Epigenetic-based combination therapy and liposomal codelivery overcomes osimertinib-resistant NSCLC via repolarizing tumor-associated macrophages

Abstract

Osimertinib (Osi) is widely used as a first-line treatment for non-small cell lung cancer (NSCLC) with EGFR mutations. However, the majority of patients treated with Osi eventually relapse within a year. The mechanisms of Osi resistance remain largely unexplored, and efficient strategies to reverse the resistance are urgently needed. Here, we developed a lactoferrin-modified liposomal codelivery system for the combination therapy of Osi and panobinostat (Pan), an epigenetic regulator of histone acetylation. We demonstrated that the codelivery liposomes could efficiently repolarize tumor-associated macrophages (TAM) from the M2 to M1 phenotype and reverse the epithelial-mesenchymal transition (EMT)-associated drug resistance in the tumor cells, as well as suppress glycolysis, lactic acid production, and angiogenesis. Our results suggested that the combination therapy of Osi and Pan mediated by liposomal codelivery is a promising strategy for overcoming Osi resistance in NSCLC.

Introduction

Lung cancer is a leading cause of cancer-related death worldwide, with non-small cell lung cancer (NSCLC) accounting for about 85% of cases [1]. Epidermal growth factor receptor (EGFR) mutations are the main oncogenic drivers in NSCLC, and EGFR tyrosine kinase inhibitors (EGFR-TKI) have revolutionized the treatment of advanced NSCLC patients with EGFR mutations. However, nearly 50%–60% of patients treated with the first-generation EGFR-TKI (gefitinib) develop acquired resistance owing to the EGFR^T790M secondary mutation [2, 3]. Osimertinib (Osi), a third-generation EGFR-TKI, was designed to overcome the EGFR^T790M-associated drug resistance [4, 5]. Osi exhibits a superior efficacy in the previously untreated advanced NSCLC with an EGFR mutation with a median overall survival of 38.6 months [6]. Despite the significantly improved treatment outcomes, the acquired resistance to Osi is inevitably developed [7, 8]. The mechanisms underlying this resistance are complex and have not been fully revealed.

TAM plays an important role in the development of TKI resistance [9], which is highly plastic and can be reversibly polarized into the M1 and M2 phenotypes. TAM is often polarized towards the M2 phenotype, which contributes to tumor progression. M2 TAM secretes abundant TGF-β into the tumor microenvironment (TME) and promotes epithelial-mesenchymal transition (EMT) [10]. EMT is associated with drug resistance in cancer cells [11]. Specifically, EMT mediates EGFR-TKI resistance, and thus the biomarkers of EMT may serve as a prognostic factor in advanced stages of NSCLC [12].

Furthermore, TAM-derived TGF-β and cancer cell-derived vascular endothelial growth factor (VEGF) can induce angiogenesis [13, 14]. Notably, the activated VEGF/VEGFR pathway also mediates EGFR-TKI resistance [15]. Moreover, VEGF can also elicit EMT [16], and angiogenic VEGF-VEGFR2 signaling can increase ERK1/2 phosphorylation and promote EMT process [13]. It indicates that downregulation of VEGF and anti-angiogenesis could be a potential treatment strategy for treating Osi-resistant NSCLC.

Epigenetic regulation is closely related to TAM polarization [17]. In particular, histone deacetylase (HDAC) is heavily involved in M2 macrophage differentiation, and panobinostat (Pan) was reported to potently inhibit M2 polarization by suppressing MEK/ERK signaling and PPARγ expression as well as the consequent retinoic acid signaling [18]. In addition, epigenetic regulation can affect tumor metabolism. Histone deacetylation is conducive to the transcriptional inhibition of gluconeogenesis, thus promoting the activation of glycolysis in cancer cells, also known as Warburg effect [19]. The increased glycolysis and lactic acid production can also contribute to TKI resistance [20]. Our previous study also revealed that Pan can block M2 polarization and regulate aerobic glycolysis in cancer cells [21].

Pan is a pan-HDACi approved by the US Food and Drug Administration (FDA) in 2015 to treat patients with multiple myeloma. Clinical studies have revealed the safety and efficacy of Pan in combination with EGFR-TKIs, such as erlotinib and gefitinib [22, 23]. Here we propose a combination therapy of Pan and Osi using a targeting codelivery liposomal system. Lactoferrin (Lf) is an iron-carrying protein and specifically binds to low-density lipoprotein receptor-related protein 1 (LRP-1), which is overexpressed on both tumor cells and TAM [24]. We thus developed an Lf-modified liposome system for Osi-resistant NSCLC treatment. It was expected that Osi resistance could be reversed via M2-to-M1 repolarization of TAM, inhibition of EMT and glycolysis, and anti-angiogenesis.

Materials and methods

Materials

Osimertinib and panobinostat were from CSN Pharm (Arlington Heights, USA). Soybean phosphatidylcholine (SPC), cholesterol, distearoyl phosphoethanolamine (DSPE)-PEG2000, DSPE-polyethylene glycol (PEG)-NHS ester (DSPE-PEG-NHS) were purchased from A. V. T. Pharmaceutical Co., Ltd (Shanghai, China). Lactoferrin was purchased from Shengsai Chemical Co., Ltd. (Nanjing, China). RPMI-1640 cell culture medium and Dulbecco’s Modified Eagle’s Medium (DMEM) cell culture medium and fetal bovine serum (FBS) were purchased from Gibco (ThermoFisher Scientific, Waltham, USA). Cocktail protease inhibitors and 3-(4, 5-dimethyl-2-thiazolyl)-2, 5-diphenyltetrazolium bromide (MTT) were obtained from Sigma-Aldrich (St. Louis, USA). Recombinant murine macrophage colony-stimulating factor (M-CSF) was purchased from PeproTech (Rocky Hill, USA). The primary antibodies of EGFR, phospho-EGFR, Akt, phosphor-Akt, VEGF, cleaved-caspase3, TGF-β, GAPDH, C -MYC, HDAC2, LDHA, CD31, BCL-XL, and iNOS were purchased from Cell Signal Technology (Boston, USA). Anti-β-Actin was purchased from Sigma-Aldrich (St. Louis, USA), and antibodies of mannose receptor (CD206), PKM2, LRP-1, Ki67, and TGF-β were purchased from Abcam (Cambridge, UK). The horseradish peroxidase (HRP)-conjugated goat anti-rabbit/mouse, donkey anti-goat IgG secondary antibody, and Annexin V-FITC Apoptosis Detection Kit were purchased from Beyotime (Shanghai, China). All primers for qPCR were synthesized by Generay Biotech (Shanghai, China).

Cell lines

H1975 human NSCLC cells and human umbilical vein endothelial cells (HUVEC) were obtained from the Shanghai Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Osi-resistant human H1975 NSCLC cell lines (H1975/AZDR) were developed and provided by the Department of Oncology, Shanghai Chest Hospital. The cells were cultured in RPMI-1640 medium or DMEM added with 10%FBS, 100 U/ml of streptomycin, and 100 U/ml of penicillin in a humidified incubator with 5% CO₂ at 37 °C_.

Animals

Female BALB/c nude mice (3–4 weeks) and male BALB/c mice (6–8 weeks) were purchased from Shanghai Laboratory Animal Center, Chinese Academy of Sciences and housed at a specific-pathogen-free (SPF) care facility with sterilized food and water under a 12-h light/dark cycle. All the animal experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC), Shanghai Institute of Materia Medica (SIMM), Chinese Academy of Sciences.

Mouse bone marrow-derived macrophage (BMDM) culture and polarization

The BMDMs were collected from the BALB/c mice (male, 6–8 weeks) using a standard protocol based on the previous reports [25]. The cells were cultured in a DMEM medium with 20% FBS and 20 ng/mL M-CSF for four days. The medium was then replaced with DMEM containing IFN-γ (20 ng/mL) and LPS (100 ng/mL) and incubated for another 24 h to induce M1 polarization or with DMEM containing IL-4 (40 ng/mL) to induce M2 polarization.

Preparation of drugs-loaded liposome

The liposomes were prepared using a thin-film-hydration method. SPC, cholesterol, and DSPE-PEG₂₀₀₀ (or DSPE-PEG₂₀₀₀-NHS) at a ratio of 30:1:1 were dissolved in chloroform. The drugs at an optimized ratio of 2:1 (Osi/Pan) were dissolved in methanol. The solutions were mixed and the organic solvents were then removed using a rotary evaporator under a vacuum condition. The thus-formed film was hydrated by 1 mL phosphate buffer saline (PBS). The suspension was subjected to water-bath ultrasound and then extruded through a polycarbonate membrane (0.2 μm) using a liposome extruder (Avanti Polar Lipids, Alabaster, USA). The liposomes modified with DSPE-PEG₂₀₀₀-NHS were further incubated with Lf at a ratio of 5:1 (Lf/NHS) overnight at 4 °C. Afterward, the liposomes were further purified using an Amicon ultrafiltration device (MWCO 100 kDa, Merck, Darmstadt, German) at 3500 rpm for 30 min to remove free lactoferrin. The non-modified liposomes were prepared in the same manner without the addition of lactoferrin.

Characterization of the liposomes

The modification efficiency of Lf in Lf-Lipo was measured by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Briefly, 5 mg Lf was dissolved in PBS at a final concentration of 5 mg/mL. It was then diluted with PBS to prepare the samples with varying concentrations (2, 1.2, 0.8, 0.4, and 0.2 mg/mL). The samples were characterized by SDS-PAGE with Coomassie blue staining, and quantitatively calculated by ImageJ software (National Institutes of Health, Bethesda, USA).

The particle size and polydisperse index (PDI) were determined using a Zeta Size Nanoparticle Analyzer (Nano-ZS90, Malvern, UK). The morphology of the liposomes was measured by transmission electron microscopy (TEM, FEI Talos L120C, Waltham, USA). The encapsulation efficiency and drug-loading capacity of the liposomes were determined by HPLC (1260 Infinity, Agilent Technologies, Santa Clara, USA), and calculated according to the following formula:

$$\mathrm{Encapsulation}\mathrm{efficiency}\left(\%\right)=\frac{\mathrm{weight}\mathrm{of}\mathrm{encapsulated}\mathrm{drug}}{\mathrm{weight}\mathrm{of}\mathrm{total}\mathrm{drug}}\times 100\%$$

$$\mathrm{Drug}\mathrm{loading}\mathrm{capacity}\left(\%\right)=\frac{\mathrm{weight}\mathrm{of}\mathrm{encapsulated}\mathrm{drug}}{\mathrm{weight}\mathrm{of}\mathrm{liposomes}}\times 100\%$$

The stability of the liposomes was evaluated by dispersing the samples in PBS (pH 7.4) containing 10% FBS and with a shaking speed of 150 rpm in an incubator at 37 °C. The liposome size was monitored by a Zeta Size Nanoparticle Analyzer at different time points.

The in vitro drug release was assessed using a standard dialysis method (MWCO 10–14 kDa) in PBS containing SDS (0.5% v/v) under a sink condition at 37 °C with 150 rpm. The drug concentration was determined by HPLC.

Cellular uptake efficiency

The H1975/AZDR cells were incubated with the liposomes (Lipo and Lf-Lipo) that were labeled with coumarin 6 for 1 h. In addition, free Lf was pre-incubated with the cells for competitive inhibition of the Lf-Lipo to investigate Lf-mediated delivery ability. Afterward, the cells were washed with PBS, fixed in 4% paraformaldehyde (PFA) for 20 min, and stained with DAPI. Fluorescent images were observed with a fluorescence microscope (Leica DM 6B fluorescence microscope, Wetzlar, Germany). Besides, the M2 cells were treated for fluorescent imaging and for flow cytometry (Agilent ACEA NovoCyte 3000, Santa Clara, USA) to determine the cellular uptake efficiency.

Macrophage repolarization and EMT reversal

EMT reversal in H1975/AZDR cells was investigated by incubating the cells with Lf-Lipo for 24 h. The H1975 cells and H1975/AZDR cells treated with Lf-Lipo were used to analyze the difference in E-cadherin expression by a standard Western blotting method. The quantification of protein bands was conducted using densitometry with ImageJ. The macrophages were seeded in a 12-well plate at a density of 1 × 10⁵ cells per well and incubated for 24 h. The cells were treated with free drugs Osi (2 μM), Pan (1 μM), Osi (2 μM) + Pan (1 μM), Lipo, or Lf-Lipo for 12 h. After treatment, the M1 or M2 cells were collected and then subjected to a standard flow cytometry analysis and Western blotting assay.

In vitro cytotoxicity assay, apoptosis analysis and glycolysis regulation by Lf-Lipo

Cytotoxicity was measured using a standard MTT assay in H1975/AZDR cells. The samples were measured at 490 nm using a microplate reader (Multiskan, ThermoFisher, Waltham, USA). IC₅₀ values were calculated.

Furthermore, the H1975/AZDR cell apoptosis was investigated. The cells were seeded in a 6-well plate and cultured for 24 h. The cells were then incubated with the liposomes (equal to Osi 2 µM and Pan 1 µM) or the free drugs for 24 h. After a thorough wash with PBS, the cells were collected and stained with a FITC-Annexin V Apoptosis Detection Kit (Beijing Solarbio Science & Technology Co., Ltd., China) according to the manufacturer’s protocol. In addition, the cells treated with the drugs were also collected for Western blotting assay.

The H1975/AZDR cells were seeded in the 24-well plates at a density of 5 × 10⁴ cells per well and incubated for 24 h. The cells were treated with the drugs at a dose of 2 μM Osi and 1 μM Pan. After 12 h of incubation, the supernatant was collected to measure the lactic acid concentrations using a lactic acid assay kit (Jiancheng Bioengineering Institute, Nanjing, China). Meanwhile, the cells were collected to analyze the levels of HDAC2 and PKM2 by Western blotting.

Anti-angiogenesis study in vitro

The HUVCEs were seeded in the 24-well plates that were pre-coated with Matrigel matrix (Corning, USA) and M1 (or M2) were seeded in an upper chamber of Transwell cassette (Corning, USA) for 12 h. The cells were then treated with free drugs Osi (2 μM), Pan (1 μM), Osi (2 μM) +Pan (1 μM), Lipo, or Lf-Lipo for 12 h. The endothelial cell tube formation was observed by an inverted microscope in a bright field (10×) and the further analysis using ImageJ. In addition, the cells were collected to detect the expression of VEGF by Western blotting assay.

The Osi-resistant xenograft NSCLC -bearing animal models

The Osi-resistant xenograft NSCLC-bearing mouse model was established by subcutaneously injecting the H1975/AZDR cells (5 × 10⁶) into the back of the BALB/c nude mice. The mice were housed under standard conditions.

In vivo imaging and biodistribution

The Osi-resistant tumor xenograft-bearing mice were intravenously injected with an equal dose of the Cy5.5-labeled Lipo or Lf-Lipo, and the biodistribution of the liposomes was monitored by an imaging system (IVIS SPECTRUM, Caliper PerkinElmer, USA). At the endpoint, the mice were humanely sacrificed to collect the organs and tumors. The organs and tumors were dissected for ex vivo imaging and then used for Western blotting. The tumor tissues were then mechanically chopped and digested with hyaluronidase and collagenase for the preparation of single-cell suspension. The cells were then stained with anti-LRP-1 antibody followed by Alexa Fluor 488 fluorescent antibody staining. The quantitative analysis of the Cy5.5⁺LRP-1⁺ cells was carried out using flow cytometry.

In vivo therapeutic efficacy and mechanism studies

The therapy study was performed 8 days, with the tumor volume reaching 100−200 mm³, after subcutaneous injection of the H1975/AZDR cells. The Osi-resistant xenograft-bearing mice were randomly divided into six groups, and each group contained four mice. The mice were treated with Osi (6 mg/kg), Pan (2 mg/kg), Osi + Pan (6 mg/kg +2 mg/kg), Lipo, or Lf-Lipo (equal dose with the combination therapy) via tail vein injection every other day. PBS was used as a control. Tumor size and bodyweight were measured every other day and calculated according to the formula below:

$$V={ab}^2/2\left(a\mathrm{represents}\mathrm{the}\mathrm{length}\mathrm{and}b\mathrm{represents}\mathrm{the}\mathrm{width}\right).$$

The treatment endpoint was set at the tumor volume exceeding 2000 mm³ or the bodyweight loss larger than 15%. The mice were euthanized and the major organs including heart, liver, spleen, lung, kidney, and tumor were dissected for weighing to calculate organ coefficients at the end of treatment. The tumor tissues collected from each group were photographed and weighed to calculate the inhibition rate of tumor growth and for further study.

Histopathological examination

At the treatment endpoint, the major organs (heart, liver, spleen, lung, and kidney) were collected for histopathological examination by hematoxylin and eosin (H&E) staining method to evaluate the adverse effects.

Mechanism studies

To investigate the mechanisms underlying the anti-tumor efficacy of the Lipo and Lf-Lipo, the tumors were dissected for detection of TAM at the endpoint. M1 and M2 TAM in the tumor tissues were observed by Western blotting as described above. The antibodies anti-CD206 and anti-iNOS were used to label M2 and M1, respectively. Besides, the tumor tissues were used for detection of the mRNA expression of CD206, iNOS, TGF-β, TNF-α, IFN-γ, and VEGF using quantitative real-time polymerase chain reaction (qRT-PCR).

The intratumoral concentrations of lactate were measured by a lactic acid detection kit. The protein expression of Ki67, VEGF, CD31, PKM2, HDAC2, C-Myc, LDHA, EGFR/Akt, and their phosphorylated forms were detected by Western blotting as described above.

Statistical analysis

The statistical analysis was performed by t-test and one-way ANOVA. Data were expressed as mean ± SD(n ≥ 3). Statistical difference was defined as *P < 0.05, **P < 0.01, ***P < 0.001.

Results

Characterization of the Lf-modified liposomes and cellular uptake study

The Lf-modified liposomes (Lf-Lipo) with co-encapsulation of Pan and Osi were prepared using a thin-film-hydration method. The Lf-Lipo had a mean diameter of 122.3 nm, a spheroidal shape, and a zeta potential around 20 mV (Fig. 1a, b). The modification ratio of Lf in Lf-Lipo was about 0.2:1 (w/w, Fig. 1c), as detected by Coomassie brilliant blue staining in SDS-PAGE. The drug-loading capacity (DL%) and entrapment efficiency (EE%) are listed in Fig. 1d. Both the Lipo and Lf-Lipo displayed good stability in a physiologically mimic medium containing 10% of FBS (Fig. 1e). Both liposomes had a sustained release pattern (Fig. 1f).

Fig. 1. Characterization of the liposomes and cellular uptake study.

a Particle size and TEM of the Lipo and Lf-Lipo (scale bar: 200 nm). b Zeta potential of Lipo and Lf-Lipo. c Modification ratio of Lf in Lf-Lipo. d Encapsulation efficiency and drug-loading capacity. e Stability of Lipo and Lf-Lipo. f Cumulative drug release of Lipo and Lf-Lipo. g LRP-1 and HDAC2 expression. h Fluorescence images in H1975/AZDR cells after incubation with the Coumarin-6-labeled liposomes (scale bar: 100 μm). i Quantitative analysis of cellular uptake efficiency via flow cytometer. j Statistical analysis of uptake efficiency. Data are presented by mean ± SD (n = 3), *P < 0.05, **P < 0.01.

The fluorescence imaging results demonstrated the enhanced uptake efficiency of Lf-Lipo in the H1975/AZDR cells (Fig. 1h). The flow cytometry results showed that the Lf-Lipo group had a 1.4-fold increase in uptake compared to the Lipo group (Fig. 1i, j). The cellular uptake of Lf-Lipo was dramatically decreased by approximately 60% when the H1975/AZDR cells were pretreated with free Lf (Fig. 1j). The flow cytometry results also showed that the cellular uptake of Lf-Lipo in M2 was significantly higher than that of Lipo. The Lf-Lipo displayed a 1.4-fold higher uptake efficiency than the Lipo in M2 cells (Fig. S1). The enhancement of Lf-Lipo in cellular uptake was attributed to the cell surface LRP-1 overexpressed in the H1975/AZDR cells and M2 cells (Figs. 1g and 2b). The results revealed that the promise of Lf-mediated targeting the cancer cells and TAMs.

Fig. 2. Macrophage repolarization and EMT reversal.

a MΦ repolarization from M2 to M1 by liposomes treatment. b The LRP-1 and M2-related marker CD206 expression. c Western blotting analysis of of EMT markers (E-cadherin and Vimentin) in H1975 cells and H1975/AZDR cells induced by Lf-Lipo. d The grayscale analysis of E-cadherin by ImageJ software. e The grayscale analysis of Vimentin by ImageJ software. Data are expressed as mean ± SD (n = 3). ***P < 0.001.

Macrophage repolarization and EMT reversal

TAM is a major component of the TME in NSCLC [26]. M2-to-M1 repolarization led to a reduction of TGF-β secretion, which thus suppressed TGF-β-driven EMT and yielded an effect of reversing drug resistance [27]. Therefore, the regulation of TAM is a promising therapeutic strategy [28]. The close association between epigenetic regulation and macrophage-mediated immunity has been well-documented [29]. For instance, HDAC inhibition can promote the anti-tumor activity of macrophages and amplify anti-PD-L1 immunotherapy [30]. Our previous works revealed that HDACi (e.g., vorinostat and Pan) can effectively repolarize TAM from M2 to M1 phenotype [21, 31]. In this work, the results showed that Lf-Lipo treatment increased the ratio of M1 (iNOS⁺)/M2 (CD206⁺) macrophages, signifying the effect on repolarization from M2 to M1 (Fig. 2a). Meanwhile, Lf-Lipo treatment decreased the level of M2-related marker CD206, and the targeting delivery of Lf-Lipo to M2 macrophages was related to the overexpression of LRP-1 in M2 (Fig. 2b). Notably, the ratio of M1/M2 macrophages can be used as a prognostic predictor for cancer therapy, and the high ratio represents a favorable outcome, while the low ratio indicates a poor prognosis [32].

The expression of EMT-associated proteins was tested in the H1975/AZDR cells and the parental H1975 cells. The results showed that the H1975/AZDR cells exhibited EMT phenotype, as characterized by low E-cadherin and high vimentin (Fig. 2c), indicating the connection of EMT with Osi resistance. Lf-Lipo treatment was effective to reverse EMT, as evidenced by the up-regulation of E-cadherin and downregulation of vimentin (Fig. 2d, e).

Anti-tumor effect in Osi-resistant H1975/AZDR cells

Given the crucial role of HDAC in drug resistance of tumors [33], we detected the HDAC2 expression of H1975 and H1975/AZDR. The results indicated that HDAC2 expression was increased in the H1975/AZDR cells compared to the parental H1975 cells (Fig. 1g). HDACi has potent therapeutic effects by inducing cell cycle arrest, apoptosis, and immunogenicity, as well as inhibiting angiogenesis and DNA repair on various malignancies [34]. HDACi treatment showed anti-proliferative activity in NSCLC [35, 36]. The efficacy of combination therapy of Osi/Pan in the Osi-resistant H1975/AZDR cells was examined. The Lf-Lipo displayed the highest anti-tumor efficacy, with the IC₅₀ of 5.7 μΜ, compared to 10.2 μΜ of the combination of free Osi/Pan (Fig. 3a). By contrast, single use of Osi or Pan showed the IC₅₀ of 22.9 or 33.5 µM. The cell apoptosis results further confirmed the most potent efficacy of Lf-Lipo (Fig. 3b, c). It showed that Lf-Lipo-mediated combination therapy could overcome the drug resistance and re-sensitize the H1975/AZDR cells to the treatment.

Fig. 3. In vitro anti-tumor effect and glycolysis regulation by Lf-Lipo.

a MTT assay of H1975/AZDR. b Flow cytometric analysis of H1975/AZDR cells apoptosis. c The total apoptosis rate of different groups. d Apoptotic protein levels of BCL-2, BCL-XL, and cleaved caspase3. e The production of lactic acid in H1975/AZDR cells. f PKM2 and HDAC2 expression in H1975/AZDR cells. g Downregulation of p-EGFR/p-AKT. Data are expressed as mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001.

Besides, the Western blotting results showed the regulation effect of Lf-Lipo on the apoptosis-related proteins; the level of apoptosis-inhibiting protein Bcl-xl was downregulated while the apoptosis-promoting protein cleaved caspase3 was up-regulated (Fig. 3d).

Moreover, lactic acid is the primary end product of tumor glycolysis, which plays a crucial role in tumor immunosuppression and drug resistance [37, 38]. Specifically, lactic acid is a key regulator in macrophage polarization and drives the polarization of protumor M2 phenotype [39]. Our results showed that Lf-Lipo treatment reduced lactic acid secretion from the H1975/AZDR cells (Fig. 3e). The mechanism could be associated with the downregulated HDAC2 and pyruvate kinase M2 (PKM2) (Fig. 3f).

Notably, EGFR/Akt pathway is an important growth signaling in cancer cells [40]. Both the combination of Pan/Osi and the liposomes strongly inhibited the EGFR/Akt phosphorylation, and Lf-Lipo showed the highest efficiency (Fig. 3g). Therefore, the results indicated that the Lf-Lipo could inhibit the activation of EGFR pathway and induce the apoptosis of the Osi-resistant NSCLC cells.

In vitro anti-angiogenesis study

Cancer cell-derived VEGF is a driving factor of angiogenesis [41] and the accompanying activation of VEGF/VEGFR pathway can drive EMT and TKI resistance [13, 15, 16]. As shown in Fig. 4b, Lf-Lipo treatment greatly reduced the VEGF secretion from the H1975/AZDR cancer cells.

Fig. 4. In vitro anti-angiogenesis study.

a Illustration of the HUVEC/BMDM co-culture system. b Expression level of VEGF in H1975/AZDR. c HUVEC tube formation inhibited by Lf-lipo (scale bar: 100 μm). Anti-angiogenesis image analysis. d Number of meshes. e Number of junctions. f Total of segment length. Data are expressed as mean ± SD (n = 3). **P < 0.01, ***P < 0.001.

TAM-derived TGF-β can also induce angiogenesis; moreover, the activation of TGF-β signaling leads to EMT and drug resistance [14, 42]. The anti-angiogenesis effect of Lf-Lipo was examined in a HUVEC/BMDM co-culture system (Fig. 4a). M2 macrophages induced angiogenesis, showing significant tube formation, while M1 displayed an anti-angiogenesis effect (Fig. 4c). Lf-Lipo treatment efficiently inhibited tube formation (Fig. 4c) and significantly reduced amounts of junctions, meshes, and total segment length compared to the M2 group (Fig. 4d–f).

In vivo imaging and biodistribution study

The tumor-targeting ability of the Lf-Lipo labeled with Cy5.5 dye was examined using in vivo imaging. The accumulation of Lf-Lipo and Lipo in the tumor sites was rapid (Fig. 5a, b). The Lf-Lipo group showed a greater intratumoral accumulation than the Lipo group (Fig. 5e, f). The Western blotting results showed that LRP-1 was highly expressed in the H1975/AZDR tumor tissues (Fig. 5d), which could be responsible for the increased tumor delivery of Lf-Lipo. Besides, the flow cytometry analysis showed that the percentage of intratumoral Cy5.5⁺LRP-1⁺ cells in the Lf-Lipo group was 1.56-fold higher than the Lipo group (Fig. 5g, h), further confirming the preferential uptake of the Cy5.5-labeled Lf-Lipo mediated by LRP-1 that was overexpressed in the target cells.

Fig. 5. In vivo imaging and distribution of Lf-Lipo.

a Distribution of Lipo and Lf-Lipo in H1975/AZDR tumor-bearing mice (n = 3). b In vivo radiant efficiency at the tumor sites. c Representative photomicrographs of ex vivo imaging of major organs. d Western blotting analysis of LRP-1 expression in the tumor tissues and major organs. e Ex vivo imaging of the tumor tissues. f The tumor radiant efficiency. g The population of Cy5.5⁺ LRP-1⁺ cells measured by flow cytometry. h Statistical analysis of Cy5.5⁺ LRP-1⁺ cells. Data are expressed as mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001.

Anti-tumor efficacy of Lf-Lipo in the Osi-resistant NSCLC mouse model

The mice bearing Osi-resistant H1975/AZDR subcutaneous tumor were treated according to the therapeutic schedule in Fig. 6a. Osi treatment only yielded marginal efficacy; by contrast, the treatment of Lf-Lipo or Lipo efficiently retarded the tumor growth (Fig. 6b–d). At the endpoint, the tumor inhibition rate was calculated, showing 82% and 65% for the Lf-Lipo and Lipo groups, respectively, compared to 51% of the combo-free drugs and 45% for the single use of Osi (Fig. 6e). In addition, after Lf-Lipo treatment, the cell proliferation marker Ki67 was downregulated, which suggested the inhibition of tumor cell growth (Fig. 6h).

Fig. 6. Anti-tumor efficiency of Lf-Lipo in Osi-resistant human NSCLC in vivo.

a Therapeutic schedule. b Images of Osi-resistant tumors at the treatment endpoint. c Tumor growth curves. d The tumor weight at the endpoint. e Inhibition rate of tumor growth. f Bodyweight changes during treatment. g Organ coefficients. h The expression of Ki67 in the tumors. i Histopathological examination of major organs after treatment (scale bar: 100 μm). Data are expressed as mean ± SD (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001.

The biosafety of Lf-Lipo was evaluated by monitoring the bodyweight changes during treatment, and there was no statistical difference compared to the control group (Fig. 6f). At the endpoint, the organ coefficient was calculated and there was no significant difference among the groups (Fig. 6g). H&E staining showed no obvious pathological abnormalities in the main organs of the mice in all groups (Fig. 6i). These results indicated the safety of the treatment.

Remodeling tumor microenvironment

The intratumor TAMs was analyzed and the results showed Lf-Lipo effectively polarized M2 to M1 phenotype, with the downregulation of M2-related marker CD206 but up-regulation of M1-related marker iNOS (Fig. 7d–f). As a result, the immunosuppressive status was reversed (e.g., TGF-β downregulation), and the anticancer immunity was activated (e.g., IFN-γ and TNF-α up-regulation) (Fig. 7b, c, h).

Fig. 7. Remodeling tumor microenvironment.

The mRNA levels of VEGF (a), TGF-β (b), IFN-γ (c), CD206 (d), and iNOS (e). f Western blotting analysis of M1- and M2-associated markers. g Western blotting analysis of CD31 and VEGF. h The expression of TNF-α and TGF-β. i Downregulation of p-EGFR/p-AKT. j The intratumoral levels of lactic acid. k, l PKM2, HDAC2, c-Myc, LDHA, and p-LDHA expression in the tumors after treatment. Statistically significant differences are presented as: *P < 0.05, **P < 0.01, ***P < 0.001.

Moreover, the intratumor lactate levels were significantly reduced in the Lf-Lipo group, indicating that the aerobic glycolysis was efficiently inhibited by Lf-Lipo (Fig. 7j). HDAC2, PKM2, and LDHA expression were remarkably reduced while p-LDHA was up-regulated (Fig. 7k, l). Besides, c-Myc is a key factor in mediating glycolysis and promoting M2 macrophage polarization [43]. Our results showed c-Myc was downregulated by Lf-Lipo treatment (Fig. 7k). It suggested that inhibition of HDAC activity was associated with aerobic glycolysis suppression.

The anti-angiogenic results showed that Lf-Lipo downregulated the intratumor VEGF level (Fig. 7a, g), and consequently, the expression of an angiogenesis marker CD31 was reduced (Fig. 7g), suggesting the reduced neovascular density.

The Western blotting results showed that Lf-Lipo treatment inhibited the EGFR/Akt signaling pathway that facilitates tumor growth (Fig. 7i), which was consistent with the in vitro results.

Discussion

Osi has been approved as a first-line treatment for NSCLC patients with EGFR mutations. Despite its high efficacy, therapeutic resistance to Osi rapidly develops, and there are no clear-cut therapeutic options other than chemotherapy and locally ablative therapy [44]. The mechanisms of resistance can be broadly classified into two categories: EGFR-dependent and EGFR-independent mechanisms [45, 46]. However, drug resistance generally involves multiple mechanisms, and therefore, Osi-based combination therapy could be a potential solution.

We developed a nanomedicine-based combination therapy to codeliver Osi and Pan to the tumor based on Lf-mediated binding to the overexpressed LRP-1 on the surface of both tumor cells and TAM to regulate the tumor immune microenvironment. Our results showed that Lf-Lipo was able to repolarize TAM from the M2 to M1 phenotype and inhibit EMT and neovascularization, yielding a synergistic effect against Osi-resistant tumor growth. This was due to the downregulation of protumor and pro-angiogenesis cytokines (e.g., TGF-β and VEGF) and the suppression of the EGFR/Akt/Erk signaling pathway.

Recently, the relationship between resistance to TKI and the tumor microenvironment has been revealed [47]. TAMs are the major group of immune cells in the tumor microenvironment, and they have been well-documented to induce TKI resistance. For example, a high TAM count correlates with a poor response to TKI treatment and can serve as a predictor of TKI treatment outcomes in NSCLC [48, 49]. Repolarizing TAMs from the M2 to M1 phenotype can reverse therapeutic resistance to the first-generation TKI gefitinib [28, 31, 40].

The mechanisms involved in TAM-associated drug resistance are highly complicated and multifaceted. For example, TAM-derived TGF-β drives EMT and angiogenesis, which are associated with therapeutic resistance. In this work, we focused on M2-to-M1 repolarization of TAMs using a HDACi Pan and demonstrated that reversal of Osi resistance could be associated with the suppression of EMT and angiogenesis induced by TAM. EMT could be an important mechanism responsible for Osi resistance, as the H1975/AZDR cells exhibited EMT phenotype. Our strategy was to repolarize the M2 phenotype and thus diminish the secretion of TGF-β.

In addition, the glycolytic end product lactic acid is known to be a strong immunosuppressor that promotes TAM2 differentiation and TGF-β expression, as well as a potent angiogenic factor [50]. Specifically, glycolysis and EMT are intertwined [51]. The increased intratumor lactic acid stimulates Snail and EMT and activates the TGF-β pathway [52]. The activated TGF-β signaling not only induces EMT but also upregulates EGFR expression [53, 54]. EGFR activation promotes nucleus PKM2 phosphorylation that upregulates cyclin D1 and c-Myc and facilitates cancer cell growth [54, 55]. The crosstalk between PKM2 and TGF-β–induced factor homeobox 2 (TGIF2) leads to the suppressed E-cadherin and induced EMT [56]. Our previous study showed that the mechanisms of the EMT reversal could inhibit tetramer PKM2-mediated glycolysis and reduce lactic acid-driving mesenchymal-like differentiation via blocking the TGF-β/EGF-cascaded PKM2 nuclear ectopic [57].

In this work, Lf-Lipo could inhibit glycolytic lactic acid production and subsequently reverse EMT, thus overcoming Osi resistance. Drug resistance involves not only genetic mutations but also epigenetic regulations. Although there has been a significant increase in research on the relationship between epigenetics and drug resistance, the mechanisms have not been fully revealed yet. It has been reported that cisplatin resistance can be overcome by combining cisplatin with Pan [58]. However, the effect of HDACi on reversing Osi resistance was not reported.

HDAC is involved in cancer cell proliferation, angiogenesis, DNA repair, metabolism, and immunogenicity in various malignancies [34]. Our results indicated that pan could overcome Osi resistance via multiple effects, such as TAM repolarization, EMT reversal, glycolysis inhibition, and anti-angiogenesis, suggesting a potential combination therapy. However, further investigation is needed to elucidate the mechanisms in detail.

Conclusion

To summarize, we have developed a liposome-based codelivery system (Lf-Lipo) for Pan and Osi to regulate the tumor immune microenvironment and reverse therapeutic resistance in an Osi-resistant NSCLC experimental model. We demonstrated that Lf-Lipo can promote tumor cell apoptosis, inhibit aerobic glycolysis, reverse EMT, repolarize TAM, and suppress angiogenesis. This study proposes an HDACi-based combination therapeutic strategy with enhanced drug delivery efficiency and treatment efficacy against Osi-resistant NSCLC.

Author contributions

TTL and YZH designed the research and performed data analysis. TTL, WX, and GHC carried out the experiments. LL, YH, and JLZ participated part of the experiments. TTL wrote the manuscript. WWL and XFG analyzed the manuscript formal. YZH revised the manuscript. All of the authors have read and approved the final manuscript. The authors declare no competing financial interest.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41401-023-01205-4.