Vinyl ether acrylate (VEA) and VEA-functionalized PVA (PVA-g-VEA) were prepared based on the previous reports (Chen et al., 2017). For the synthesis of DMA-functionalized PVA-g-VEA (PVA-g-VEA-DMA), DMA and Et₃N (20 μl) were added to a 10-ml solution of PVA-g-VEA (0.413 g, 0.024 mmol) in methanol under stirring, and the reaction was continued for 12 h. Afterwards, the above mixture solution was condensed by rotary evaporation. Finally, PVA-g-VEA-DMA was purified using the precipitant of ice-cold diethyl ether and then dried using the vacuum drier.

To synthesize TPP-modified PVA-g-VEA-DMA, a mixture of TPP (70.5 mg, 0.16 mmol), DCC (49.1 mg, 0.24 mmol), and DMAP (20.3 mg, 0.17 mmol) was dissolved in dimethyl sulfoxide (DMSO) (10 ml), and the obtained solution was vigorously stirred for 1 h for activation of the carboxyl group in TPP. Subsequently, a solution of PVA-g-VEA-DMA (100 mg) in DMSO (5 ml) was slowly added for a 24-h reaction under stirring. After that, PVA-g-VEA-DMA/TPP was purified via extensive dialysis [molecular weight cut off (MWCO): 3.5 kDa] against methanol for 24 h, concentrated through rotary evaporation, precipitated in ice-cold diethyl ether, as well as vacuum dried.

Acid-responsive nanogels were prepared using the UV-crosslinking method. Briefly, the nanogels were obtained by dissolving the polymer of PVA-g-VEA-DMA/TPP in water (1 mg/ml) with the addition of a photoinitiator (I2959, 5 wt% of polymer), adjusting solution pH to 6.8, and stirring under UV exposure for 10 min. To prepare drug-loaded nanogels, a solution of PTX and LND in DMSO (10 mg/ml) was added in the aqueous polymer solution with a feeding ratio of 10%. Through adjusting the solution pH to 6.8 and in situ crosslinking of acrylate in the polymers under UV light for 10 min, PTX and LND co-loaded nanogels were obtained. A Millipore ultrafiltration centrifugal tube with a MWCO of 10,000 was used to remove the free drugs.

pH-induced degradation of nanogels was studied by detecting the nanogel size change via dynamic light scattering (DLS) in different conditions (pH 7.4 and 5.0) for different periods (0, 6, and 24 h). The unleash behavior of LND and PTX-loaded nanogels was explored in different conditions (pH 7.4 and 5.0) through the dialysis procedure. In brief, a dialysis bag (MWCO: 3,500) containing 1 ml of PTX and LND co-loaded nanogel suspension was placed into a large-volume tube with 20 ml of the corresponding medium under a continuous shaking at 100 rpm at 37°C. At various time intervals (0, 2, 4, 6, 8, 10, 24, 52, 56, and 72 h), 5 ml of the medium in the tube of each group was extracted and supplemented by the newly prepared medium. The collected media of each group were concentrated and subjected to HPLC for quantification of the released PTX and LND.

MCF-7/MDR cells (1 × 10⁵ cell/well) were seeded into a 24-well plate containing a rounded coverslip in each well. After 24 h, fluorescein isothiocyanate (FITC)-labeled NGs, D-NGs, and T-D-NGs were added and incubated with the cells for 8 h. Then, the medium was removed, and the cells were incubated with 500 μl of serum-free media containing 200 nM MitoTracker® Red CM-H2XRos probes at 37°C for 20 min or 500 μl of serum-free media containing 50 nM LysoTracker® Red DND-99 probes at 37°C for 30 min. After washing thrice in phosphate-buffered saline (PBS) and fixation of the cells using 4% paraformaldehyde for 20 min, 4′,6-diamidino-2-phenylindole (DAPI) was added for 10-min staining of the cell nucleus, and the excess DAPI was removed by washing thrice in PBS. Finally, the fluorescence images of each group were obtained by confocal laser microscope (CLSM, LSM700, Zeiss, Germany) and processed using the ZEN imaging software.

To measure the mitochondrial depolarization, the cationic fluorochrome JC-1 was employed to detect mitochondrial membrane potential change. Briefly, MCF-7/MDR cells were seeded in a 12-well plate (1 × 10⁵ cells/well) and incubated for 24 h. Then, free LND, NGs@LND, D-NGs@LND, and T-D-NGs@LND at a LND dosage of 50 μg/ml were severally added and incubated with the cells for 12 h. Thereafter, the cells were washed with PBS followed by incubation with 0.0125 M JC-1 at 37°C for 20 min. After washing thrice in PBS, the cells were observed using a fluorescence microscope (IX73, Olympus, Japan).

A ROS assay kit (Beyotime, China) was utilized for the detection of the intracellular ROS level. Briefly, MCF-7/MDR cells were cultured in a 12-well plate (1 × 10⁵ cells/well) and incubated for 24 h. Then, free LND, NGs@LND, D-NGs@LND, and T-D-NGs@LND at a LND dosage of 50 μg/ml were respectively added and incubated with the cells for 12 h. Afterwards, the cells were washed with PBS for two times and stained with dichlorodihydro-fluorescein diacetate (DCFH-DA) for 20 min. After washing thrice in PBS and fixation of the cells using 4% paraformaldehyde for 20 min, DAPI was added for a 10-min staining of the cell nucleus, and the excess DAPI was removed by washing thrice in PBS. Finally, the cells were observed using a fluorescence microscope (IX73, Olympus, Japan).

An ATP assay kit (Beyotime, China) was utilized for the detection of the intracellular ATP levels. MCF-7/MDR cells were cultured in a six-well plate (1 × 10⁵ cell/well) for 24 h. After that, free LND, NGs@LND, D-NGs@LND, and T-D-NGs@LND at a LND dosage of 50 μg/ml were respectively added and incubated with the cells for 12 h. The intracellular ATP level was measured according to the manufacturer’s protocol.

To evaluate the apoptosis-inducing effect of different LND formulations, MCF-7/MDR cells were seeded into 12-well plates (1 × 10⁵ cell/well) and cultured for 24 h at 37°C. Then, different LND formulations (50 μg/ml LND) were added and incubated with the cells for 12 h. After that, the cells were digested and collected for Annexin V-FITC/propidium iodide (PI) staining based on the manufacturer’s instructions and analyzed by flow cytometry.

MTT assay was utilized to study the cytotoxicity of MCF-7/MDR cells receiving various formulations. The cells were cultured in 96-well plates for 24 h and then incubated with various formulations or free drug. After 48-h incubation, MTT solution was added for 4-h incubation. Then, the culture medium was replaced by DMSO for detection under a microplate reader with the absorbance at 490 nm.

Data were expressed as mean ± standard deviation. Differences between groups were assessed by a two-tailed unpaired Student’s t-test or one-way ANOVA with Tukey’s post hoc test. The level of significance was set at probabilities of *p < 0.05, **p < 0.01, and ***p < 0.001.

We started with the synthesis of PVA-g-VEA/VEA-DMA/TPP polymer, in which the pendent VEA, positively charged DMA, and mitochondria-recognizable TPP were successively conjugated onto the water-soluble PVA backbone (Scheme 1). The VEA was firstly grafted onto PVA polymers via an acetalization reaction between vinyl ether and hydroxyl group in the presence of PTSA. The signals at δ 6.6 and δ 5.85–6.45 attributable to the vinyl protons of VEA and newly formed acetal methine proton, respectively, revealed the successful conjugation of VEA to the polymers, and the VEA grafting ratio was determined to be about 6.1% by comparing the integral of the peaks at δ 1.3 and at δ 6.6 (Figure 1A). DMA was further conjugated with acrylate in PVA-VEA, yielding a DMA functionality of about 2.66% with 3.44% acrylate left for UV-mediated crosslinking according to the ¹H NMR spectra (Figure 1B). Lastly, the obtained polymeric conjugates were coupled with TPP on the backbone using DCC and DMAP via esterification, and the content of the TPP moiety on PVA-VEA-DMA-TPP was about 1.3 wt% as revealed from the integrals of the peaks at δ 7.68 of benzene protons and at δ 1.3 (Figure 1C).

T-D-NGs were then obtained by dissolving the polymer of PVA-g-VEA/VEA-DMA/TPP in water, adjusting pH to 7.8, crosslinking acrylate in the presence of I2959 under UV light. D-NGs and NGs were prepared in a similar way. As displayed in Figure 2A–C, T-D-NGs, D-NGs, and NGs presented a typical spherical shape with uniform sizes of about 170 nm. pH-responsive T-D-NGs were explored by detecting the size change of T-D-NGs in acetate buffer (pH 5.0) for different times via DLS measurement. T-D-NGs displayed a swelled structure with a size of about several hundred nanometers after 6 h and later a small-sized particle with a size of about 17 nm after 24 h, whereas the size of T-D-NGs maintained almost the same at pH 7.4 after 24-h incubation (Figure 2D). These results revealed that T-D-NGs possessed a good colloidal stability in the physiological condition while suffered from the destruction at pH 5.0 that mimicked intracellular acidic condition. It should be noted that the swelled structure of T-D-NGs in response to acidic condition in their initial stage was ascribed to their partial cleavage of the acetal linkers in the network of T-D-NGs, which ultimately led to complete disassociation and small-sized conjugates in the acidic condition.

LND, which is a small-molecule hexokinase inhibitor, and chemotherapeutic PTX were co-loaded into the T-D-NGs via hydrophobic and/or electrostatic interactions. LND and PTX-loaded nanogels showed increased sizes in comparison to the blank nanogels as determined by DLS (Table 1). The introduction of DMA and TPP moieties in the nanogels dramatically improved drug loading content (DLC) from 2 wt% to about 5 wt% at a theoretical DLC of 10 wt%, which was probably due to their increased hydrophobicity and electropositivity.

The in vitro LND and PTX release profiles from T-D-NGs were studied at various pH, i.e., pH 5.0 and pH 7.4. As displayed in Figure 2E, F, a small amount of LND (31.7%) and PTX (34.5%) was leaked from T-D-NGs at pH 7.4 after 72 h, respectively, whereas more than 70% of LND and PTX were released at pH 5.0 after 72 h, further confirming pH-triggered disassociation of the nanogels and the subsequent payload release.

The cellular behaviors of T-D-NGs were investigated by using CLSM in MCF-7/MDR cells. T-D-NGs, D-NGs, and NGs were labeled with FITC for CLSM imaging, and the results showed that FITC-labeled NGs (FITC-NGs, green signal) were captured by lysosomes (red signal) with colocalized orange signal within cells for 8-h incubation, while most of the green signal of FITC-labeled T-D-NGs (FITC-T-D-NGs) and FITC-labeled D-NGs (FITC-D-NGs) were separated from the red signal of lysosomes after 8 h, indicating the successful escape of FITC-T-D-NGs and FITC-D-NGs from lysosomes with the aid of DMA module in the nanogels (Figure 3). In the following, we evaluated the mitochondrial-targeting ability of FITC-T-D-NGs, FITC-D-NGs, and FITC-NGs by staining the mitochondria with MitoTracker (red). After 12 h, a large area of FITC-T-D-NGs were overlapped with the mitochondria, resulting in intense orange fluorescence in the merged image (Figure 4). In contrast, FITC-D-NGs presented separated green and red fluorescence, and hardly any green signal was observed in the FITC-NG group with only red fluorescence in the merged image. All the above results demonstrated that FITC-T-D-NGs could efficiently escape from the lysosomes and target the mitochondria for facilitating LND delivery, while FITC-NGs and FITC-D-NGs either got trapped in the lysosomes or lacked mitochondrial orientation.

To explore the underlying mechanism of LND-loaded T-D-NGs (T-D-NGs@LND) in interfering mitochondrial functions, the measurements of MMP, ROS production, and ATP secretion in MCF-7/MDR cells were performed. JC-1 staining was firstly employed, which served as a fluorescence probe to detect MMP. JC-1 forms aggregates with red fluorescence at a high MMP, while remains monomers with green fluorescence at a low potential level, which is indicative of early apoptosis. After 12-h incubation, free LND displayed minimal green signal and intense red signal level comparable to the PBS group in MCF-7/MDR cells due to its slow diffusion and poor mitochondrial targeting (Figure 5A; Supplementary Figure S1). NGs@LND slightly increased the green fluorescence to some extent and maintained the strong red fluorescence, revealing an insufficient dissipation of MMP because of being trapped in lysosomes and limited accumulation in the mitochondria as demonstrated above. D-NGs@LND ameliorated this phenomena because of its lysosomal escape aided by DMA moieties in the nanogels. T-D-NGs@LND was the most effective in dissipating MMP with the strongest green fluorescence concomitant with the weakest red fluorescence, which was ascribed to its lysosomal escape, mitochondrial accumulation, and effective LND release triggered by acid within cells.

ROS production causes oxidative cell damage at a high level, which was another mechanism involved in cancer cell apoptosis via the mitochondrial pathway. ROS production level was detected by a fluorescence probe of DCFH-DA via CLSM imaging. The trend of the intracellular ROS production level was similar with the results of the dissipation of MMP (Figure 6; Supplementary Figure S2), verifying that T-D-NGs@LND induced a large amount of intracellular ROS production in the mitochondria, potentially for anti-tumor applications.

Cumulative evidences demonstrated that LND had the potential to inhibit hexokinase enzyme that was implicated in ATP generation and mitochondrial respiration. ATP secretion from the MCF-7/MDR cells receiving various LND formulations was estimated through the ATP assay kit. As expected, T-D-NGs@LND significantly decreased ATP secretion from the cells with a decline of about 60.9% versus 43.48%, 30.44%, and 15.95% for D-NGs@LND, NGs@LND, and free LND severally after 12-h incubation (Supplementary Figure S3). Such a pronounced inhibition of ATP secretion in MCF-7/MDR cell induced by T-D-NGs@LND further confirmed the precise and efficient delivery of LND into the mitochondria and also laid a solid foundation for MDR reversal by suppressing P-gp that was dependent on ATP.

The cell apoptosis with the treatments of various LND formulations via the mitochondrial pathway was explored via Annexin V-FITC/PI staining through flow cytometry. LND could induce cell apoptosis via several mechanisms like destruction of MMP and oxidative stress of overproduced ROS in the mitochondria as noted above. Total apoptotic rates of about 47.8% was found in T-D-NGs@LND-treated cells, which was higher than the D-NGs@LND (44%), NGs@LND (35.5%), and free LND (35.8%) groups (Figure 5B). Moreover, T-D-NGs@LND caused both higher early (41.4%) and late (6.4%) apoptosis than other treatments. MTT assay was also carried out and showed that all of the LND formulations led to a dose-dependent cytotoxicity toward MCF-7/MDR cells, among which T-D-NGs@LND displayed the most potent anti-tumor activity with a cell viability of about 33.47% at a LND concentration of 100 μg/ml for 48 h (Figure 7A). Taken together, T-D-NGs@LND was demonstrated to exert direct tumor cell-killing ability of LND by overcoming lysosomal trap, navigating into the mitochondria and efficiently releasing LND.

Furthermore, the potential role of LND for drug resistance reversal and chemo-sensitization was utilized to reactivate the treatment ability of PTX in MCF-7/MDR cells. Thus, the combination of LND and PTX in T-D-NGs could bring about a synergistic anti-tumor activity by fully exerting tumor cell killing of both LND and PTX. To verify the above assumption, MTT assay was performed in MCF-7/MDR cells receiving PTX- and/or LND-loaded nanogels for 48 h. As displayed in Figure 7B, all the PTX formulations and free PTX revealed mild cytotoxicity toward the cells with cell viabilities more than 40% even at a high PTX concentration of 20 μg/ml, reflecting drug resistance of the cells and inactivation of PTX formulations. The combination of LND and PTX in various formulations showed cytotoxicity toward the cells in a PTX dose-dependent fashion at a fixed LND concentration of 50 μg/ml (Figure 7C). As predicted, T-D-NGs@LND and PTX remarkably boosted the tumor cell-killing activity with a cell viability of about 18.59% at LND and PTX concentrations of 50 and 20 μg/ml for 48 h, which was much lower than T-D-NG-enabled monotherapy (T-D-NGs@LND and T-D-NGs@PTX), revealing the reversal of drug resistance and the synergistic therapeutic efficacy of unleashed LND and PTX. The half maximal inhibitory PTX concentration (IC₅₀) of T-D-NGs@LND and PTX (LND: 50 μg/ml) was also calculated, which was determined to be 7.44 μg/ml. T-D-NGs@LND and PTX also displayed a much higher anti-tumor activity as compared with D-NGs@LND and PTX (IC₅₀: 10.70 μg/ml), NGs@LND and PTX (IC₅₀: 11.96 μg/ml), and LND and PTX (IC₅₀: 11.49 μg/ml), further verifying T-D-NG-mediated lysosomal escape, mitochondrial accumulation, and acid-triggered payload release for fully exerting the efficacy of both drugs. In the meanwhile, the cytotoxicities of drug-free nanogels toward the cells were also assessed, and the results showed cell viabilities of 80%–100% at nanogel concentrations ranging from 0.15 to 2 mg/ml (Figure 7D), confirming the superior biocompatibility of the nanogels, which was mainly made up of FDA-approved biocompatible PVA backbone.