-STIM with ConA (Corning, US). All cells were maintained at 37 °C in a humidified atmosphere with 5% CO₂.

To assess PD-L1 expression on the membrane of MCA205 cells, immunofluorescence staining was carried out. MCA205 cells were seeded at a density of 4 × 10⁵ cells per well in a 96-well plate and incubated at 37 °C with 5% CO₂ for at least 16 h. Following incubation, the cells were fixed with 4% paraformaldehyde (PFA) and stained using an PD-L1/CD274 monoclonal antibody to label PD-L1 maker.

To confirm the PD-1/PD-L1 interaction, immunofluorescence staining was performed for qualitative analysis of PD-1 and CD8 on CTLL-2 cells. CTLL-2 cells were seeded at a density of 4 × 10⁵ cells per well in a 96-well plate and cultured at 37 °C with 5% CO₂ for at least 16 h. The cells were then fixed with 4% PFA and immunostained using PD-1/CD279 monoclonal antibody or anti-CD8 antibody to target the respective PD-1 or CD8 markers.

Assessment of siRNA sequences for PD-L1 inhibition in MCA205 cells

To evaluate the effectiveness of various siRNAs (PD-L1 siRNA-1: 5’-GCCGACUACAAGCGAAUUA-3’; PD-L1 siRNA-2: 5’-CCAGCACACTGAGAATCAA-3’; PD-L1 siRNA-3: 5’-GCCACAGCGAATGATGTTT-3’; PD-L1 siRNA-4: 5’-CCUAUAUGUGGUAGAGUAU-3’) for targeted PD-L1 silencing using Maestrofectin transfection reagent/siRNA complexes, PD-L1 levels were subsequently quantified by Mouse PD-L1 DuoSet ELISA (R&D System). For the transfection procedure, cells were seeded into 96-well plates at a density of 1.5 × 10⁴ cells per well and incubated overnight. The next day, the medium was removed, and the cells were rinsed twice with culture medium (without FBS, pH 7.4). The transfection medium containing 0.4 µg siRNA per well, along with various samples, was added. Following a 48-hour incubation, the treated MCA205 cells were subjected to qualitative and quantitative analysis of PD-L1. Untreated cells served as the control group. Based on these observations, we concluded that the selected sequence of PD-L1 siRNA-3 is suitable for use in future experiments.

Production, purification, and titration of AAV2-PD-L1 shRNA

The plasmid pAAV-PD-L1 shRNA was constructed by first amplifying the PD-L1 fragment from siRNA using polymerase chain reaction (PCR). EcoRI and SalI restriction sites were introduced into the fragment using the following primer sequences: 5’-GGCGAATTCGCCACCATGGGTTCAGAGGGCGGCCCCGCCC-3’ and 5’-ACGCGTCGACTTAATCCTCGTCGCTACCGATGGCGCTGGT-3’. The resulting PCR product, a 0.71 kb PD-L1 siRNA cDNA, was then cloned into the EcoRI-SalI sites of the pAAV-MCS plasmid, generating pAAV-PD-L1 shRNA. The size of constructed pAAV-PD-L1 shRNA fragment is 6,097 base pairs (bp) (Fig. S3). Downstream of the PD-L1 shRNA sequence is an enhanced green fluorescent protein (EGFP) gene, which allows for the observation of EGFP expression in transfected cells.

The cell line used in this study was 293T (CRL-3216, ATCC) for production of AAV2. Cells were cultured in DMEM supplemented with 100 µg mL^{− 1} streptomycin, 10% FBS, and 100 U mL^{− 1} penicillin. The cells were maintained under standard conditions at 37 °C in a 5% CO₂ atmosphere. AAV2-PD-L1 shRNA was generated using the AAV-2 Helper Free Packaging System (Cell Biolabs) as described in previous studies [15, 16, 19]. The production process involved co-transfecting 293T cells with three plasmids (pHelper, pAAV-RC2, and pAAV-PD-L1 shRNA) using polyethyleneimine as the gene delivery reagent. For each 100-mm dish, 20 µg of pHelper, 10 µg of pAAV-RC2, and pAAV-PD-L1 shRNA were mixed with 40 µg of polyethyleneimine in serum-free medium, vortexed for 30–60 s, and incubated for 20–30 min. The transfection process lasted 30 min, after which the cells were harvested three days later. Purification and titration of AAV2-PD-L1 shRNA were performed using the ViraBind AAV Purification Kit and QuickTiter AAV Quantitation Kit (Cell Biolabs). The viral stock from each batch (eight 10-cm dishes) yielded 10¹¹ to 10¹² genome copies (GC) per milliliter (encapsulation efficiency of 0.85; total viral particles ~ 1.18 × 10¹¹ to 1.18 × 10¹²), and the purified virus was stored at -80 °C until use.

Synthesis of AAV2-Dox

We conjugated Dox to AAV2 surface thiol groups using the sulfo-SMCC crosslinker. First, Dox was activated by adding 0.25 mg (0.0005 mmol) of sulfo-SMCC to 1 mL samples of Dox solution at concentrations of 0.001, 0.005, 0.01, 0.02, or 0.05 µg mL^− 1 and incubating for 30 min at room temperature. Excess sulfo-SMCC was removed via dialysis (Biotech CE Dialysis Kits, 0.5–1.0 kDa MWCO) in PBS at 4 °C overnight. The dialysis buffer was exchanged at least twice, with each interval lasting 3–4 h. Subsequently, this maleimide-activated Dox (Dox-SMCC) was combined with 0.5 µL of thiol-containing AAV2 (1 × 10¹² GC mL^− 1). The mixture was gently agitated and incubated for 30 min at room temperature to facilitate conjugation between the maleimide on Dox-SMCC and sulfhydryl groups on AAV2. To quantify the conjugation efficiency, the Dox-SMCC, AAV2, or AAV2-Dox was analyzed by size exclusion chromatography on a Superdex 200 Increase 10/300 GL column (Cytiva) equilibrated with PBS. The conjugation efficiency of Dox-SMCC to AAV2 was calculated by analyzing the Dox-SMCC elution fractions. The resulting AAV2-Dox was purified using a 100 kDa molecular weight cutoff desalting column.

To mimic AAV2 nuclear uncoating, the AAV2-Dox conjugates were disassembled by heating at 70–75 °C for 10–20 min. This thermal treatment physically disrupted the AAV2 capsid, triggering the release of Dox. To quantify the release, samples were analyzed by size exclusion chromatography (SEC) on a Superdex 200 Increase 10/300 GL column equilibrated with PBS. The release efficiency of Dox was calculated by quantifying the free Dox in the elution fractions.

Cytotoxicity

To verify the cytotoxic effect of the prepared AAV2-Dox, we used the CellTiter Aqueous One Solution Cell Proliferation Assay to evaluate the impact of this viral treatment combined with Dox on cell viability. The assay operates on the principle that the tetrazolium compound in the reagent reacts with succinate dehydrogenase (SDH) in the mitochondria to produce a colored formazan product, which is only generated by viable cells. Since dead cells lack SDH activity, differences in total metabolic activity are measured by detecting absorbance at a specific wavelength. MCA205 cells were seeded in a 96-well plate at a density of 10⁴ cells per well and incubated at 37 °C in a 5% CO₂ atmosphere for at least 16 h. The MCA205 cells were treated with AAV2-PD-L1 shRNA alone, AAV2-PD-L1 shRNA combined with Dox pretreatment (Dox^pre-AAV2, Dox concentration: 0.001 µg mL^− 1, 0.005 µg mL^− 1, 0.01 µg mL^− 1, 0.02 µg mL^− 1, 0.05 µg mL^− 1), or AAV2-PD-L1 shRNA-Dox (Dox concentration: 0.001 µg mL^− 1, 0.005 µg mL^− 1, 0.01 µg mL^− 1, 0.02 µg mL^− 1, 0.05 µg mL^− 1) treatment, followed by continued incubation. In parallel, we also assessed the cytotoxicity of Dox and Dox-SMCC. Additionally, we assessed how Dox dosage affects cancer cells by comparing single-dose and triple-dose groups. We administered three daily doses to the triple-dose group of Dox^pre-AAV2 (Dox concentration: 0.02 µg mL^− 1) or AAV2-Dox (Dox concentration: 0.005 µg mL^− 1). After post-treatment, we cultured the single-dose group or the triple-dose group for an additional 72 h. On designated observation days, the culture medium in each well was replaced with 100 µL of fresh medium, followed by measurement using the CellTiter Aqueous One Solution Cell Proliferation Assay. Metabolic activity was measured as the optical density (OD) of formazan at 490 nm and taken as an indirect measure of cell viability. The MTS reduction in untreated cells was set as 100%, and the OD of treated cells was expressed as a percentage relative to the untreated control.

Measurement of EGFP or PD-L1 levels in MCA205 cells treated with AAV2, Dox^pre-AAV2, or AAV2-Dox

To investigate the effects of AAV2, Dox^pre-AAV2 (0.02 µg mL^− 1) pretreatment, AAV2-Dox (0.005 µg mL^− 1), or AAV2-PD-L1 shRNA alone on cancer cells, MCA205 cells were seeded in a 96-well culture plate at a density of 1.5 × 10⁴ cells and incubated at 37 °C with 5% CO₂ for at least 16 h before various treatments. AAV2 was administered at a dose of 8.5 × 10⁸ GC per well for each treatment. After 72 h incubation under various conditions, treated MCA205 cells were analyzed using the Mouse PD-L1 DuoSet ELISA kit (R&D Systems) to quantify PD-L1 protein levels. PD-L1 expression in untreated cells was defined as 100%, and the expression in treated cells was reported as a percentage of this control. To evaluate the impact of virus dosage on cancer cells, various conditions were established for single-dose and triple-dose (one dose per day for 3 consecutive days) treatments. The single-dose group was cultured for 72 h post-treatment, while the triple-dose group was cultured for 144 h. Afterwards, treated cells were performed to determine viral transduction efficacy and to quantitatively and qualitatively assess PD-L1 protein levels. Alternatively, to qualitatively assess changes in PD-L1 protein expression in MCA205 cells following various treatments, immunofluorescence staining was performed using a primary PD-L1/CD274 monoclonal antibody and a secondary goat anti-mouse IgG H&L antibody conjugated with Alexa Fluor 555.

Establishment of the red fluorescent protein-expressed MCA205 cell line

To evaluate the anti-tumor effects of drugs on cancer cells within a co-culture system, we aimed to generate a stable MCA205 cell line expressing red fluorescent protein (RFP). This was achieved by infecting MCA205 cells with an RFP Control Lentivirus Puro (abm, CA, V23N23T). The resulting MCA205-RFP cell line would enable real-time visualization of MCA205 cells dynamics during co-culture with CTLL-2 cells. MCA05 cells (2 × 10⁵) were seeded in a 3 cm culture dish and incubated for at least 16 h. Based on the manufacturer’s recommendations, we selected a multiplicity of infection of 10 for viral transduction. RFP expression was assessed using a fluorescence microscope 2–3 days post-infection. Once most cells exhibited fluorescence, the standard culture medium (supplemented with 10% FBS, 1% Penicillin/Streptomycin, 1% NEAA, and 25 mM HEPES) was replaced with culture medium containing 2 µg mL^− 1 puromycin for antibiotic selection. Since the lentivirus carries a puromycin resistance gene, only successfully infected cells survived and proliferated in the presence of puromycin. After five days of antibiotic selection, cells were harvested and fixed using 0.125% Trypsin-EDTA. Uninfected MCA205 cells were prepared as a control group. Flow cytometry analysis confirmed that all surviving MCA205 cells stably expressed RFP, validating their suitability for further experiments.

Coculture of CTLL-2 cells and MCA205-RFP cells treated with AAV2-PD-L1 shRNA-DOX

To investigate the impact of PD-L1 silencing on anti-tumor immune responses, murine CTLL-2 T cells were used as effector T cells and co-cultured with AAV2, Dox^pre-AAV2, or AAV2-Dox treated MCA205-RFP cells to block the PD-1/PD-L1 interaction. After the various treatments, CTLL-2 cells were seeded in each well of a 96-well plate at densities of 1.5 × 10⁴ cells, along with a fixed number of AAV2, Dox^pre-AAV2, or AAV2-Dox treated MCA205-RFP cells (1.5 × 10⁴ cells), followed by continued incubation. At various time points, the culture medium was refreshed and cancer cell viability was measured using the CellTiter Aqueous One Solution Assay. Viability was expressed relative to untreated controls (set at 100%). After treatment, cells were fixed with 4% PFA and imaged via confocal microscopy.

Synthesis and computational analysis of PEGylated AAV2-Dox

PEGylated AAV2-Dox was prepared via chemical conjugation. Reaction mixtures containing amine-functionalized mPEG48 (0.32 µg, 0.15 nmol) and AAV2 stock (0.5 µL, 1 × 10¹² GC mL^− 1) in PBS were gently stirred at 25 °C for 15 min to achieve homogeneity. Subsequently, terephthalaldehyde (0.05, 0.1, 0.2, 0.3, or 0.5 nmol) dissolved in PBS was added dropwise, and the reaction proceeded at 25 °C for another 30 min. PEGylated AAV2 was synthesized in parallel as a control to validate the in vivo VDC system. Alternatively, non-pH-sensitive PEGylated AAV2-Dox was synthesized via chemical conjugation (Fig. S8). Briefly, mPEG48-carboxylic acid (0.32 µg, 0.15 nmol) was reacted with varying amounts of EDC (0.05, 0.10, 0.15, 0.50, or 1.00 nmol) in MES-buffered saline. Sulfo-NHS was added gently under stirring at 25 °C for 15 min to form a homogeneous solution of amine-reactive PEG-sulfo-NHS ester. AAV2 stock (0.5 µL, 1 × 10¹² GC mL^− 1) in PBS was then added dropwise to the mixture and allowed to react at 25 °C for 120 min. The resulting conjugate was purified using a size-exclusion desalting column (100 kDa MWCO) equilibrated with PBS.

For transmission electron microscopy (TEM), AAV2 alone or PEGylated AAV2-Dox solutions (at pH 7.4 or pH 6.7) were first deposited on a Formvar-carbon-coated 200 mesh copper grid. Imaging was then performed with a JEOL JEM-1400 microscope at an accelerating voltage of 100 kV. The hydrodynamic diameter and zeta potential of PEGylated AAV2-Dox (pH 7.4 or pH 6.7) or AAV2 (pH 7.4 or pH 6.7) were determined at 25 °C using a Malvern Zetasizer Nano ZS. Stability was evaluated by measuring the hydrodynamic diameter after 6 h of incubation at 37 °C in either PBS (pH 7.4) or culture medium (pH 7.4). Furthermore, we evaluated the hydrodynamic diameter of PEGylated AAV2-Dox by incubating the samples in PBS (pH 7.4) and culture medium (pH 7.4) at 37 °C for 12, 24, 48, or 72 h. Molecular dynamics (MD) simulations were performed using the NAMD program, with parameters adapted from the CHARMM27 force field [15, 16], to investigate the self-assembly of PEGylated AAV2-Dox over time in different pH environments (pH 7.4 or pH 6.7).

Dot blot and neutralization assay

To ensure uniform AAV2-Dox or PEGylated AAV2-Dox with varying molar ratios of PEG to terephthalaldehyde (3, 1.5, 0.75, 0.5, and 0.3), samples were supplemented with 0.05% SDS prior to spotting. Polyvinylidene difluoride (PVDF) membranes were activated with methanol, rinsed with water, and equilibrated in Tris buffered saline with Tween 20 (TBST). To prevent uneven drying or diffusion during manual spotting, the membrane was maintained in a semi-hydrated state by placing it on TBST-saturated filter paper. Samples were applied sequentially to the membrane. After air-drying at room temperature for 10–30 min, the membranes were blocked with 5% non-fat milk in TBST for 60 min at room temperature and subsequently incubated with anti-VP1 antibody overnight at 4 °C. Following three washes with TBST, membranes were incubated with the appropriate HRP-conjugated secondary antibody for at least 60 min. Immunoreactive dots were visualized using enhanced chemiluminescence (ECL) reagents.

MCA205 cells were seeded in 24-well plates at a density of 1 × 10⁵ cells well^− 1 and cultured overnight. The following day, AAV2 or PEGylated AAV2-Dox (5 × 10⁸ GC) were incubated with an anti-VP1 antibody (100 µg mL^− 1) in a total volume 10 µL PBS for 30 min at room temperature. To evaluate sequential neutralization, this incubation process was repeated for a second and third round, respectively, prior to transduction. The AAV2-VP1 antibody or PEGylated AAV2-Dox-VP1 antibody mixtures was then diluted in 0.5mL of culture medium containing 10% FBS, 100 U mL^− 1 penicillin, and 100 µg mL^− 1 streptomycin adjusted to either pH 7.4 or pH 6.7 and added to the seeded cells. Transduction was allowed to proceed for 3 days. Post-transduction, EGFP reporter expression was quantified using flow cytometry (Beckman Coulter, Fullerton, CA, USA).

Transduction efficiency of PEGylated AAV2-Dox

All viral infection and transduction experiments were conducted using a culture medium supplemented with 10% FBS, 100 U mL^− 1 penicillin, and 100 µg mL^− 1 streptomycin. To evaluate the transduction efficiency of PEGylated AAV2-Dox at different PEG/terephthalaldehyde molar ratios, non-pH-sensitive PEGylated AAV2-Dox at different PEG/EDC molar ratios, AAV2-Dox, or AAV2 alone, EGFP expression was used as the reporter. MCA205 cells were seeded at 1 × 10⁵ cells well^− 1 in 24-well plates and infected the following day. Test samples were added to the cells in culture medium with 10% FBS, and transduction was allowed to proceed for 3 days. Post-transduction, EGFP expression was quantified by flow cytometry (Beckman Coulter, Fullerton, CA, USA). For analysis, cells were detached using 0.025% trypsin, transferred to microtubes, and fixed with 4% paraformaldehyde. Cells were gated based on forward and side scatter, and 10,000 events were collected per sample. Untreated cells served as a negative control.

Cytotoxicity of PEGylated AAV2-Dox

MCA205 cells (7 × 10⁴ cells well^− 1) were seeded in 24-well plates and incubated in culture medium for 12 h. Subsequently, cells were exposed to PEGylated AAV2-Dox (at various molar ratios of PEG/terephthalaldehyde) or non-pH-sensitive PEGylated AAV2-Dox (at various molar ratios of PEG/EDC) at pH 7.4 for 24 h. After this incubation, the treatment media containing the test samples were removed. Cell viability was assessed using the CellTiter Aqueous One Solution Cell Proliferation Assay. The viability of untreated cells was set as 100%, and the viability of treated cells was expressed as a percentage relative to this control.

In vitro viral transduction and analysis

The pH-responsive release of AAV2 from PEGylated AAV2-Dox was evaluated by monitoring EGFP expression following AAV2 infection. MCA205 cells were seeded (1 × 10⁵ cells well^− 1) into 24-well plates. On the subsequent day, cells underwent three PBS washes. Viral transduction was then initiated by incubating the cells for 6 days with various treatments (AAV2: 5 × 10⁸ GC well^− 1) in media whose pH was adjusted to either 7.4 or 6.7 using 1.0 M lactate. Post-transduction, EGFP expression was quantified using flow cytometry (Beckman Coulter, Fullerton, CA, USA), with a minimum of 10,000 events acquired for each analysis. Cells not subjected to viral treatment were used as a negative control. Alternatively, to assess changes in PD-L1 protein expression, PD-L1 levels in treated MCA205 cells were quantified using the Mouse PD-L1 DuoSet ELISA kit. PD-L1 expression in treated cells was expressed as a percentage relative to untreated control cells, whose expression was set to 100%.

Mice study

The Institutional Animal Care and Use Committee of the College of Medicine, National Taiwan University, approved all animal procedures (IACUC-20160366), which were performed in accordance with the Guide for the Care and Use of Laboratory Animals. Six-week-old male C57BL/6 mice were procured from the National Laboratory Animal Center in Taiwan. Animals were housed under a 12-hour light/dark cycle in controlled conditions, with up to five mice per cage and free access to standard chow and water. To evaluate the potential therapeutic effects of AAV2, Dox^pre-AAV2, AAV2-Dox, PEGylated AAV2, and PEGylated AAV2-Dox, subcutaneous MCA205 cells (2 × 10⁶ cells)-derived tumors were injected subcutaneously into C57BL/6 mice to establish the xenograft tumors. Once tumors reached approximately 200 mm³, mice were assigned to one of four treatment groups: PBS (control), Dox^pre-AAV2, AAV2, AAV2-Dox, PEGylated AAV2, or PEGylated AAV2-Dox (AAV2: 5 × 10⁹ GC mouse^− 1). Each mouse then received an injection of 100 µL of its assigned treatment in PBS. Tumor growth was subsequently monitored at least three times per week. Mice in the control group received PBS injections. Tumor volumes were measured using calipers and calculated with the formula: Volume = 0.5 × L² × W (where L is tumor length and W is tumor width).

In vivo analysis

To assess the impact of different treatments, C57BL/6 mice received either PBS, AAV2, Dox^pre-AAV2, AAV2-Dox, PEGylated AAV2, or PEGylated AAV2-Dox on Day 0. Blood samples were subsequently collected from the orbital sinus of these mice on Day 14 to obtain serum for analysis. To evaluate long-term toxicity, additional blood samples were obtained from the PBS or PEGylated AAV2-Dox group on Days 21 and 28. Several biochemical markers were evaluated in the collected serum. These included glutamic-oxaloacetic transaminase (GOT), glutamic-pyruvic transaminase (GPT), total bilirubin (TBIL), creatinine (CRE), and blood urea nitrogen (BUN). The levels of GOT and GPT were determined using an enzymatic method that measures the relative activity of aspartate aminotransferase (AST) and alanine aminotransferase (ALT), respectively. TBIL levels, which serve as an indicator of liver cell damage such as that seen in hepatoma and hepatitis, were quantified using Randox Diagnostic Test kits following the manufacturer’s provided instructions. CRE and BUN levels, an indicator of kidney function, were also measured using commercial kits (BioAssay Systems, Hayward, CA).

Tumor or organ tissue was harvested from the treated mice. Total RNA was isolated using TRIzol reagent (Thermo Fisher Scientific) and reverse-transcribed into cDNA using the GoScript Reverse Transcription Mix (Promega). Quantitative polymerase chain reaction (qPCR) was conducted with GoTaq qPCR Master Mix (Promega) on a QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific). To assess the biodistribution of AAV2, specific primers targeting the EGFP transgene (forward: 5’-CATGGTCCTGCTGGAGTTCGTG-3’; reverse: 5’-CGTCGCCGTCCAGCTCGACCAG-3’) were used for this quantification. Alternatively, relative PD-L1 expression in treated tumor cells was determined and normalized to GAPDH as an internal housekeeping control. Primer sequences included: PD-L1 (forward: 5’-GAGTGCAGATTCCCTGTAGAAC-3’; reverse: 5’-CTCTCCTGCCACAAACTGAA-3’) and GAPDH (forward: 5’-TCTTGGGCTACACTGAGGAC-3’; reverse: 5’-CATACCAGGAAATGAGCTTGA-3’). For the analysis of tumor-infiltrating lymphocytes (tumor-infiltrating leukocytes stained by CD45⁺, cytotoxic T lymphocytes stained by CD8⁺ and CD3⁺, helper T lymphocytes stained by CD4⁺ and CD3⁺, and macrophages stained by F4/80⁺ and CD11b⁺), tumor tissues were dissociated using a gentleMACS Dissociator (Miltenyi Biotec, Germany) and isolated via density gradient centrifugation. Single-cell suspensions were stained with antibodies against surface markers prior to fixation and permeabilization. Data acquisition was performed using the BD Accuri C6 flow cytometer (BD Biosciences).

Cytokine concentrations were determined using commercial ELISA kits targeting Granzyme B and IFN-γ (Abcam). The procedures were carried out following the standard instructions provided by the manufacturer.

Harvested xenograft tumors or different organs underwent formalin fixation, paraffin embedding (FFPE), and sectioning for comprehensive analysis, including standard histopathological imaging. For immune cell analysis, FFPE sections were subjected to multiplex immunohistochemistry (mIHC) using the Polaris system (Akoya Biosciences) with the Opal 7-Color and anti-Rabbit Manual IHC Kits (NEL810001KT and NEL840001KT, respectively). Specific immune cell markers were detected using the following rabbit primary antibodies and assigned Opal dyes: anti-CD4 (Abcam, Opal620), anti-CD8 (Bioss, Opal570), anti-MHC II (Bioss, Opal650), anti-CD19 (HistoSure, Opal520), anti-CD206 (Bioss, Opal520), and anti-F4/80 (HistoSure, Opal690). Multispectral images were acquired with Phenochart and analyzed using inForm software (Akoya Biosciences). The inForm software’s machine-learning algorithm, trained on ten manually annotated representative tumor images, identified and quantified cell types based on marker expression: B cells (CD19⁺), CD4⁺ T cells (CD4⁺), CD8⁺ T cells (CD8⁺), M1 macrophages (F4/80⁺MHC II⁺CD206⁻), and M2 macrophages (F4/80⁺MHC II⁻CD206⁺). DAPI nuclear staining was used for cell segmentation.

Statistical analysis

All experiments were performed in triplicate or sextuplicate, and the results are presented as mean ± standard deviation (s.d.) or standard error of the mean (s.e.m.). Differences between the control and experimental groups were evaluated using a two-tailed Student’s t-test. Statistical significance was defined as p < 0.05 (*).

Author contributions

All authors made substantial contributions to the conception, design, data acquisition, analysis, and interpretation of the work. S.J.T, I.K, P.C.Y., and Z.X.L. prepared Figs. 1, 2, 3, 4 and 5. Y.C.H prepared figure 3 F. All authors contributed to drafting and critically revising the manuscript, approved the final version for publication, and agree to be accountable for all aspects of the work.

Funding

This work was supported by the National Science and Technology Council (111-2628-E-110-007-MY3, 112-2221-E-006-196, 112-2628-B-110-004-MY3, 113-2622-B-006-009, 114-2223-E-110-003-MY3, and 114-2314-B-006-109-MY3).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.