Proton pump inhibitors enhance macropinocytosis‐mediated extracellular vesicle endocytosis by inducing membrane v‐ATPase assembly

Abstract

Besides participating in diverse pathological and physiological processes, extracellular vesicles (EVs) are also excellent drug‐delivery vehicles. However, clinical drugs modulating EV levels are still lacking. Here, we show that proton pump inhibitors (PPIs) reduce EVs by enhancing macropinocytosis‐mediated EV uptake. PPIs accelerate intestinal cell endocytosis of autocrine immunosuppressive EVs through macropinocytosis, thereby aggravating inflammatory bowel disease. PPI‐induced macropinocytosis facilitates the clearance of immunosuppressive EVs from tumour cells, improving antitumor immunity. PPI‐induced macropinocytosis also increases doxorubicin and antisense oligonucleotides of microRNA‐155 delivery efficiency by EVs, leading to enhanced therapeutic effects of drug‐loaded EVs on tumours and acute liver failure. Mechanistically, PPIs reduce cytosolic pH, promote ATP6V1A (v‐ATPase subunit) disassembly from the vacuolar membrane and enhance the assembly of plasma membrane v‐ATPases, thereby inducing macropinocytosis. Altogether, our results reveal a mechanism for macropinocytic regulation and PPIs as potential modulators of EV levels, thus regulating their functions.

1. INTRODUCTION

Extracellular vesicles (EVs) are mainly produced by the outward budding of the plasma membrane (PM) or by an endosomal pathway involving the fusion of multivesicular bodies (MVBs) with the PM (Mathieu et al., 2019). As transmitters of information between different cells, EVs have diverse functions in many pathophysiological situations, including cancers, immune responses, nervous system diseases and cardiovascular diseases (Kalluri & LeBleu, 2020). We and others have reported that the advancement of EV‐related diseases can be altered by regulating EV biogenesis. For instance, inhibiting intestinal epithelial cell‐derived EVs (IEC‐EVs) released by neutral sphingomyelinase inhibitor GW4869 aggravated inflammatory bowel disease (IBD) development (Jiang et al., 2016). Reducing tumour cell‐derived EVs (TEVs) biogenesis by knocking out Coro1a enhanced antitumor immunity effectively, thereby restricting tumour progression (Fei et al., 2021). Furthermore, Rab27a silence‐mediated decrease in TEV secretion suppressed tumour metastasis (Peinado et al., 2012).

Except for transmitting information to acceptor cells by acting at the cell surface (Zhang et al., 2019), EV uptake by effector cells is the prerequisite for EV functions involving intracellular cargoes (Mathieu et al., 2019). Therefore, we supposed that besides regulating EV biogenesis, EV functions could also be modulated by interfering with effector cells to take up EVs. In addition, enhanced EV uptake by non‐effector cells will accelerate EV clearance, thereby ceasing EV functions. Clathrin‐ and caveolin‐dependent pathways are reported to mediate EV uptake (Mulcahy et al., 2014; Svensson et al., 2013). Macropinocytosis is also engaged in EV internalisation (Costa et al., 2017). However, whether EV uptake can be modulated by regulating these pathways via clinical compounds is undeciphered.

Macropinocytosis allows cells to ingest proteins in the external milieu via the fluid phase. Activation of RAS and PI3K signalling and downstream recruitment of RAC1 are essential for the induction of macropinocytosis (King & Kay, 2019). PM v‐ATPases matter for RAS‐induced macropinocytosis due to their crucial role in the PM translocation of RAC1, and PKA activation by RAS is necessary for the PM accumulation of v‐ATPases (Ramirez et al., 2019). V‐ATPases are proton pumps (PPs) driven by ATP hydrolysis to transport protons into intracellular organelles, thereby regulating pH in the organelles. V‐ATPase subunits are broadly organised into a soluble ATP‐hydrolytic domain (V₁) and a membrane‐embedded proton‐translocation domain (V₀). In addition to being primarily located on the membrane of vacuoles, v‐ATPases are also expressed on the PM (Forgac, 2007; Vasanthakumar & Rubinstein, 2020). The regulation of the balance between vacuolar and PM v‐ATPases is an untouched area.

PPs mainly include P‐type H⁺/K⁺‐ATPases and V‐type H⁺‐ATPases (v‐ATPases). P‐type H⁺/K⁺‐ATPases are located on the PM, and v‐ATPases especially accumulate on the vacuole membrane (Martinez‐Zaguilan et al., 1993). Due to their powerful effect on inhibiting gastric acid secretion, P‐type H⁺/K⁺ PP inhibitors (PPIs) have been used to treat gastritis and gastric ulcer in clinics for a long time (Li et al., 2013; Yatime et al., 2009). Although v‐ATPase inhibitors have been demonstrated to inhibit macropinocytosis, the role of PPIs in macropinocytosis is still unclear. As clinically approved drugs, if PPIs can regulate macropinocytosis, treatment with EV‐related diseases by regulating EV functions will come true.

Here, we show that a PPI‐mediated decrease in cytosolic pH induces the dissociation of ATP6V1A from the vacuolar membrane. Then, ATP6V1A translocates to PM and promotes the assembly of PM v‐ATPases, leading to enhanced macropinocytosis. In this way, PPIs accelerate the elimination of immunosuppressive IEC‐EVs and TEVs, which aggravates IBD or mounts antitumor immunity, respectively. In addition, PPIs promote the uptake of drug‐loaded red blood cell‐derived EVs (RBC‐EVs) by recipient cells, resulting in EVs’ enhanced drug delivery efficacy. Altogether, our results demonstrate that PPIs can regulate endogenous EV functions and improve the activity of exogenous EVs as delivery vehicles.

2. MATERIAL AND METHODS

2.1. Mice

C57BL/6J, BALB/c, nude mice (6–8 weeks old) were purchased from SLAC Laboratory Animal Co. LTD (Shanghai, China). Rab27a^fl/fl , Atp6v1a^fl/fl , Cmv‐Cre, Villin‐Cre and Rosa‐LSL‐tdtomato mice were purchased from Cyagen Biosciences (Suzhou, Jiangsu, China). All mice were housed in a specific pathogen‐free facility, and the Animal Care and Use Committee of Zhejiang University School of Medicine approved experimental protocols.

2.2. Human samples

Human blood samples from gastric tumour patients with PPI administration were obtained from the Second Affiliated Hospital, Zhejiang University School of Medicine. The local Ethics Committee and the Review Board of the Second Affiliated Hospital, Zhejiang University School of Medicine, approved the collection of human samples. Information on the 183 patients with advanced lung cancer treated with single‐agent anti‐PD‐1 therapy or combined with PPI treatment was obtained from Zhejiang Cancer Hospital. All patients were informed of the usage of their samples, and signed consent forms were obtained. Information regarding the human samples is listed in Tables S1 and S2.

2.3. Cell culture

4T1, HeLa, B16F10, A549, MC38, RAW264.7, HEK293T and THP‐1 cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). 4T1, HeLa, B16F10, A549, MC38, RAW264.7 and HEK293T cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% (v v⁻¹) foetal calf serum (FBS) and 1% penicillin/streptomycin. THP‐1 cells were cultured in RPMI‐1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin. To obtain bone marrow‐derived dendritic cells (BMDCs), the tibias and femurs were removed with sterile techniques, and the bone marrow was flushed with fresh RPMI‐1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin. The bone marrow cells were then plated in RPMI‐1640 medium supplemented with GM‐CSF (50 ng mL⁻¹) for 6 days for subsequent experiments. To obtain primary mouse peritoneal macrophages (PEMs), mice (6–8 weeks old) were injected intraperitoneally with 3% fluid thioglycolate medium (Merck, Darmstadt, Germany). Three days later, peritoneal lavage fluid was collected and centrifuged. Cells were cultured in RPMI‐1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin; nonadherent cells were removed 2 h later, and the adhesive monolayer cells were washed with RPMI‐1640 medium and used as PEMs.

2.4. Cell treatment

Cells were treated with 10 μM rabeprazole (Rabe) or omeprazole (Ome) to inhibit PPs (Selleck, Shanghai, China). Cells were treated with 10 μM Cytochalasin D to inhibit phagocytosis (MedChemExpress, Shanghai, China). Cells were treated with 10 μM indomethacin (MedChemExpress) or chlorpromazine (Selleck) to inhibit caveolae‐ or clathrin‐mediated endocytosis, respectively. Macropinocytosis was inhibited by treatment cells with 10 μM LY294002 or amiloride (Selleck). Cells were treated with 10 μM KM91104 or 10 nM Baf‐A1, or 10 μM azathioprine (Selleck) to inhibit the activity of v‐ATPase or RAC1, respectively. Cells were treated with 10 μM Dynasore (MedChemExpress), CDC42‐IN‐1 (Selleck) or NAV‐2729 (MedChemExpress) to inhibit the activity of dynamin, CDC42 and ARF6, respectively. Cell cholesterol was extracted by treatment with 10 μM MeβCD (Aladdin). Equal pH gradients in cells were induced by 10 μM 2, 4‐DNP (Merck). Cells were treated with 10 μM H89 (Selleck) to inhibit PKA activation. In most experiments, cells were pretreated with the above reagents for 2 h, followed by subsequent experiments.

2.5. Isolation of EVs

For EV isolation, cell culture supernatants were centrifuged at 500 × g for 10 min, 2000 × g for 15 min and 10,000 × g for 30 min at 4°C. Then, the supernatants were passed through 0.22 μm syringe filters (Millipore) and collected in 35 mL ultracentrifuge tubes (Beckman Coulter, Brea, CA, USA). The EVs were concentrated using ultracentrifugation with an SW32Ti rotor (L‐90K with SW32Ti rotor, Beckman Coulter) at 100,000 × g for 70 min at 4°C. The EV protein contents were quantified by a BCA protein assay kit (Thermo Fisher Scientific). For IEC‐EVs isolation, intestinal tissues were collected and digested for 2 h with 1 mg mL⁻¹ type II collagenase (Sigma) and centrifuged at 300 × g for 10 min, the supernatants were collected, and IEC‐EVs were separated as described above.

2.6. Electron microscopy

For negative EV staining, 200‐mesh carbon films were hydrophilised with a glow discharge instrument at 15 mA for 25 s. An EV solution was pipetted onto 200‐mesh carbon‐coated copper grids and kept at RT for 1 min. After removing any excess suspension with filter paper, the EVs were negatively stained with 2% uranyl acetate at RT for 1 min. Any extra suspension was removed before the grids were air‐dried. Images were acquired by electron microscopy (Tecnai G2 Spirit 120 kV, Thermo FEI, Hillsboro, OR, USA).

2.7. Nanoparticle tracking analysis

To measure particle sizes and concentrations, EVs purified from DMSO‐ and Rabe‐treated 4T1 cells were analysed by nanoparticle tracking analysis using a NanoSight NS300 system (Malvern PANalytical) configured with a 488 nm laser and high‐sensitivity sCMOS camera and were finally analysed with NTA 3.2 or 3.3 software.

2.8. ELISA

Sandwich ELISA measured serum EVs in the individual samples. Briefly, 96‐well ELISA plates were coated with 4 μg mL⁻¹ purified anti‐CD63 antibodies in coating buffer and incubated overnight at 37°C. After blocking with assay diluent, serum samples in triplicate were added to individual wells and incubated overnight at 37°C. The plates were washed, and the bound EVs were detected by incubation with anti‐CD9 or PD‐L1 and HRP for 2 h at 37°C. In some experiments, 96‐well ELISA plates were coated with 4 μg mL⁻¹ purified anti‐A33⁺ or CD11c⁺ antibodies, and the bound EVs were detected by incubation with anti‐CD63 and HRP for 2 h at 37°C. Then, the signal was developed with tetramethylbenzidine, and the samples were blocked with 2 M H₂SO₄, after which absorbance at 450 nm was measured with a SpetraMax M5 microplate reader (Molecular Devices, San Jose, CA, USA). The antibodies used are listed in Table S3.

2.9. Immunofluorescence staining and confocal microscopy

4T1 and HeLa cells were plated on glass coverslips in 24‐well plates. Cells were fixed with 4% paraformaldehyde for 30 min at room temperature (RT), permeabilised using 0.1% Triton X‐100, blocked with 5% BSA in PBS for 1 h and stained with specific antibodies. Primary antibodies were detected using iFluor 488‐ or iFluor 594‐conjugated secondary antibodies (Huabio, Hangzhou, Zhejiang, China). Nuclei were stained with DAPI. The cell membrane was labelled with iFluor 488 phalloidin (Yeasen, Shanghai, China). Cells were incubated with 10 mM pHrodo Green AM (Thermo) for 30 min at room temperature for pH‐sensitive fluorescence detection. Fluorescence was finally detected with an Olympus FV3000 laser confocal microscope (Olympus). Images were analysed with ImageJ software (NIH).

2.10. Total internal reflection fluorescence microscopy

4T1 cells were transfected with the mCherry‐CD63‐HA plasmid for 36 h, and the cells were incubated in PBS before analysis by TIRF microscopy. For TIRF microscopy with an Olympus IX83 microscope (Olympus), the penetration depth (δ) of the evanescent field used to excite the fluorophore was set to 150 nm. Frames were acquired at 10 Hz in stacks of 400 images with an exposure time of 100 ms. Fluorescence image acquisitions were collected with CellSens software (Olympus). mCherry⁺ CD63 vesicles were quantified using ImageJ software (NIH).

2.11. Flow cytometric analysis

Cells were pretreated with the corresponding inhibitors to analyse phagocytosis, followed by incubation with 10 μg mL⁻¹ PKH26⁺ EVs for 1 h or 50 μg mL⁻¹ TRITC‐Dextran for 20 min. Then, labelled EVs were washed with a large volume of PBS to remove residual dyes by centrifugation at 100,000 × g for 1 h at 4°C thrice. To analyse EV amounts in cultured supernatants, EVs were incubated with 4‐μm anti‐CD63‐coated aldehyde sulphate latex beads (Thermo Fisher Scientific), then were washed and collected by centrifugation at 3500 × g for 5 min at 4°C, and then incubated with the corresponding fluorescence‐conjugated primary antibodies in the dark at 4°C for 20 min. Single‐cell suspensions prepared from the dLNs or blood were resuspended in PBS supplemented with 1% FBS and stained with the indicated fluorescent antibodies in the dark at 4°C for 20 min. The cells or beads were washed thrice with PBS and analysed using a FACS Athena flow cytometry system. The antibodies used are listed in Table S3, and the gating strategies are shown in Figure S9.

2.12. Isolation of RBC‐EVs, production of RBC‐EVs/Dox and RBC‐EVs/miR‐155‐ASO

For RBC‐EVs isolation, RBCs were separated from plasma by centrifugation at 500 × g for 10 min and passed through a leukodepletion filter (Terumo, Tokyo, Japan). Isolated RBCs were diluted in RPMI‐1640 medium and treated with 2 μM calcium ionophore (Merck) for 48 h at 37°C. Then, the RBCs and cell debris were removed by centrifugation at 600 × g for 20 min, 1600 × g for 15 min, 3260 × g for 15 min and 10,000 × g for 30 min (all steps at 4°C). RBC‐EVs were concentrated using ultracentrifugation with a SW32Ti rotor (L‐90K with an SW32Ti rotor, Beckman Coulter) at 100,000 × g for 70 min at 4°C. Subsequently, the EV pellets were resuspended in sterile PBS. For the production of RBC‐EVs/Dox or RBC‐EVs/miR‐155‐ASO, RBC‐EVs were coincubated with Dox or cholesterol‐modified miR‐155‐ASO at 37°C for 5 h and washed twice by centrifugation at 10,000 × g for 30 min at 4°C.

2.13. Lipidomic assay and analysis

4T1 cells treated with DMSO or Rabe were analysed for untargeted absolute quantitative lipidomic analysis. Lipids were extracted and analysed using an UHPLC Nexera LC‐30A ultra‐performance liquid chromatography system (SHIMADZU, Japan) coupled to Q‐ExactivePlus (Thermo Scientific) in Shanghai Applied Protein Technology. The lipidomic data are presented in Supplementary Table Lipidomic Data.

2.14. IBD model experiments

Acute IBD was induced by giving 2.5% DSS (MP Biomedicals, Solon, OH, USA; w v⁻¹) with mol wt. 36,000−50,000 in acidified drinking water for 7 days. The day mice started to drink the DSS solution was regarded as day 0. To analyse the Rabe or Ome effects on IBD, mice received intraperitoneal injections of DMSO, 5 mg kg⁻¹ Ome or Rabe every 2 days from day −1. Acute IBD was induced in Rab27a^fl/fl × Villin‐Cre mice to analyse IEC‐EVs’ effects. In some experiments, mice received intraperitoneal injections of DMSO or Rabe with 30 mg kg⁻¹ TGF‐βRI inhibitor SB525334 or 100 μg IEC‐EV combinatorial intravenous injection every 2 days from day −1.

2.15. Tumour model experiments

To observe the effect of Rabe on EVs endocytosis in vivo, BALB/c mice were injected subcutaneously with 1 × 10⁶ RFP⁺ 4T1 cells on day 0, and then mice received an intratumoral injection of 5 μL Rabe (10 mM) and an intravenous injection of 10 μg PKH67⁺ EVs, DSFCs were installed on these mice for MP‐IVM analysis. To analyse the effect of ATP6V1A on EV endocytosis in vivo, 1 × 10⁶ 4T1 and Atp6v1a ^−/− 4T1 cells were injected subcutaneously in each flank of the same mouse (BALB/c mice) on day 0. Then, these mice received an intravenous injection of 10 μg VivoTrack 680‐labeled 4T1‐EVs on day 9, and the distribution of EVs was detected using an in vivo imaging system. To analyse the effect of ATP6V1A on Rabe‐induced EV endocytosis in vivo, 1 × 10⁶ 4T1 or Atp6v1a ^−/− 4T1 cells were injected subcutaneously in both flanks of the same mouse (BALB/c mice) on day 0. Then, one flank tumour received an intratumoral injection of 5 μL Rabe (10 mM) and an intravenous injection of 10 μg VivoTrack 680‐labeled 4T1‐EVs on day 9, and the distribution of EVs was detected using an in vivo imaging system. To analyse the effect of ATP6V1A on tumour growth, 1 × 10⁶ 4T1 and Atp6v1a ^−/− 4T1 cells were injected subcutaneously in each flank of the same mouse (BALB/c mice or Pd1 ^−/− mice) on day 0, and tumour size was monitored every other day by measurement with dial callipers from day 9. To analyse the effect of Rabe on tumour growth, 1 × 10⁶ 4T1 or Atp6v1a ^−/− 4T1 cells were injected subcutaneously in both flanks of the same mouse (BALB/c mice or Pd1 ^−/− mice) on day 0, 1 × 10⁶ 4T1 cells were injected subcutaneously in both flanks of the same mouse (nude mice) on day 0. Then, these mice were treated with Rabe from day 9 every 2 days, and tumour size was monitored every other day by measurement with dial callipers from day 9. To deplete CD8⁺ T cells in vivo, 1 × 10⁶ 4T1 cells were injected subcutaneously in both flanks of the same mouse (BALB/c mice) on day 0, 100 μg anti‐mouse CD8‐InVivo antibodies were injected intraperitoneally from day 3 every 2 days, these mice were later treated with Rabe from day 9 every 2 days, and tumour size was monitored every other day by measurement with dial callipers from day 9. To analyse the role of Rabe in the antitumor effects of anti‐PD‐1 therapy, BALB/c mice were injected subcutaneously with 1 × 10⁶ 4T1, and C57BL/6J mice were injected subcutaneously with 1 × 10⁶ B16F10 on day 0. Then, these mice received an intraperitoneal injection with isotype control antibodies or 50 μg anti‐PD‐1 together with or without intravenous injection of 5 mg kg⁻¹ Rabe from day 11 every 2 days, and tumour size was monitored every other day by measurement with dial callipers from day 11. To analyse the role of Rabe in the antitumor effects of RBC‐EVs/Dox, 1 × 10⁶ 4T1 or Atp6v1a ^−/− 4T1 cells were injected subcutaneously in both flanks of the same mouse (Nude mice) on day 0. Then, these mice received an intravenous injection of RBC‐EVs/Dox together with or without an intratumoral injection of Rabe from day 9 every 2 days, and tumour size was monitored every other day by measurement with dial callipers from day 9.

2.16. Rosa‐LSL‐tdtomato mice experiments

To analyse tdTomato expression in colons, Rosa‐LSL‐tdtomato mice received an intravenous injection of 10 μg IEC‐EVs or IEC‐EVs/Cre combined with DMSO or 5 mg kg⁻¹ Rabe, and colons were collected for immunofluorescence staining and confocal microscopy 24 h later. To analyse tdTomato expression in livers, Rosa‐LSL‐tdtomato mice bearing 4T1 or Cre⁺ 4T1 tumour received an intratumoral injection of DMSO or 5 μL Rabe (10 mM) for consecutive 5 days. Livers were then collected for immunofluorescence staining and confocal microscopy.

2.17. Acute liver failure (ALF) model therapeutic experiments

ALF model was introduced into mice by intraperitoneally injecting with 50 μg kg⁻¹ Escherichia coli 0111:B lipopolysaccharide (LPS, Sigma‐Aldrich) and 400 mg kg⁻¹ D‐galactosamine (D‐GaIN, Sigma‐Aldrich) on day 0. Then, the mice were intravenously injected with 50 μg RBC‐EVs/miR‐NC‐ASO, 50 μg RBC‐EVs/miR‐NC‐ASO and 5 mg kg⁻¹ Rabe, 50 μg RBC‐EVs/miR‐155‐ASO, or 50 μg RBC‐EVs/miR‐155‐ASO and 5 mg kg⁻¹ Rabe on days −3, −2 and −1. The mice were sacrificed 12 h after treatment.

2.18. Plasmids

The cDNA sequences were cloned by PCR from the cDNA of PEMs and inserted into the PCDNA3.1 backbone: ATP6V0A2 and ATP6V0A4. pcDNA3.1‐CMV‐CFP; UBC‐Cre25nt and pLV‐CMV‐LoxP‐DsRed‐LoxP‐eGFP plasmids were purchased from Addgene (Watertown, MA, USA). All constructs described above were verified by sequencing.

2.19. Transfection of siRNA

Transient siRNA transfection was performed in vitro using TransIT‐TKO (Mirus Bio, Madison, Wisconsin, USA) according to the manufacturer's instructions. The cells were used for experiments after 48‐h transfection. The siRNAs used are listed in Table S3.

2.20. Real‐time PCR analysis

According to the manufacturer's instructions, total RNAs were isolated from cells or tissues using TRIzol reagent (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Specific primers of miR‐155 and U6 (GenePharma, Shanghai, China) were used to reverse the transcription of mature miRNAs. Single‐stranded cDNA was synthesised using a cDNA Synthesis Kit (Vazyme, Nanjing, Jiangsu, China) for reverse transcription of mRNAs. Real‐time PCR was performed with the CFX Touch system Bio‐Rad) using SYBR Green Master Mix (Vazyme). The primers used are listed in Table S3.

2.21. Nuclear and cytoplasmic protein extraction

Nuclear and cytoplasmic protein extraction was performed using a Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime, Shanghai, China). Briefly, 2 × 10⁶ cells were resuspended in 200 μL of cytoplasmic protein isolation solution A and homogenised on ice. Next, 10 μL cytoplasmic protein isolation solution B was added, and the cells were homogenised on ice. The homogenate was centrifuged at 10,000 × g for 5 min at 4 °C. The resulting supernatant was the cytoplasmic protein fraction. The pellet was resuspended in 50 μL nuclear protein isolation solution, homogenised on ice and centrifuged at 10,000 × g for 10 min. The resulting supernatant was the nuclear protein fraction.

2.22. In vitro flag pull‐down assay

A flag pull‐down assay was employed to identify the interactions between Flag‐ATP6V1A, HA‐ATP6V0A2 and HA‐ATP6V0A4. The pull‐down proteins were mixed in the reaction buffer and incubated in a rotating incubator at 4°C overnight. The proteins were pulled down with anti‐Flag M2 beads on the rotating incubator at 4°C for 3 h. After removing the supernatant, the beads were washed with the reaction buffer four times. Immunoprecipitates were eluted by boiling with 1% (w v⁻¹) SDS sample buffer, followed by western blot with anti‐Flag or anti‐HA antibodies.

2.23. Standard curve for intracellular pH analysis

4T1 cells were cultured in a 96‐well plate with 10 mM pHrodo Green AM (Thermo) for 30 min at room temperature. The intracellular pH was clamped with extracellular buffer at pH 4.5, 5.5, 6.5 and 7.5 using the Intracellular pH Calibration Buffer Kit (Thermo). Relative fluorescence units were measured with Thermo Scientific Varioskan® Flash. The pH standard curve was obtained by plotting the average of 3 data points and fitting a linear trend line.

2.24. Western blot analysis

Cell lysates or immunoprecipitates were separated by SDS–PAGE and transferred to PVDF membranes (Millipore, Massachusetts, USA). After blocking with 5% BSA, the membranes were incubated with primary antibodies at 4°C overnight and were then incubated with the corresponding secondary antibodies for 1 h at RT. The membranes were washed three times for 10 min each, incubated with SuperSignal Chemiluminescent Substrate (Pierce, Dallas, Texas, USA) and scanned with a Tanon 4500 Gel Imaging System (Tanon, Shanghai, China). The antibodies are listed in Table S3.

2.25. Co‐immunoprecipitation

Cells were lysed in a co‐immunoprecipitation buffer containing protease inhibitor cocktail tablets (Roche). After centrifugation for 10 min at 12,000 × g and 4°C, the supernatants were collected and incubated with protein A/G magnetic beads (Thermo Fisher Scientific) and the corresponding antibodies at 4°C overnight. The next day, the beads were washed three times with IP buffer. Immunoprecipitates were eluted by boiling with 1% (w v⁻¹) SDS sample buffer.

2.26. Histological analysis

Murine colon and liver were dissected, fixed in 4% paraformaldehyde, embedded in paraffin, sectioned, and stained with H&E solution. Murine colon and tumour were dissected, fixed in Tissue‐Tek® O.C.T. Compound (SAKURA, USA), frozen sectioned, and stained with specific primary antibodies at 4°C overnight. Primary antibodies were detected using iFluor 488‐ or iFluor 594‐conjugated secondary antibodies (Huabio). Nuclei were stained with DAPI and examined by Olympus FV3000 laser confocal microscope (Olympus). Images were analysed with ImageJ software.

2.27. Histological score

The degree of colitis was scored without any prior knowledge of the experimental groups as follows: inflammation (scale of 0–3), lesion depth (scale of 0–3), crypt damage (scale of 0–4), degree of epithelial regeneration (scale of 0–4), percentage of the pathological area (scale of 0–4). The total score ranged from 0 points (no colitis) to 18 points (severe colitis).

2.28. Lentivirus production and stable cell line generation

For ATP6V1A knockout, 4T1 cells were cotransfected with plv5‐Cas9‐Blast (Merck) and U6‐gRNA:hPGK‐puro‐2A‐tBFP (Merck) containing the specific gRNA sequences. After 24 h, single BFP⁺ cells were sorted into 96‐well plates with a Beckman MoFlo Astrios EQ cell sorter (Beckman). The levels of endogenous proteins were then verified by western blot analysis. For RFP overexpression, 4T1 cells were transfected with RFP expression plasmids, and RFP⁺ cells were selected under 2.5 μg mL⁻¹ puromycin media.

2.29. Measurement of cytokines

ELISA was used to measure the levels of murine serum IL‐1β, IL‐6, TNF (BioLegend, San Diego, California, USA), AST and ALT (Nanjing Jiancheng, Nanjing, Jiangsu, China) according to the manufacturer's instructions.

2.30. Statistical analysis

Statistical differences were compared by students’ t‐tests between two groups and one‐way ANOVA followed by Tukey tests among multiple groups. A log‐rank test was used for survival rate analysis. All data are expressed as the mean ± SD values and were analysed using GraphPad Prism 8.0 (GraphPad Software Inc., San Diego, CA, USA). Differences with p < 0.05 were defined as significant.

3. RESULTS

3.1. PPIs promote EV uptake by various types of cells

PPI Ome has been demonstrated to reduce EV release (Chalmin et al., 2010). Besides Ome, we also confirmed that PPI Rabe reduced EV levels in supernatant from murine 4T1 breast cancer cells (4T1‐EVs), and Rabe had a more potent suppressive effect on EV levels than Ome (Figure 1a,b and Figure S1a–c), which is relevant to its higher inhibitory potential in PPs (Horn, 2000). Besides the decrease in exosome markers, Annexin A1, a specific marker of microvesicles (Jeppesen et al., 2019), was also reduced by Rabe. However, Annexin A1 was less enriched in the EVs relative to cells than exosome markers (Figure S1c). Therefore, these EVs are mainly exosomes mixed with a few microvesicles. Then, 10 μM Rabe was used in the following experiments because it possessed the optimal EV inhibitory effect without affecting cell survival (Figure S1d, e). To elucidate how PPIs regulate EV levels, we first examined the effect of PPIs on MVBs and found that Rabe did not affect the number of CD63⁺ or hepatocyte growth factor‐regulated tyrosine kinase substrate (HRS)⁺ MVBs in 4T1 cells (Figure 1c and Figure S1f). The MVB numbers are collectively determined by the biogenesis and elimination including plasma membrane (PM) fusion and lysosomal degradation of MVBs. We found that Rabe did not affect MVB biogenesis, as evidenced by the observation of similar numbers of EEA1⁺ early endosomes (EEs) in Rabe‐treated 4T1 cells (Figure S1g). TIRF microscopy showed comparable numbers of CD63⁺ MVBs in the subplasmalemmal region of 4T1 cells with or without Rabe treatment (Figure S1h), suggesting that PPIs do not affect the fusion of MVBs and PM. Rabe also did not affect the colocalization of MVBs and LAMP1⁺ lysosomes (Figure S1i), indicating the unchanged degradation of MVBs. In addition, PPIs did not affect the intraluminal vesicle (ILV) budding of MVBs when visualised by overexpression of constitutively active Rab5^Q79L mutant (Figure S1j). These results imply that PPIs are unlikely to regulate EV biogenesis.

FIGURE 1.

PPIs promote EV uptake by various types of cells. (a) Flow cytometric analysis of CD9 percentage on EVs captured by anti‐CD63‐coated latex beads from the supernatants of 4T1 cells treated with DMSO, Ome or Rabe. (b) EVs were isolated from equal numbers of 4T1 cells treated with DMSO, Ome or Rabe and then the protein amount of these EVs was determined by BCA assay. (c) Confocal microscopy analysis of CD63⁺ MVBs in 4T1 cells treated with DMSO or Rabe. Scale bar, 10 μm. (d) Flow cytometric analysis of PKH26‐labeled EVs taken by 4T1 cells treated with DMSO, Ome or Rabe. (e) Real‐time fluorescence microscopy analysis of the uptake kinetics of PKH26‐labeled EVs by PEMs treated with DMSO or Rabe. Scale bar, 10 μm. (f) Confocal microscopy analysis of DsRed and eGFP in HEK293T reporter cells with DMSO or Rabe pretreatment, followed by coculture without or with Cre⁺ HEK293T cells. Scale bar, 10 μm. (g) Flow cytometric analysis of CD9 percentage on EVs captured by anti‐CD63‐coated latex beads from the supernatants of 4T1 cells with Rabe alone or combined with Cytochalasin D treatment. (h) Flow cytometric analysis of PKH26‐labeled EVs taken by the indicated cells treated with DMSO, Ome or Rabe. Representative results from three independent experiments are shown. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001 (one‐way ANOVA followed by Tukey test except for unpaired two‐tailed Student's t‐test in c, e and g; mean ± SD).

We then investigated whether the PPI‐mediated decrease in EVs is due to elevated EV uptake by 4T1 cells. Actually, PPIs significantly promoted 4T1 cells to take up PKH26‐labeled EVs rather than EV‐free PKH26 dye (Figure 1d and Figure S1k,l). Real‐time fluorescence microscopy also validated that Rabe induced increased EV uptake by PEMs (Figure 1e and Movie 1–2). Next, we developed a Cre‐Loxp system to monitor the effect of PPIs on EV uptake by HEK293T cells (Figure S1m). After Cre⁺ HEK293T cell‐derived EV uptake, a red‐to‐green colour switch can be detected in HEK293T reporter cells (Mohr et al., 2015). By this assay, we found that more EGFP⁺ reporter cells with pretreatment of Rabe were induced by Cre⁺ HEK293T cells (Figure 1f), indicating the increased uptake of EVs from Cre⁺ cells. Subsequently, we found that Rabe no longer reduced EV levels in the 4T1 cell supernatant when EV uptake was inhibited by Cytochalasin D (Figure 1g). As an extension, we confirmed that PPIs could also promote EV uptake by a variety of cells, including murine RAW264.7 macrophages, BMDCs and MC38 colon cancer cells; human THP‐1 monocyte leukaemia cells, A549 lung cancer cells and HeLa cervical cancer cells (Figure 1h). Altogether, these results suggest that PPIs promote EV uptake rather than inhibit EV secretion by various cells.

3.2. PPIs promote EV uptake by increasing v‐ATPase‐dependent macropinocytosis

We then explore how PPIs regulate EV uptake. First, we verified that Rabe did not affect the attachment and membrane fusion of EVs and 4T1 cells (Figure S2a,b). Thus, we dissected the role of PPIs in EV endocytosis and found that neither caveolae‐mediated endocytosis inhibitor indomethacin nor the clathrin‐mediated endocytosis inhibitor chlorpromazine affected EV uptake induced by Rabe (Figure S2c,d). Moreover, we found that inhibitors of dynamin (dynasore), CDC42 (CDC42‐IN‐1) and ARF6 (NAV‐2729) hardly affected Rabe‐induced EV uptake by 4T1 cells (Figure S2e), suggesting that RhoA‐, CDC42‐ and ARF6‐regulated endocytosis (Mayor & Pagano, 2007) were minimally affected by Rabe, either. However, the macropinocytic inhibitors LY294002 or amiloride completely abolished the Rabe‐induced increase in EV uptake (Figure 2a). Correspondingly, Rabe did not reduce microvesicles from LY294002‐treated 4T1 cells (Figure S2f). Moreover, enhanced macropinocytosis of TRITC‐Dextran by Rabe‐treated PEMs was also eliminated by LY294002 (Figure 2b and Movie 3–6). V‐ATPases are essential for macropinocytosis. We found that Rabe no longer promoted EV uptake by 4T1 cells with ATP6V1A, a V₁A subunit of v‐ATPases, knockout (Atp6v1a^−/− 4T1‐1) (Figure 2c and Figure S2g). Similar results could be obtained in v‐ATPase inhibitor KM91104‐ or Bafilomycin A1 (Baf‐A1)‐treated 4T1 cells (Figure 2d). We also observed more PM localization of ATP6V1A in Rabe‐treated 4T1 cells (Figure 2e,f). Increased membrane‐associated ATP6V1A and decreased cytoplasmic ATP6V1A induced by Rabe were further confirmed by western blotting (Figure 2g). Then, we compared the effect of different endocytosis pathways on EV uptake. We investigated this by using cells with different phagocytic activity, including PEMs (high), 4T1 cells (medium) and T cells (low). Although inhibition of clathrin‐ and macropinocytosis‐mediated endocytosis similarly reduced EV uptake by PEMs, inhibition of macropinocytosis dominantly suppressed EV uptake by all cells with the inhibitory percentages of 18.57 ± 1.141, 51.16 ± 2.921 and 37.95 ± 6.991 (mean ± SD) in PEMs, 4T1 cells and T cells, respectively (Figure 2h). Thus, v‐ATPase‐mediated macropinocytosis, which PPIs can induce, is crucial for EV uptake by recipient cells.

FIGURE 2.

PPIs promote EV uptake by increasing v‐ATPase‐dependent macropinocytosis. (a) Flow cytometric analysis of PKH26‐labeled EVs taken by 4T1 cells with Rabe alone or combined with LY294002 or amiloride treatment. (b) Real‐time fluorescence microscopy analysis of the uptake kinetics of TRITC‐Dextran by PEMs treated with Rabe alone or combined with LY294002. Scale bar, 10 μm. (c) Flow cytometric analysis of PKH26‐labeled EVs taken by DMSO‐ or Rabe‐treated 4T1 or Atp6v1a^−/− 4T1‐1 cells. (d) Flow cytometric analysis of PKH26‐labeled EVs taken by 4T1 cells with Rabe alone or combined KM91104 or Baf‐A1 treatment. (e) Flow cytometric analysis of ATP6V1A on PM of 4T1 cells treated with DMSO or Rabe. (f) Confocal microscopy analysis of the ATP6V1A localization in Phalloidin‐labelled 4T1 cells treated with DMSO or Rabe. Scale bar, 10 μm. (g) Western blotting analysis of ATP6V1A proteins in cytoplasm and membrane of 4T1 cells treated with DMSO or Rabe, Na⁺/K⁺ ATPase used as a membrane marker. (h) Flow cytometric analysis of PKH26‐labeled EVs taken by the indicated cells treated with DMSO, Baf‐A1, indomethacin or chlorpromazine. (i and j) Flow cytometric analysis of PKH26‐labeled EVs taken by 4T1 cells with Rabe alone or combined with MeβCD (i) or azathioprine (j) treatment. Representative results from three independent experiments are shown. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001 (unpaired two‐tailed Student's t‐test except for one‐way ANOVA followed by Tukey test in c and h; mean ± SD).

PM translocation of v‐ATPases will increase PM trafficking of cholesterol and subsequent PM‐associated RAC1, which are prerequisites for macropinocytosis (Ramirez et al., 2019). We found that Rabe indeed increased the PM distribution of cholesterol (Figure S2h). Cholesterol is rich in the lipid raft (RF) (Ishitsuka et al., 2005). When detecting RF marker Flotillin‐1, we found that Rabe seemed to facilitate RF formation and Rabe‐induced PM accumulation of ATP6V1A was inclined to colocalise with RF (Figure S2i). MeβCD can extract cholesterol from the plasma membrane (Han et al., 1999). We discovered that MeβCD treatment abolished Rabe to induce EV endocytosis (Figure 2i). We also detected increased PM localization of RAC1 (Figure S2j). In addition, Rabe‐induced EV endocytosis is eliminated by RAC1 inhibitor azathioprine (Figure 2j). Altogether, increased PM v‐ATPases are a crucial step for PPI‐induced macropinocytosis.

3.3. PPIs increase PM v‐ATPase assembly by decreasing cytosolic pH

We next sought to identify the upstream that mediates PM accumulation of v‐ATPases. PKA has been demonstrated to be required in this process (Ramirez et al., 2019). However, we did not detect the activation of PKA by Rabe (Figure S3a). In addition, PKA inhibitor H89 did not eliminate Rabe‐induced EV uptake by 4T1 cells (Figure S3b). Bicarbonate transporter SLC4A7 is indispensable for PKA‐induced macropinocytosis (Ramirez et al., 2019). Due to low knockdown efficacy in 4T1 cells, we performed these experiments in HeLa cells and found that Rabe still induced EV uptake by SLC4A7‐silenced HeLa cells (Figure S3c,d). PPIs inhibit the transport of protons from cytoplasm to extracellular space, thereby reducing cytosolic pH. We did detect reduced cytosolic pH in Rabe‐treated 4T1 cells by using the pH‐sensitive fluorescent pHrodo Green AM (Figure 3a). Based on the standard curve (Figure S3e), we confirmed that Rabe treatment declined cytosolic pH from 7.4 to 6.7. These results motivated us to characterise the role of pH in Rabe‐induced EV endocytosis. First, we confirmed that low pH (6.4) media decreased cytosolic pH (Figure S3f), induced ATP6V1A PM accumulation and increased EV uptake by 4T1 cells (Figure 3b,c). Next, 4T1 cells were first treated with cholesterol to induce ATP6V1A PM accumulation and then treated with high pH (7.6) media. As a result, high pH (7.6) media increased cytosolic pH (Figure S3g), reduced ATP6V1A PM accumulation and decreased EV uptake by 4T1 cells (Figure 3d,e). Furthermore, protonophore 2, 4‐dinitrophenol (2, 4‐DNP)‐induced equal pH gradients across biological membranes (Dechant et al., 2010; Nilsson, 1995) eliminated Rabe‐mediated increase in ATP6V1A PM accumulation and EV uptake (Figure 3f,g). These results indicate that Rabe induces EV endocytosis in a pH‐dependent manner.

FIGURE 3.

PPIs increase PM v‐ATPase assembly by decreasing cytosolic pH. (a) Confocal microscopy analysis of pH value in 4T1 cells treated with DMSO or Rabe indicated by pH‐sensitive fluorescent probe pHrodo Green AM. Scale bar, 10 μm. (b–e) Confocal microscopy analysis of ATP6V1A localization in Phalloidin‐labeled 4T1 cells (b and d), and flow cytometric analysis of PKH26‐labeled EVs taken by 4T1 cells (c and e), cultured in normal pH (7.2) and low pH (6.4) media (b and c), or cultured in normal pH (7.2) and high pH (7.6) media in the presence of cholesterol (d and e). Scale bar, 10 μm. (f) Confocal microscopy analysis and flow cytometric analysis of ATP6V1A localization in Phalloidin‐labeled 4T1 cells with Rabe alone or combined with 2,4‐DNP treatment. Scale bar, 10 μm. (g) Flow cytometric analysis of PKH26‐labeled EVs taken by 4T1 cells with Rabe alone or combined with 2,4‐DNP treatment. (h and i) Confocal microscopy analysis of ATP6V1A and LAMP1⁺ lysosome colocalization in 4T1 cells (h) or ATP6V1A localization in Phalloidin‐labeled 4T1 cells (i) treated with DMSO or Rabe, cultured in high pH (7.6) (h) or normal pH (7.2) (i) media. Scale bar, 10 μm. (j) Confocal microscopy analysis of ATP6V1A and LAMP1⁺ lysosome colocalization in HeLa cells (left) or ATP6V1A localization in Phalloidin‐labeled HeLa cells (right) transfected without or with ATP6V0A2 plasmids, treated without (left) or with (right) cholesterol. Scale bar, 10 μm. (k) Confocal microscopy analysis of ATP6V1A localization in Phalloidin‐labeled HeLa cells (left) or ATP6V1A and LAMP1⁺ lysosome colocalization in HeLa cells (right) transfected without or with ATP6V0A4 plasmids, cultured in normal pH (7.2) (left) or high pH (7.6) (right) media. Scale bar, 10 μm. (l) Affinity precipitation of recombinant HA‐ATP6V0A2 and HA‐ATP6V0A4 proteins with Flag‐ATP6V1A protein under different pH conditions, captured by anti‐Flag M2 magnetic beads and analyzed by western blotting. Representative results from three independent experiments are shown. ns, not significant; **p < 0.01; ***p < 0.001 (unpaired two‐tailed Student's t‐test; mean ± SD).

PPIs are pro‐drugs and need to be activated in an acidic environment (Lugini et al., 2016; Milito et al., 2010). However, we did detect declined cytosolic pH induced by PPIs in a pH 7.2 medium, which prompted us to speculate that PPIs probably had weak activity in a pH 7.2 medium. To test this, we treated 4T1 cells with PPIs in a pH 6.6 or pH 7.9 medium, respectively. In a pH 7.2 medium, PPIs reduced cytosolic pH from 7.4 to 6.7; in a pH 6.6 medium, PPI‐mediated cytosolic pH reduction was more pronounced, from 6.9 to 6.3, accompanied by more EV uptake by 4T1 cells. However, PPIs maintained cytosolic pH at 7.6 and could not affect EV uptake of 4T1 cells in a pH 7.9 medium (Figure S3h). Furthermore, comparable cell viability was observed in different pH media within experimental time (Figure S3i). Thus, PPIs have activity in a pH 7.2 medium to a certain degree.

Then, we explored the mechanisms responsible for pH‐mediated PM accumulation of v‐ATPases and found that Rabe did not affect the protein and mRNA levels of ATP6V1A (Figure S3j,k). ATP6V1A is recruited to the membrane by binding to the V₀ subunit. The Mammal V ₀ subunit has four isoforms (a1–a4) that target different biological membranes. a2 presents in endosomes/vacuoles, while a3 and a4 contribute to PM targeting of v‐ATPases (Vasanthakumar & Rubinstein, 2020). A decrease in cytosolic pH results in the disassembly of vacuolar v‐ATPases but leaves the vacuolar membrane (VM) localization of the V₀ subunit unchanged (Dechant et al., 2010). We found that Rabe notably decreased LAMP1⁺ lysosome (indicating VM) distribution (Figure 3h) but increased PM distribution of ATP6V1A (Figure 3i). Therefore, we supposed that V₀ of PM and VM competitively bind ATP6V1A and reduced VM localization of ATP6V1A may enhance the interaction of ATP6V1A and PM V₀. We then tested this hypothesis by ATP6V0A2 overexpression. Due to low overexpression efficacy in 4T1 cells, we performed these experiments in HeLa cells instead and found that ATP6V0A2 overexpression notably increased VM localization of ATP6V1A along with decreased PM localization of ATP6V1A in cholesterol‐treated HeLa cells (Figure 3j and Figure S3l). In contrast, ATP6V0A4 overexpression caused increased PM and declining VM localization of ATP6V1A in HeLa cells cultured in high pH (7.6) media (Figure 3k and Figure S3l). We obtained opposite results in ATP6V0A2‐ or ATP6V0A4‐silenced HeLa cells (Figure S3m,n–q). We also excluded Rabe affecting the protein expression of ATP6V0A2 and ATP6V0A4 (Figure S3r). Besides, we confirmed that ATP6V0A2 overexpression reduced ATP6V0A4 pulled down by ATP6V1A and vice versa (Figure S3s). More importantly, we found that ATP6V1A pulled down more ATP6V0A4 and less ATP6V0A2 in low pH (6.4) media than in high pH (7.6) media (Figure 3l). Collectively, we concluded that PPI‐mediated reduced cytosolic pH promotes the disassembly of ATP6V1A from VM, thereby enhancing the assembly of PM v‐ATPases.

Lipids are involved in membrane fusion and macropinocytosis (Parolini et al., 2009; Ramirez et al., 2019). Low pH induces EV and cell fusion, likely by altering lipid composition (Parolini et al., 2009). Because EVs for 4T1 cell uptake were pre‐isolated, PPI‐mediated increased EV uptake is not due to altered EV lipid composition. Low pH causes membrane phospholipid fatty acid remodelling of tumour cells (Urbanelli et al., 2020). Therefore, we performed lipidomics and found that Rabe did not alter the major components of lipid class total lipid level (Figure S3t). However, nine subclasses of lipids showed certain degrees of increase or decrease in content (Figure S3u). To analyse the lipidomic data further, we screened 37 lipid species with significant differences after Rabe treatment using p < 0.05 and VIP > 1 as screening criteria. In this way, we found that sphingomyelin (SM) showed a substantial increase among these lipids after Rabe treatment (Figure S3v). SM has been reported to enhance phagocytosis (Bryan et al., 2021; Lauer et al., 2000). Thus, SM may be involved in Rabe‐mediated increased phagocytosis of EVs.

3.4. PPIs aggravate IBD by accelerating the clearance of IEC‐EVs

We previously found that TGF‐β1‐containing IEC‐EVs maintained immune balance through the TGF‐β receptor (TGF‐βR) signalling pathway (Jiang et al., 2016). Therefore, we assumed that accelerated IEC‐EV clearance would inhibit IEC‐EV TGF‐β1 and TGF‐βR binding, prevent TGF‐βR signalling activation, and cause abnormal activation of intestinal immunity, which probably worsens IBD. We found that either Rabe or Ome treatments decreased colonic EV and A33⁺ IEC‐EV levels (Figure 4a,b). Consistently, Rabe and Ome accelerated the development of DSS‐induced IBD, as indicated by the increased body weight loss, shorter colonic length and more severe histological damage (Figure 4c–e). To characterise the IEC‐EV role in this process, we introduced IBD into Rab27a^fl/fl × Villin‐Cre mice, in which the secretion of A33⁺ IEC‐EVs but not DC‐derived CD11c⁺ EVs was specifically suppressed (Figure S4a). In these mice, we found that Rabe hardly involved IBD development (Figure 4f–h). We also found that Rabe no longer affected the severity of IBD in mice treated with TGF‐βRI inhibitor SB525334 (Figure S4b–d). To elucidate that Rabe affects IBD progression by regulating IEC‐EV uptake, we transferred IEC‐EVs from Cmv‐cre mice (IEC‐EVs/Cre) into Rosa‐LSL‐tdtomato mice (Figure S4e). After the uptake of Cre⁺ EVs by cells of these mice, the cells will express tdTomato (Zomer et al., 2015). We found that IEC‐EVs/Cre could be taken up by intestinal CD45⁺ and CD45⁻ cells, which was significantly enhanced by Rabe (Figure 4i). In addition, the inhibitory effects on IBD progression achieved by the transfer of IEC‐EVs/Cre significantly inhibited IBD development were eliminated in Rabe‐ or Ome‐treated mice (Figure 4j–l). Therefore, these results demonstrate that PPIs promote uptake‐mediated clearance of IEC‐EVs, thereby aggravating IBD symptoms.

FIGURE 4.

PPIs aggravate IBD by accelerating the clearance of IEC‐EVs. (a and b) Mice received intraperitoneal injection of DMSO, Ome or Rabe, and then EVs were isolated from the colon of these mice. The protein amount of these EVs was determined by BCA assay (a). Flow cytometric analysis of A33 percentage on EVs captured by anti‐CD63‐coated latex beads from the cultured supernatants of colons from those mice (b). (c–e) Mice were administered with 2.5% DSS for 7 days to induce acute colitis and simultaneously received intraperitoneal injections of DMSO, Ome or Rabe every 2 days from day −1. Body weight changes (c), colonic length (d), and hematoxylin and eosin (H&E) stained colonic tissue sections (e) of these mice. Scale bar, 200 μm. (f–h) Villin‐Cre Rab27a^f/f mice were administered with 2.5% DSS for 7 days to induce acute colitis and simultaneously received intraperitoneal injections of DMSO or Rabe every 2 days from day −1. Body weight changes (f), colonic length (g) and H&E stained colonic tissue sections (h) of these mice. Scale bar, 200 μm. (i) Representative immunofluorescence of CD45 staining and tdTomato expression in colons from Rosa‐LSL‐tdtomato mice received an intravenous injection of IEC‐EVs or IEC‐EVs/Cre. Scale bar, 50 μm. (j–l) Mice were administered with 2.5% DSS for 7 days to induce acute colitis, accompanied by intravenous injection with IEC‐EVs, intraperitoneal injection with Ome or Rabe, or corresponding combination injection every 2 days from day −1. Body weight (j), colonic length (k) and H&E stained colonic tissue sections (l) of these mice. Scale bar, 200 μm. Representative results from three independent experiments are shown. Ctrl and PBS indicate mice with plain water or 2.5% DSS drinking, respectively. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001 (one‐way ANOVA followed by Tukey test; mean ± SD).

3.5. PPIs improve antitumor immunity by promoting TEV clearance

TEVs play critical roles in tumour immunosuppression (Chen et al., 2018; Poggio et al., 2019). We assumed that the PPI‐mediated decrease in TEVs by enhanced endocytosis would promote antitumor immune responses. To test this, we first intravitally imaged the endocytosis of TEVs by tumour cells and assessed the effect of Rabe on this process. We subcutaneously inoculated RFP⁺ 4T1 cells into mouse backs and injected these mice with Rabe intratumorally. Then, dorsal skinfold chambers (DSFCs) were installed on these mice for multiphoton intravital microscopy (MP‐IVM) analysis. Rabe led to faster EV uptake (Figure 5a and Movie 7–8), accelerated the arrival of the uptake platform, and enhanced the uptake intensity in the platform (Figure 5b,c and Movie 9–12). Then, we established another ATP6V1A‐knockout 4T1 cell line (Atp6v1a^−/− 4T1‐2) (Figure S2g) and subcutaneously inoculated 4T1 or both ATP6V1A‐deficient 4T1 tumour in each flank of mice, respectively. Then, these mice were intravenously injected with VivoTrack 680‐labeled 4T1‐EVs. We found that the uptake of 4T1‐EVs by both ATP6V1A‐deficient 4T1 was significantly inhibited (Figure S5a). Next, both flanks of mice were subcutaneously inoculated with 4T1 or both ATP6V1A‐deficient 4T1 tumour, followed by Rabe intratumoral injection in one flank tumour. After intravenous injection with VivoTrack 680‐labeled 4T1‐EVs, we found that the 4T1 tumour with Rabe injection endocytosed more 4T1‐EVs than that without Rabe injection, which could not be detected in both ATP6V1A‐deficient 4T1 tumour mice (Figure 5d). These results suggest that Rabe promotes TEV uptake by tumour via macropinocytosis.

FIGURE 5.

PPIs improve antitumor immunity by promoting TEV clearance. (a–c) Mice received a subcutaneous injection of RFP⁺ 4T1 tumour in both flanks and one flank tumour received an intratumoral injection of Rabe. Then, these mice were intravenously injected with PKH67‐labeled 4T1‐EVs and DSFCs were used for MP‐IVM analysis at indicated times. Scale bar, 10 μm. (d) Mice received a subcutaneous injection of 4T1, Atp6v1a^−/− 4T1‐1 or Atp6v1a^−/− 4T1‐2 cells in both flanks and one flank tumour received an intratumoral injection of Rabe. Then, these mice were intravenously injected with VivoTrack 680‐labeled 4T1‐EVs. The distribution of EVs in the indicated organs was detected using an in vivo imaging system. (e–h) Mice received a subcutaneous injection of 4T1, Atp6v1a^−/− 4T1‐1 (e, f) or Atp6v1a^−/− 4T1‐2 (e) cells in each flank respectively, or 4T1 tumour‐bearing mice in both flanks received an intratumoral injection of DMSO or Rabe from day 9 every 2 days (g and h). Tumour sizes (e and g), flow cytometric analysis of the percentages of PD‐1⁺, IFN‐γ⁺ and Ki‐67⁺ T cells among the CD45⁺CD8⁺ T cell populations in dLNs (f and h) of these mice. Representative results from three independent experiments are shown. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001 (unpaired two‐tailed Student's t‐test except for one‐way ANOVA followed by Tukey test in d; mean ± SD).

Subsequently, we evaluated the effect of macropinocytosis‐mediated TEV clearance on antitumor immunity. Although ATP6V1A‐deficient 4T1 cells showed comparable viability relative to 4T1 cells in vitro (Figure S5b), both ATP6V1A‐deficient 4T1 tumour exhibited accelerated tumour growth (Figure 5e). We also detected lower proportion of CD8⁺ T cells in the draining lymph nodes (dLNs) and tumour tissues of Atp6v1a^−/− 4T1‐1 tumour‐bearing mice (Figure S5c,d). In addition, increased exhaustion markers and decreased activation and proliferation markers were detected in CD8⁺ T cells in dLNs of these mice (Figure 5f). On the contrary, intratumoral injection of Rabe retarded tumour growth in 4T1 tumour‐bearing mice (Figure 5g), without any adverse effects on gastrointestinal damage evidenced by similar body weight (Figure S5e). Correspondingly, we found increased proportion of CD8⁺ T cells in the dLNs and tumour tissues (Figure S5f,g), and reduced exhaustion markers and enhanced activation and proliferation markers in CD8⁺ T cells in dLNs of these mice (Figure 5h). This corresponded to the results that intratumoral injection of Rabe did not restrict 4T1 tumour growth in thymus‐deficient mice as well as in CD8⁺ T cells depleted mice (Figure S5h). Furthermore, Rabe did not affect the progression of both ATP6V1A‐deficient 4T1 tumour (Figure S5i), which suggests that Rabe‐mediated enhancement of antitumor T‐cell responses is macropinocytosis dependent.

Macropinocytosis is believed to facilitate tumour progression by supplying nutrients during rapid tumour growth (Finicle et al., 2018). Although we detected enhanced macropinocytosis in the 4T1 tumour of Rabe‐treated mice (Figure 5d), we found neither decreased TUNEL staining and autophagosomes nor increased Ki‐67 staining in 4T1 tumour tissues from Rabe‐treated mice (Figure S6a–c), which indicated that Rabe did not affect the apoptosis and proliferation of tumour cells and nutrient‐sensitive autophagy in tumours. In addition, we found comparable apoptosis, proliferation and autophagy between Atp6v1a^−/− 4T1‐1 and 4T1 tumours (Figure S6d–f). These results suggest that macropinocytosis probably does not directly affect the biological characteristics of tumour cells.

3.6. PPIs attenuate the resistance to anti‐PD‐1 therapy of tumour patients

PD‐L1⁺ TEVs are involved in the suppression of systemic antitumor immunity and contribute to anti‐PD‐1 therapy resistance (Poggio et al., 2019; Xie et al., 2019). Before a noticeable difference in tumour size was observed, the general quantity of EVs was enhanced in Atp6v1a^−/− 4T1‐1 tumour‐bearing mice compared with 4T1 tumour‐bearing mice, evidenced by increased serum CD63⁺CD9⁺ EVs. Furthermore, the CD63⁺PD‐L1⁺ EV subset was also significantly enhanced (Figure S7a). Opposite results were obtained in 4T1 tumour‐bearing mice with Rabe treatment relative to that without Rabe treatment (Figure S7b). Rabe‐mediated TEV reduction was further confirmed by the observation of reduced serum RFP⁺ EVs of RFP‐expressing 4T1 tumour mice (Figure S7c). In Pd1^−/− mice, Atp6v1a^−/− 4T1‐1 and 4T1 tumours or 4T1 tumours with or without Rabe treatment showed similar growth tendencies (Figure S7d,e). Systemic administration of Rabe also decreased serum CD63⁺CD9⁺ and CD63⁺PD‐L1⁺ EVs and slowed 4T1 tumour growth (Figure S7f,g). Thus, Rabe effectively prevents PD‐L1⁺ TEV‐induced immunosuppression. Based on these data, we wondered whether Rabe could abolish anti‐PD‐1 therapy resistance of 4T1 tumours. We first excluded the adverse effects of Rabe on gastrointestinal damage indicated by similar body weight (Figure S7h,i). As expected, anti‐PD‐1, along with the systemic injection of Rabe, significantly inhibited anti‐PD‐1 therapy‐resistant 4T1 tumour growth (Figure 6a). Similar results were obtained in B16F10 tumour‐bearing mice (Figure 6b). Rabe treatment also increased CD8⁺ T cell infiltration in both anti‐PD‐1 therapy‐resistant 4T1 and B16F10 tumour tissues (Figure S7j). To elucidate the functional relevance of PPIs on tumour patient TEVs, we collected blood from gastric tumour patients with PPI treatment and found that PPI treatment significantly reduced CD63⁺CD9⁺ and CD63⁺PD‐L1⁺ EVs in circulation (Figure 6c and Table S1). As expected, PPI treatment increased the proportion of circulating CD8⁺ T cells (Figure 6d), reduced the exhaustion markers, and enhanced activation and proliferation markers in CD8⁺ T cells (Figure 6e,f). Therefore, PPIs probably alleviate anti‐PD‐1 therapy resistance of tumour patients by reducing systemic TEVs. We retrospective analysed the survival of 183 patients with advanced lung cancers treated with single‐agent anti‐PD‐1 therapy as second‐ or subsequent‐line therapy. During treatment, 28 patients received Ome (20 mg, bid) oral administration due to gastric ulcers (Table S2). Compared with the patients with anti‐PD‐1 treatment alone, these 28 patients had a favourable progress‐free survival (PFS) compared with that without PPI treatment (Figure 6g). Altogether, PPIs attenuate anti‐PD‐1 therapy resistance by clearing immunosuppressive TEVs.

FIGURE 6.

PPIs attenuate the resistance to anti‐PD‐1 therapy of tumour patients. (a and b) BALB/c mice were injected subcutaneously with 4T1, and C57BL/6J mice were injected subcutaneously with B16F10 on day 0. Then, these mice received an intraperitoneal injection with isotype control antibodies or anti‐PD‐1 together with or without intravenous injection of Rabe from day 11 every 2 days. Tumour sizes of 4T1 (a) or B16F10 (b) tumour‐bearing mice. (c–f) ELISA analysis of CD63⁺CD9⁺ and CD63⁺PD‐L1⁺ EVs (indicated by OD 450 nm values) in sera (c), representative flow cytometric plots (left) or statistical analysis (right) of the percentages of circulating CD8⁺ T cells (d), PD‐1⁺CD8⁺ T and Tim3⁺CD8⁺ T (e) and Ki‐67⁺CD8⁺ T and granzyme B (Gzm B)⁺CD8⁺ T cells (f) in the blood (d–f) of gastric tumour patients before (Day 0) and after (Day 2) PPI treatment. (g) PFS of lung cancer patients received anti‐PD‐1 therapy along with or without PPI treatment. Representative results from three independent experiments are shown. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001 (one‐way ANOVA followed by Tukey test in a and b; paired two‐tailed Student's t‐test in c–f; log‐rank test in g; mean ± SD).

3.7. PPIs enhance the delivery efficacy of therapeutic drugs by EVs

EVs are ideal delivery vehicles for RNA and chemotherapeutic reagents (Mir & Goettsch, 2020). Increased endocytosis of EVs will probably enhance their delivery efficiency, so we investigated the effect of PPIs on this. Since Rabe promoted EV uptake by tumours, we examined whether Rabe improves the antitumor outcome of chemotherapeutic drug‐loaded RBC‐EVs. First, we confirmed that Rabe‐induced RBC‐EVs uptake by 4T1 cells was also eliminated by LY294002 and KM91104 (Figure S8a). We confirmed that systemically administration of Rabe promoted the uptake of RBC‐EVs by 4T1 tumour (Figure S8b). Then, we loaded doxorubicin into RBC‐EVs (RBC‐EVs/Dox) according to our previous publication (Zhang et al., 2020). We found that Rabe treatment resulted in more pronounced cytotoxicity in 4T1 but not Atp6v1a^−/− 4T1 cells caused by RBC‐EVs/Dox (Figure S8c). Correspondingly, Rabe significantly enhanced the antitumor effect of RBC‐EVs/Dox on 4T1 but not Atp6v1a^−/− 4T1 tumour in nude mice (Figure 7a,b), without any adverse effects on gastrointestinal damage indicated by similar body weight (Figure S8d,e). In addition, we detected increased Dox uptake and apoptosis in the Rabe‐treated 4T1 tumour, which was markedly abolished in the Atp6v1a^−/− 4T1 tumour (Figure 7c).

FIGURE 7.

PPIs enhance the delivery efficacy of therapeutic drugs by EVs. (a–c) 4T1 or Atp6v1a ^−/− 4T1 cells were injected subcutaneously in both flanks of the same mouse (Nude mice) on day 0. Then, these mice received an intravenous injection of RBC‐EVs/Dox together with or without an intratumoral injection of Rabe from day 9 every 2 days. Tumour sizes (a and b), representative immunofluorescence images of Tunnel staining and RBC‐EVs/Dox in tumour tissues (c) from 4T1 (a and c) or Atp6v1a^−/− 4T1 (b and c) tumour‐bearing nude mice. (d–f) ALF mice induced by D‐GalN and LPS (D‐GalN/LPS) treatment were intravenously injected with RBC‐EVs/miR‐NC‐ASO, RBC‐EVs/miR‐NC‐ASO and Rabe, RBC‐EVs/miR‐155‐ASO, or RBC‐EVs/miR‐155‐ASO and Rabe. H&E stained liver tissue sections (d), plasmatic ALT and AST levels (e), ELISA analysis of IL‐1β, IL‐6 and TNF protein levels in sera (f) of these mice. Representative results from two independent experiments are shown. Scale bar, 50 μm. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001 (one‐way ANOVA followed by Tukey test except for unpaired two‐tailed Student's t‐test in a and b; mean ± SD).

We have reported that RBC‐EVs loaded with antisense oligonucleotides of microRNA‐155 (RBC‐EVs/miR‐155‐ASOs) can treat ALF (Zhang et al., 2020). Since Rabe also increased liver accumulation of RBC‐EVs, we tested whether Rabe can improve the therapeutic effect of RBC‐EVs/miR‐155‐ASOs. As expected, Rabe has a synergistic impact on RBC‐EVs/miR‐155‐ASO‐induced downregulation of miR‐155 levels in the liver of ALF (Figure S8f). In keeping with this result, Rabe notably improved the therapeutic effects of RBC‐EVs/miR‐155‐ASOs on ALF, as evidenced by the liver histopathology, function and levels of pro‐inflammatory cytokines (Figure 7d–f). However, we could not observe these effects in Atp6v1a^−/− mice (obtained by crossing Cmv‐cre mice with Atp6va1^fl/fl mice) (Figure S8g–i). In summary, Rabe enhances the EV‐mediated delivery of therapeutic drugs by promoting EV endocytosis.

4. DISCUSSION

PPIs have been reported to inhibit EV release (Chalmin et al., 2010; Iero et al., 2008). In keeping with these publications, we detected reduced EV production from PPI‐treated tumour cells. However, we found that PPIs did not affect EV biogenesis, including origin, degradation, PM fusion of MVBs and ILV budding. Unexpectedly, we found that PPI treatment substantially enhanced the EV uptake by tumour cells, thereby reducing EV amount. Therefore, our results demonstrate that the EV amount in the cell culture supernatants and body fluid was collectively determined by cells’ secretion and uptake of EVs, which alarms that EV uptake should be routinely assessed in EV biogenesis studies.

Low pH has been reported to increase EV release by tumour cells (Logozzi et al., 2018), and our results suggest that PPIs promote tumour cells’ EV uptake by declining cytoplasmic pH, thereby reducing EVs. Unlike PPIs that downgrade cytoplasmic pH, a lasting external acidic environment profoundly affects tumour cells. For instance, an acidic environment drives transcriptome alterations of tumour cells (Rohani et al., 2019). A proton‐activated chloride channel, a pH sensor, induces malignant properties of human osteosarcoma cells (Peng et al., 2021). Therefore, the acidic environment probably upregulates the expression of genes involved in EV biogenesis in tumour cells to increase TEV secretion.

EVs are involved in the progression of various diseases, so therapeutic effects can be achieved by regulating EV biogenesis. Although many studies have revealed the mechanisms for EV biogenesis, the strategies for modulating EV release are still in experimental feasibility (Han et al., 2022). To exert their functions, EVs are necessary to be taken up by effector cells. Therefore, EV functions can also be controlled by regulating effector cells to take up EVs or EV clearance by non‐effector cells. In this study, we found that PPI‐mediated enhanced uptake of EVs by tumour cells was due to the induction of macropinocytosis, which is responsible for the uptake of most EVs by recipient cells. Given that PPIs are familiar and safe clinical drugs for treating gastritis and gastric ulcers, PPIs represent promising candidate drugs for treating EV‐related diseases.

To further support our hypothesis, we validated the regulatory effects of PPIs on EV functions in murine tumour and IBD models. Our data showed that PPI treatment distinctly reduced systemic TEV levels by enhancing TEV uptake in a macropinocytosis‐dependent manner, thus successfully mounting antitumor immunity and subsequent suppression of tumour progression. More importantly, when combined with PPIs, therapy resistance of anti‐PD‐1 was effectively abolished. In addition, PPIs also aggravated IBD symptoms by promoting IEC‐EV clearance via macropinocytosis. Therefore, these results verify that reducing systemic EV levels can be achieved by enhancing EV uptake. However, our results simultaneously raise an unavoidable debate that PPI‐mediated enhancement of EV uptake is not cell‐specific and that effector cells will also internalise increased EVs, which may amplify EV functions. IEC‐EVs and TEVs exert their immunosuppressive functions by activating TGF‐βR and PD‐1 signalling, respectively, which requires the interaction of ligands on EVs and receptors on T cells. Therefore, the fast internalisation of IEC‐EVs and TEVs is beneficial to eliminate their effects on T cells. Furthermore, regarding T cells, PPI‐induced macropinocytosis is relatively highly specific because macropinocytosis showed minimal impact on the EV uptake of T cells. In addition to engaging receptors, EVs also function through intracellular cargo delivery to effector cells. However, PPIs are more likely to promote TEV recycling by tumour cells due to the high macropinocytosis‐dependent EV uptake of tumour cells. In addition, EVs should be inclined to be internalised in situ by their parent cells rather than effector cells after release because of spatial distance and homology advantages. This is similar to the finding that Rabe did not affect Atp6v1a^−/− 4T1 tumour progression, suggesting that PPI‐induced tumour‐cell macropinocytosis is essential for TEV clearance.

According to the classical point of view, macropinocytosis is beneficial in improving the nutrient deficiency of tumours, which facilitates tumour survival (Finicle et al., 2018). However, our results demonstrated that enhanced macropinocytosis of tumour cells prevents tumour growth by clearing the immunosuppressive TEVs, thereby raising the opposite role of macropinocytosis in tumour progression. In addition, we found that macropinocytosis did not seem to affect the starvation status of tumours, as evidenced by the unchanged autophagy in tumour tissues with different macropinocytic activity. Moreover, macropinocytosis affected neither apoptosis nor proliferation of tumour cells in vivo. Combined with previous publications in which macropinocytosis alone showed no effect on tumour growth in vivo (Huang et al., 2023; Jayashankar & Edinger, 2020), we assumed that the starvation status of tumours probably needs to reach a threshold at which macropinocytosis‐mediated nutrient supply is vital for tumour survival. Then, the oncogenic effects of macropinocytosis will show after that. Thus, more delicate studies are required to clarify the dynamic influence of macropinocytosis during tumour progression.

EVs are popularly accepted as prominent vehicles for delivering therapeutic drugs (Herrmann et al., 2021; O'Brien et al., 2020). Relevant to their functions on EVs, PPIs markedly increased EV delivery efficiency by promoting their internalisation. According to our results, PPIs significantly increased the EV accumulation in tumours and the liver, thus improving the antitumor effects of RBC‐EVs/Dox or the therapeutic effects of RBC‐EVs/miR‐155‐ASOs on ALF. Mesenchymal stem cell‐derived EVs (MSC‐EVs) have potential in regenerative medicine applications and hold ideal translational prospects due to their low immunogenicity. However, enhancing MSC‐EV production is substantially challenging (Chen et al., 2020; Keshtkar et al., 2018; Tsiapalis & O'Driscoll, 2020). The combined application of PPIs probably lowers the therapeutic dosage of MSC‐EVs, thereby accelerating their clinical translation. Overall, we presented a proof‐of‐concept for PPIs as potentiators of EV delivery vehicles.

Although increased PM v‐ATPases contribute to oncogenic RAS‐induced macropinocytosis via PKA in tumour cells (Ramirez et al., 2019), whether other mechanisms are responsible for macropinocytosis by regulating v‐ATPase PM translocation is unclear. When investigating PPI‐induced macropinocytosis, we revealed that the V₀ unit of PM v‐ATPases and vacuolar v‐ATPases competitively bind the v‐ATPase V₁ unit. Decreased cytosolic pH disassembled vacuolar v‐ATPase and released the V₁ unit, which then interacts with the V₀ unit of PM v‐ATPases, increasing PM v‐ATPase assembly and enhancing macropinocytosis. Thus, we unravel the mechanism responsible for the dynamic regulation of PM and vacuolar v‐ATPase assembly, which influences macropinocytosis in a PKA‐independent manner. These findings provide a universal mechanism for macropinocytic regulation in cells with or without RAS mutation.

AUTHORS CONTRIBUTION

Xinliang Lu, Zhengbo Song, Jiayue Hao, Xianghui Kong, Weiyi Yuan, Yingying Shen, Chengyan Zhang, Jie Yang and Yun Qian performed various experiments; Pengfei Yu analysed clinical data. Gensheng Zhang and Jianli Wang discussed the manuscript. Huajun Feng, Zhijian Cai, Zhenzhai Cai and Jianli Wang designed the project and supervised the study; Xinliang Lu and Zhijian Cai wrote the manuscript.

CONFLICT OF INTEREST STATEMENT

The authors declare no competing interests.

Supporting information

Supplementary Videos

Supporting Information

Supporting Information

Lu, X. , Song, Z. , Hao, J. , Kong, X. , Yuan, W. , Shen, Y. , Zhang, C. , Yang, J. , Yu, P. , Qian, Y. , Zhang, G. , Feng, H. , Wang, J. , Cai, Z. , & Cai, Z. (2024). Proton pump inhibitors enhance macropinocytosis‐mediated extracellular vesicle endocytosis by inducing membrane v‐ATPase assembly. Journal of Extracellular Vesicles, 13, e12426. 10.1002/jev2.12426

DATA AVAILABILITY STATEMENT

All data needed to evaluate the conclusions in the paper are presented in the main manuscript, the Supplementary Information and the Supplementary Data File.