Blockage of immune checkpoint molecules increases T‐cell priming potential of dendritic cell vaccine

Summary

Dendritic cell (DC) ‐based cancer immunotherapy is one of the most important anti‐cancer immunotherapies, and has been associated with variable efficiencies in different cancer types. It is well‐known that tumor microenvironment plays a key role in the efficacy of various immunotherapies such as DC vaccine. Accordingly, the expression of programmed death ligand 1 (PD‐L1) on DCs, which interacts with PD‐1 on T cells, leads to inhibition of anti‐tumor responses following presentation of tumor antigens by DCs to T cells. Therefore, we hypothesized that down‐regulation of PD‐L1 in DCs in association with silencing of PD‐1 on T cells may lead to the enhancement of T‐cell priming by DCs to have efficient anti‐tumor T‐cell responses. In this study, we silenced the expression of PD‐L1 in DCs and programmed cell death protein 1 (PD‐1) in T cells by small interfering RNA (siRNA) ‐loaded chitosan–dextran sulfate nanoparticles (NPs) and evaluated the DC phenotypic and functional characteristics and T‐cell functions following tumor antigen recognition on DCs, ex vivo. Our results showed that synthesized NPs had good physicochemical characteristics (size 77·5 nm and zeta potential of 14·3) that were associated with efficient cellular uptake and target gene silencing. Moreover, PD‐L1 silencing was associated with stimulatory characteristics of DCs. On the other hand, presentation of tumor antigens by PD‐L1‐negative DCs to PD‐1‐silenced T cells led to induction of potent T‐cell responses. Our findings imply that PD‐L1‐silenced DCs can be considered as a potent immunotherapeutic approach in combination with PD‐1‐siRNA loaded NPs, however; further in vivo investigation is required in animal models.

Expression of PD‐L1 on DCs inhibits their function. PD‐1 on T cells disrupts T cell function. Concomitant silencing of PD‐L1 and PD‐1 can induce potent anti‐tumor immune responses.

Abbreviations

BrdU

bromodeoxyuridine

CDS

chitosan–dextran sulfate

DC

dendritic cell

ELISA

enzyme‐linked immunosorbent assay

iDCs

immature DCs

mAbs

monoclonal antibodies

NPs

nanoparticles

PD‐1

programmed cell death protein 1

PD‐L1

programmed death ligand 1

qPCR

real‐time polymerase chain reaction

SEM

standard error of the mean

Treg

regulatory T

Introduction

Dendritic cell (DC) vaccines are extensively used for the immunotherapy of several cancer types, but variable outcomes have been observed, which might be related to the presence of various impediments in the tumor microenvironment.1 The tumor microenvironment is composed of several cellular and non‐cellular components, such as regulatory T (Treg) cells, tumor‐associated macrophages, tumor‐associated neutrophils, myeloid‐derived suppressor cells, hypoxia, adenosine, inhibitory cytokines, and other factors that may provide optimum conditions for rapid expansion of cancer cells.2 Expression of immune‐inhibitory checkpoint molecules such as programmed cell death protein 1 (PD‐1) and cytotoxic T‐lymphocyte antigen 4 (CTLA‐4) on tumor‐infiltrating T cells, and their ligands on antigen‐presenting cells or cancer cells is one of the most important mechanisms for tumor escape through suppression of anti‐tumor immune responses.3 Accordingly, blockage of these immune checkpoint molecules has been approved by the US Food and Drug Administration for the treatment of some cancer types such as melanoma and non‐small cell lung carcinoma.4 Several trials are underway to evaluate the efficacy of immune checkpoint blockers alone or in combination with other immunotherapeutic approaches for the treatment of various cancers.5 Therefore, blockage of these checkpoint molecules and/or their ligands may be considered as a promising therapeutic strategy to restore anti‐tumor effects of T cells in the tumor microenvironment. We have demonstrated that the blockage of CD73 and adenosine A2a receptor increases the anti‐tumor potential of DC vaccines and restores T‐cell responses in tumor‐bearing mice.6, 7, 8 We have also shown that inhibition of hypoxia‐inducible factor‐1α or adenosine receptor might enhance the efficacy of DC vaccine in animal models.9 These results imply that the attenuation of an immunosuppressive tumor microenvironment may increase the function of DC vaccine, followed by tumor elimination. Consistently, several studies have demonstrated the advantages of knocking down programmed death ligand 1 (PD‐L1) in the efficacy of DC vaccine therapy.10, 11, 12 On the other hand, there is evidence indicating the importance of PD‐1 suppression on T cells to achieve a better outcome following DC therapy.13, 14 However, there is no study regarding the concomitant silencing of PD‐L1/PD‐1 molecules on DCs and T cells, respectively, to increase the anti‐tumor T‐cell responses. Therefore, in this study, for the first time we suppressed PD‐L1 and PD‐1 on DCs and T cells, respectively, to block the inhibitory signals of these cells. To knock down target genes in DCs and T cells, we used chitosan–dextran sulfate (CDS) nanoparticles (NPs), which were loaded with specific small interfering RNAs (siRNAs). Our previous studies have shown that NPs are potent candidates for specific siRNA delivery and gene silencing in target cells.15, 16 One of the most important advantages of NPs is tumor site‐specific drug delivery, which prevents possible side effects and enhances the efficiency of treatment.17, 18 Hence, using NPs, it is possible to specifically deliver target gene‐specific siRNA molecules to suppress the expression of target molecules.

Our results showed that silencing of PD‐L1 in DC vaccines can affect their phenotypic and functional characteristics and increase T‐cell priming potential. Moreover, PD‐1 silencing could enhance T‐cell responses following recognition of tumor antigens on PD‐L1‐silenced DCs. These findings suggest the high potency of cancer immunotherapy by PD‐L1‐silenced DC vaccines in combination with PD‐1 siRNA‐loaded NPs.

Materials and methods

Materials

Nanoparticles were produced from low‐molecular‐weight chitosan and dextran sulfate (Sigma, St Louis, MO). Dimethyl sulfoxide (DMSO), sodium tripolyphosphate, hydrochloric acid, and glacial acetic acid were purchased from Merck (Darmstadt, Germany). PD‐1‐ and PD‐L1‐specific siRNAs as well as non‐targeting control siRNAs were ordered from Santa Cruz Biotechnology (Santa Cruz, CA). The toxic effect of NPs was evaluated using an MTT assay kit (Roche Applied Science, Mannheim, Germany). Cytokine measurement was carried out with enzyme‐linked immunosorbent assay (ELISA) kits (eBioscience, San Diego, CA) and fluorochrome‐conjugated monoclonal antibodies (mAbs) bought from BioLegend (San Diego, CA).

Mice and cell lines

The 4T1 murine breast cancer cells were bought from the National Cell Bank of Iran (Pasteur Institute of Iran, Tehran, Iran). Female 6‐ to 8‐week‐old BALB/c mice were purchased from the Laboratory Animal Center, Pasteur Institute of Iran. All animal‐related experiments were performed according to the instructions of the Animal Ethics Committee affiliated to Tabriz University of Medical Sciences. RPMI‐1640 medium, which was supplemented with 10% heat‐inactivated fetal bovine serum, 10 mm l‐glutamine, streptomycin (100 μg/ml), and 100 units/ml penicillin (all purchased from Gibco, Grand Island, NY) was used to culture all cells in a humidified incubator at 37° and 5% CO₂. Inoculation of tumors was performed based on our previous reports by subcutaneous injection of 7 × 10⁵ 4T1 cells into the right flank of syngeneic mice.8

Synthesis of NPs

The CDS conjugate was produced as described previously, through ionic gelation method.19 Briefly, 1 ml chitosan (50 000 MW) solution (1·5 mg/ml) at pH 5·5 was added drop‐wise to 500 μl dextran sulfate (500 μg/ml) at pH 4·5 and stirred (500 rpm) for 30 min. The CDS complex produced was then dialyzed against distilled water for 3 days, centrifuged, and the supernatant was freeze‐dried and stored at room temperature for subsequent experiments. The siRNA‐loaded CDS NPs were produced by addition of 0·8 ml siRNA/tripolyphosphate (containing 10 µg siRNA) to 1 ml CDS solution (5 mg/ml) under stirring for 20 min. Loading of siRNA molecules was investigated applying electrophoresis on the agarose gel (2%).20

Validation and characterization of the CDS NPs

Conjugation of chitosan to dextran sulfate was evaluated by fourier‐transform infrared spectroscopy (FTIR) spectra by the Nicolet FT‐IR Spectrometer (Magna IR 550; Madison, WI). Samples were mixed with KBr powder and compressed into disk plates before scanning in absorption with 4‐cm resolution in ambient conditions.

The morphology of synthesized CDS NPs was analyzed by standard error of the mean (SEM) microscopy using an Eindhoven electron microscope (XL 30; Philips, Eindhoven, the Netherlands). The fresh samples of NPs were dropped on metal stubs, coated with a layer of gold and examined by scanning electron microscopy.

The size and zeta characterization of siRNA‐loaded CDS NPs was evaluated by dynamic light scattering on a Zetasizer (Nano‐ZS; Malvern Instruments, Malvern, UK). A wavelength of 633 nm at 25° with an angle detection of 90° was used for the measurement.

The release pattern of siRNA molecules from CDS NPs was evaluated as described previously21 by dialysis of siRNA‐loaded NPs in phosphate‐buffered saline with pH 7·4 and pH 5·5 during 72 hr.

Stability of NPs in the serum was analyzed by electrophoresis in 4% agarose gel. Briefly, 400 μl CDS NPs loaded with 10 μg siRNA molecules were mixed with 200 μl fetal bovine serum and gently shaken at 37°. Twenty‐microliter samples were then collected at predetermined times (2, 6, 12, 18, 24, and 36 hr), and stored at −20° for subsequent analysis by gel electrophoresis.

Uptake of fluorescein isothiocyanate (FITC) ‐conjugated siRNA‐loaded CDS NPs by spleen‐derived T cells was evaluated using an inverted fluorescence microscope (BX40; Olympus, Tokyo, Japan).22 image j software was used for quantitative evaluation of uptake results. Flow cytometry (Becton‐Dickinson, Mountain View, CA) was used to evaluate the efficacy of cellular uptake and frequency of transfected cells by FITC‐conjugated NPs. flowjo software (FlowJo LLC, Ashland, OR) was used to analyze the data.

Cell cytotoxicity assay (MTT)

The cell cytotoxicity of siRNA‐loaded NPs on 4T1 (murine breast) and CT26 (murine colon) cell lines was investigated by MTT assay, as described previously.21 Briefly, 10⁴ cells were seeded in 96‐well flat‐bottom plates in 100 μl/well RPMI‐1640 medium and were treated with 80 pmol of siRNA‐loaded NPs, blank NPs, naked siRNA, 0·2% DMSO (as a control toxic agent) or remained untreated (as a negative control) for 48 hr. The primary medium was then substituted with fresh serum‐free medium and 10 µl of MTT solution (20 mg/ml) was added and incubated for a further 4 hr. Subsequently, 100 μl of DMSO was added to each well after aspiration of 110 μl medium from each well and plates were re‐incubated for 30 min. Viability of cells was finally determined by scanning of plates in a 96‐well microplate reader (Organon Teknika, Turnhout, Belgium) at 570 nm.

Real‐time PCR

The mRNA expression levels of target genes were assessed by quantitative real‐time polymerase chain reaction (qPCR) according to our previously reported method23 using a standard TRIzol/isopropanol protocol. β‐Actin was used as an internal control. Primer sequences are shown in Table 1.

Table 1.

Primer sequences

β‐actin	Forward	5′‐GGTCATCACTATTGGCAACG‐3′
	Reverse	5′‐ACGGATGTCAACGTCACACT‐3′
FoxP3	Forward	5′‐CACCCAGGAAAGACAGCAACC‐3′
	Reverse	5′‐ GCAAGAGCTCTTGTCCATTGA‐3′
PD‐1	Forward	5′‐GCCTGGCTCACAGTGTCAG‐3′
	Reverse	5′‐TCCAGGGCTCTCCTCGATT‐3′
PD‐L1	Forward	5′‐GGAATTGTCTCAGAATGGTC‐3′
	Reverse	5′‐GTAGTTGCTTCTAGGAAGGAG‐3′
IL‐10	Forward	5′‐AGCATTTGAATTCCCTGGGTGA‐3′
	Reverse	5′‐CCTGCTCCACTGCCTTGCTCTT‐3′
IL‐12	Forward	5′‐CCAAATTACTCCGGACGGTTCAC‐3′
	Reverse	5′‐CAGACAGAGACGCCATTCCACAT‐3′

Isolation of T cells by MACS cell separation and transfection with PD‐1 siRNA‐loaded NPs

CD3⁺ T cells were purified from the spleen and tumor samples using a T‐cell isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany), through a negative selection method as described previously.24 The purity of isolated cells was more than 95%, as investigated by flow cytometry. Transfection of T cells with PD‐1 siRNA‐loaded NPs to suppress the expression of PD‐1 was performed according to our previous report.24

Cytokine evaluation by ELISA

Concentrations of cytokines secreted by DCs or T cells co‐cultured with DCs including interferon‐γ (IFN‐γ), tumor necrosis factor (TNF), transforming growth factor‐β, interleukin‐2 (IL‐2), IL‐10, IL‐12 (P70), and IL‐17 were evaluated by ELISA kits, according to the manufacturer’s instructions (eBioscience, Affymetrix, CA).

Apoptosis assay

The effect of siRNA‐loaded NPs on apoptosis of DCs or T cells was evaluated by annexin V/propidium iodide (PI) (BD Biosciences, San Jose, CA) flow cytometry.25 T cells or DCs (2 × 10⁵) were cultured in 96‐well V‐shaped plates and treated with siRNA‐loaded NPs for 24 hr. After incubation with annexin V‐FITC and PI, the frequency of apoptotic cells was determined by flow cytometry.

Flow cytometry

The frequency of CD4⁺ FoxP3⁺ Treg cells, CD8⁺ CD107⁺ T cells, phenotypic characterization of DCs and T cells, and expression of PD‐1/PD‐L1 were determined by flow cytometry using CD4‐allophycocyanin (APC), FoxP3‐Alexa Fluor 488, CD3‐APC, PD‐1‐APC, PD‐L1‐phycoerythrin (PE), CD11c‐PE, CD80‐FITC, CD86‐PE, HLA‐DR‐FITC, CD40‐PE‐Cy5, CCR7‐PE, CD8‐FITC, CD107‐APC, and isotype‐matched control mAbs (BioLegend), as described previously.26 Briefly, 10⁶ cells were suspended in 50 µl of staining buffer and incubated with optimized amounts of fluorochrome‐conjugated mAbs at 4° for 30 min. FoxP3 intracellular staining was performed by Cytofix/Cytoperm Fixation/Permeabilization kit (BD Biosciences). After washing twice, stained cells were evaluated by FACS Calibur flow cytometer (Becton‐Dickenson) and analyzed using flowjo software.

Migration assay

Migration of control or PD‐L1‐silenced DCs in response to increasing concentrations of CCL21 was evaluated in 24‐well Transwell plates (Corning Costar Corp., Cambridge, MA). Briefly, 600 μl of complete medium containing 0·30–60 ng/ml CCL21 (PeproTech, London, UK) was added to the lower part of the wells and in the upper chamber, 100 μl of medium including 10⁵ control or PD‐L1‐silenced DCs was added to the inserts. The number of DCs migrating to the lower chamber was obtained using flow cytometry after 2 hr.

Preparation of tumor lysate

For preparation of tumor lysate, 5 × 10⁶ 4T1 cells were subcutaneously injected into the right flank of BALB/c mice and after sufficient tumor growth, mice were killed and tumor tissues were harvested surgically and single‐cell suspension was generated. Repeated freezing–thawing cycles were used to generate tumor cell lysate. Generated lysate was centrifuged (900 g, 10 min) and supernatant was collected and filtered (0·2 μm) for further use. The protein content of tumor lysate was measured by the Lowry method.27

DC preparation and PD‐L1 silencing

Dendritic cells were generated from bone marrow precursors as described previously.8 Briefly, 1 × 10⁶ bone marrow‐derived cells isolated from the femurs and tibias of mice were seeded in 24‐well plates in 1 ml RPMI‐1640 medium containing 100 U/ml of recombinant murine granulocyte–macrophage colony‐stimulating factor (BD) and 50 U/ml of recombinant murine IL‐4 (BD). Non‐adherent cells were harvested on day 3 and fresh medium was substituted. Tumor lysate (80 μg/ml) was added to immature DC (iDC) cultures on day 5. Lipopolysaccharide (Sigma) (1 μg/ml) was then added to DC culture after 10 hr. DCs were matured on day 7 and used for further experiments.

To suppress PD‐L1 expression in DCs, iDCs were transferred to six‐well plates and transfected with PD‐L1 siRNA‐loaded CDS NPs (containing 80 pm siRNA) for 6 hr. Subsequently, fresh medium containing 100 U/ml granulocyte–macrophage colony‐stimulating factor and 50 U/ml IL‐4 was added to DC culture. At day 5, cells were again transfected with PD‐L1 siRNA‐loaded NPs for 6 hr, loaded with tumor lysate and matured by lipopolysaccharide in a similar way to the control DCs.

Mixed leukocyte reaction assays

In order to evaluate T cell priming potential of control or PD‐L1‐silenced DCs, one‐way mixed leukocyte reaction was performed with DCs as stimulators and primed BALB/c tumor tissue‐derived T cells as responders, using different DC : T‐cell ratios (1 : 20, 1 : 40 and 1 : 80) in 96‐well, round‐bottomed microculture plates (Nunc, Copenhagen, Denmark) for 96 hr. A colorimetric immunoassay measuring bromodeoxyuridine (BrdU) incorporation with a BrdU ELISA kit (Roche Applied Science, Indianapolis, IN) was used to evaluate T‐cell proliferation. Briefly, 20 μl of BrdU (200 μm) was added to each well during the final 18 hr of co‐culture. Cells were then centrifuged, denatured and incubated with 100 μl/well of peroxidase‐conjugated mouse anti‐BrdU mAb for 90 min. Finally, unbound mAbs were removed and tetramethylbenzidine substrate (100 μl) was added to each well, incubated for 20 min and H₂SO₄ solution (1 m) was used to stop the reaction. The absorbance of plates was evaluated at 450 nm against a reference wavelength at 690 nm by an ELISA plate reader (Hyperion Micro Reader 4 plus, Deerfield Beach, FL).

Moreover, to evaluate the impact of PD‐L1/PD‐1 silencing on induction of CD107‐expressing CD8⁺ T cells, 5 × 10⁵ tumor‐derived T cells were co‐cultured with 10⁵ tumor‐lysate‐loaded DCs in the presence of APC‐conjugated anti‐CD107 antibody and monensin (10 μm; BioLegend) at 37° for 72 hr. Cells were treated with PD‐L1 siRNA‐loaded NPs, PD‐1 siRNA‐loaded NPs, both siRNA‐loaded NPs, or remained untreated. After incubation, cells were harvested, washed, and stained with CD8‐FITC antibody and evaluated by flow cytometry.

Statistical analysis

The SPSS statistical package (SPSS, Chicago, IL) was used for statistical analysis of results. Comparison between groups was done using one‐way analysis of variance, but a two‐group statistical comparison was performed using the Mann–Whitney U‐test. P‐values < 0·05 were considered significant. All experiments were carried out in triplicates and repeated at least twice and results have been shown as mean ± SEM.

Results

Physicochemical characteristics of NPs

Synthesized CDS NPs exhibited good physicochemical characteristics, with a size of 77·5 nm, with polydispersive index < 0·2 and zeta potential of 14·3 mV (Fig. 1a, b). This size and surface charge can facilitate the siRNA loading and cellular uptake both in vitro and in vivo.

Figure 1.

Physicochemical characteristics of chitosan–dextran sulfate (CDS) nanoparticles (NPs). NPs had a size of 77·5 nm, a polydispersive index < 0·2 (a), and a zeta potential of 14·3 (b). FTIR spectra of CDS NPs showed the proper conjugation as indicated by existence of functional groups demonstrating sulfyl peaks near 1021/cm, 1255/cm, and 815/cm (c). Scanning electron microscopy demonstrated the morphology of NPs (d). The small interfering RNA (siRNA) loading capacity of NPs is evaluated on agarose gel (e). A serum stability of CDS NPs loaded with siRNA incubated in serum is shown during pre‐determined time‐points (f). The siRNA release pattern of CDS NPs in phosphate‐buffered saline solution at pH 5·5 and pH 7·4 (g). MTT assay demonstrating the in vitro cytotoxicity of CDS NPs following 48 hr of treatment (h).

To confirm the conjugation of chitosan to dextran sulfate, FTIR spectra of CDS NPs were evaluated to investigate the presence of functional groups at peaks near 1021/cm, 1255/cm, and 815/cm demonstrating sulfyl groups, as shown in Fig. 1(c). Moreover, NPs had uniform spherical morphology, as shown using scanning electron microscopy (Fig. 1d).

The siRNA loading capacity and serum stability were other investigated characteristics of CDS NPs. As shown in Fig. 1(e), NPs could encapsulate up to 30 μg siRNA, demonstrating high potential of gene loading. Furthermore, the stability of siRNA‐loaded NPs was tested following exposure to fetal bovine serum during pre‐determined time‐points through electrophoresis on agarose gel. The results showed that NP degradation starts after 18 hr and NPs release the majority of their cargo after 36 hr (Fig. 1f).

The siRNA release profile of CDS NPs

The siRNA release pattern of CDS NPs was investigated in two acidic and neutral conditions. There was no difference between siRNA release in acidic (pH 5·5) and neutral (pH 7·4) conditions, implying the drug delivery potential of CDS NPs in the acidic tumor microenvironment. The results showed that siRNA content of NPs was gradually released up to 72 hr, and about 50% of siRNA was released in the first 24 hr (Fig. 1g).

Toxicity of CDS NPs

To investigate the cell toxicity of siRNA‐loaded NPs, MTT assay was used to evaluate NP‐related cytotoxicity in two different cancer cell lines – 4T1 (murine breast cancer) and CT26 (murine colon cancer) cells. The results showed that exposure of cell lines to siRNA‐loaded CDS NPs had no marked toxic effects up to 48 hr incubation, implying the safety of NPs for application in both in vitro and in vivo experiments. Untreated cells were used as negative control and DMSO (0·2%) was used as positive control in this assay. As shown in Fig. 1(h), none of the groups (except the DMSO group) exhibited significant cell death due to treatment with siRNA‐loaded NPs, blank NPs, or naked siRNA molecules.

siRNA‐loaded NPs efficiently suppressed expression of PD‐L1 and PD‐1 in DCs and T cells, respectively

After confirming the physicochemical characteristics of CDS NPs, their efficacy to deliver siRNA to target cells and silencing of target genes was evaluated. To study the cellular uptake of NPs by spleen‐derived T cells, CDS NPs were conjugated with FITC and their uptake by T cells was assessed by fluorescence microscopy and analyzed by image j software (Fig. 2a, b). The quantitative evaluation of transfection efficacy was also tested by flow cytometry. As shown in representative histograms (Fig. 2c), more than 70% of T cells were transfected with siRNA‐loaded CDS NPs, demonstrating the high cellular uptake of this delivery system.

Figure 2.

Nanoparticles (NPs) could efficiently deliver small interfering RNA (siRNA) molecules to dendritic cells (DCs) and T cells and suppress the expression of target genes. The representative fluorescence microscopy images demonstrating the transfection of spleen‐derived T cells with FITC‐conjugated chitosan–dextran sulfate (CDS) NPs are shown (a). Bar charts demonstrate the mean fluorescence intensity of cellular uptake, as assessed by image j software (b). The efficiency of transfection was investigated by flow cytometry assay (c). Bar charts demonstrating the relative mRNA expression of programmed death ligand 1 (PD‐L1) in DCs (d) and programmed cell death protein 1 (PD‐1) in T cells (e) following knockdown with siRNA‐loaded CDS NPs. The protein levels of PD‐L1 in DCs (f–h) and PD‐1 in T cells (i–k) after silencing with siRNA‐loaded NPs were evaluated by flow cytometry assay (blue histograms demonstrate isotype control, orange is untreated cells, and red demonstrated treated cells). (*P < 0·05 and **P < 0·01). MFI, mean fluoresce intensity.

Following confirmation of potent cellular uptake of NPs by target cells, their efficacy for silencing target genes including PD1 in T cells and PDL1 in DCs was investigated in mRNA and protein levels by qPCR and flow cytometry assays, respectively. As depicted in Fig. 2(d), treatment of DCs with PD‐L1 siRNA‐loaded CDS NPs led to potent down‐regulation of PD‐L1 mRNA levels. Similar results were observed following treatment of tumor‐derived T cells (T cells purified from tumor tissues of 4T1 breast‐tumor‐bearing mice) with PD‐1 siRNA‐loaded NPs (Fig. 2e). As demonstrated, siRNA‐loaded NPs could reduce the mRNA expression of target genes by >80% compared with the control group. The expression of PD‐L1 and PD‐1 molecules was also investigated by flow cytometry method. As shown in Fig. 2(f, g), treatment of DCs with PD‐L1 siRNA‐loaded NPs potently reduced the expression of PD‐L1 on DCs. This treatment not only reduced the frequency of PD‐L1‐expressing DCs but also markedly suppressed PD‐L1 intensity on DCs (Fig. 2h). Similar results were obtained following the treatment of tumor‐derived T cells with PD‐1 siRNA‐loaded NPs. As demonstrated in Fig. 2(i, j), transfection of T cells with PD‐1 siRNA‐loaded NPs significantly decreased the frequency of PD‐1‐expressing T cells. Moreover, evaluation of the mean fluorescence intensity of PD‐1 on T cells showed decreased expression of PD‐1 on PD‐1‐positive T cells (Fig. 2k).

Down‐regulation of PD‐L1 in DCs improved their survival, phenotype, cytokine production, and migration

After confirming the efficacy of siRNA‐loaded NPs in silencing PD‐L1 in DCs, the impact of PD‐L1 suppression on DC maturation and function was investigated. In the first step, impact of PD‐L1 silencing by NPs on viability of DCs was evaluated by flow cytometry through Annexin V/PI staining of DCs. As shown in Fig. 3(a, b), down‐regulation of PD‐L1 in DCs could slightly decrease the frequency of apoptotic cells.

Figure 3.

Silencing of programmed death ligand 1 (PD‐L1) improves dendritic cell (DC) characteristics and function. Viability of DCs was evaluated through flow cytometry. Representative dot plots demonstrating the frequency of apoptotic cells in PD‐L1‐silenced DCs and control DCs are shown (a). Bar charts depict the mean frequency of apoptotic DCs (b). Bar charts demonstrate the mean frequency of DCs expressing the maturation and effector markers in PD‐L1‐silenced and control DCs (c). The mRNA (d, e) and protein (f, g) expression levels of interleukin‐12 (IL‐12) and IL‐10 cytokines and their protein expression ratio (h) from PD‐L1 silenced and control DCs in two forms of mature DC (MDC) and immature DC (iDC) were evaluated by quantitative PCR and ELISA. CCR7‐induced migration of PD‐L1‐silenced or control DCs toward accumulating concentrations of CCL21 was analyzed by the flow cytometry. Bar charts demonstrate the mean number of migrated DCs (i). (*P < 0·05).

Maturation and stimulatory characteristics of DCs were subsequently investigated by the flow cytometry assay. As shown in Fig. 3(c), reduced expression of PD‐L1 in DCs significantly increased the expression of CD86 and HLA‐DR, slightly induced the expression of CD80 and CCR7, but no effect on CD11c and CD40 was observed.

Secretion profiles of IL‐12 and IL‐10 and their ratio are important issues in the T‐cell priming potential of DCs that were evaluated following silencing of PD‐L1 in DCs in both mRNA and protein levels as determined by qPCR and ELISA, respectively. The results showed that silencing of PD‐L1 expression could significantly increase IL‐12 and decrease IL‐10 production in PD‐L1‐silenced DCs compared with control DCs in both iDC and mature DCs (Fig. 3d, e). Similar results were observed when the production of IL‐12 and IL‐10 was investigated in supernatant of cultured DCs by ELISA (Fig. 3f, g). Moreover, the IL‐12/IL‐10 protein levels ratio was also increased significantly in PD‐L1‐silenced DCs compared with control DCs (Fig. 3h).

Finally, the impact of PD‐L1 silencing on migratory potential of DCs was investigated using transwell plates against increasing concentrations of CCL21 chemokine, which is specific for CCR7 receptors expressed on DCs. As shown in Fig. 3(i), PD‐L1 silencing increased the migration of DCs toward CCL21 in a dose‐dependent manner.

Silencing of PD‐1/PD‐L1 synergistically enhanced T‐cell responses, induced by tumor‐lysate‐loaded DCs

Following the evaluation of DC characteristics after PD‐L1 silencing, we investigated the T‐cell priming potential of control or PD‐L1‐silenced DCs on the activation of T‐cell responses following PD‐1 silencing. First, the ability of DCs to enhance T‐cell proliferation was evaluated by BrdU ELISA. T cells (as responder cells) and DCs (as stimulator cells) were co‐cultured in three ratios of responder to stimulator cells (20 : 1, 40 : 1, and 80 : 1). The results showed that down‐regulation of PD‐L1 in DCs significantly increased the proliferation of T cells in a 20/1 ratio. Moreover, silencing of PD‐1 in T cells could potently enhance T‐cell proliferation compared with PD‐1‐expressing T cells. Interestingly, co‐silencing of PD‐L1 and PD‐1 in DCs and T cells increased T‐cell expansion compared with silencing of PD‐L1 or PD‐1 molecules alone (Fig. 4a).

Figure 4.

Blockage of programmed cell death protein 1 (PD‐1) in T cells enhances T‐cell responses induced by programmed death ligand 1 (PD‐L1) ‐silenced dendritic cells (DCs). T cells and PD‐1‐silenced T cells were co‐cultured with 4T1 tumor‐lysate‐loaded PD‐L1‐silenced or control DCs in various ratios and their proliferation was evaluated by bromo‐deoxyuridine‐based ELISA (a). Apoptosis of these CD3⁺ T cells was also investigated by flow cytometry (b, c). Impact of PD‐1 silencing in T cells and PD‐L1 silencing in DCs on cytokine production by T cells was also investigated by ELISA during 96 hr of co‐culture of T cells with DCs (d). Representative dot plots show analysis method used to investigate regulatory T (Treg) cell differentiation following co‐culture of T cells and DCs (e). Silencing of PD‐1 in T cells and PD‐L1 in DCs reduced differentiation of Treg cells (f). Coinhibition of PD‐1/PD‐L1 enhanced the generation of CD107‐expressing CD8⁺ T cells (g). (DC, dendritic cell; Treg, regulatory T cell; *P < 0·05, **P < 0·01).

The effects of PD‐L1/PD‐1 silencing was then evaluated on apoptosis of T cells in co‐culture of tumor‐lysate‐loaded DCs with tumor‐derived T cells. The results showed that T‐cell exposure to PD‐L1‐silenced DCs had significantly lower apoptotic effects compared with T cells primed with control DCs. Similar results were achieved when PD‐1‐silenced T cells were stimulated with control PD‐L1‐silenced DCs. Moreover, concomitant suppression of PD‐L1 and PD‐1 could synergistically reduce apoptosis of T cells as shown in Fig. 4(b, c).

The impact of PD‐L1 silencing was subsequently investigated by secretion of various cytokines from T cells in co‐culture of tumor‐lysate‐loaded DCs and T cells during 96 hr. The results indicated that silencing of PD‐L1 markedly enhanced the secretion of IFN‐γ (T helper type 1 marker) after 72 and 96 hr of co‐culture. Moreover, down‐regulation of PD‐1 increased the secretion of this cytokine. However, there was no synergistic effect when PD‐1 and PD‐L1 were suppressed at the same time in T cells and DCs, respectively (Fig. 4d). Evaluation of IL‐2 exhibited similar results after 48 hr of culture (Fig. 4e). As shown in Fig. 4(f), the impact of PD‐1/PD‐L1 silencing on secretion of TNF was similar to secretion of IFN‐γ. Moreover, silencing of PD‐1, PD‐L1, or both of them slightly (non‐significant) increased the generation of IL‐17 (Fig. 4g). Evaluation of inhibitory cytokines including IL‐10 and transforming growth factor‐β (Treg‐derived cytokines) showed that silencing of PD‐1, PD‐L1, or PD‐1/PD‐L1 potently reduced the generation of these cytokines; however, there was no significant difference between PD‐1‐silenced T cells and the PD‐1/PD‐L1‐silenced group (Fig. 4h, i).

Finally, we evaluated the impact of PD‐1/PD‐L1 silencing on the induction of Treg phenotype in T cells co‐cultured with tumor‐lysate‐loaded DCs. Our results showed that silencing of either PD‐1 in T cells or PD‐L1 in DCs (to a lesser extent) could decrease the induction of CD4⁺ FoxP3⁺ Treg cells. On the other hand, co‐silencing of PD‐1/PD‐L1 could significantly suppress the induction of Treg cells (Fig. 4j, k). Moreover, to analyze the potential of PD‐1/PD‐L1 silencing for induction of cytotoxic T CD8⁺ cells, we analyzed the frequency of T CD8⁺ CD107⁺ cells following co‐culture of T cells and tumor‐lysate‐loaded DCs. The results showed that silencing of either PD‐L1 or PD‐1 significantly increased the frequency of CD8⁺ CD107⁺ T cells. Further, concomitant silencing of PD‐1/PD‐L1 had higher impact on induction of these cells compared with either of PD‐L1 or PD‐1 silencing (Fig. 4l).

Discussion

Various immunosuppressive mechanisms used by tumor cells in their microenvironment enhance the escape of tumor cells from the immune system and decrease the efficiency of current anti‐cancer immunotherapeutic approaches. The important role of immune checkpoint molecules in the suppression of anti‐tumor immune responses means that they have attracted extensive attention during the last two decades, leading to the awarding of the 2018 Nobel prize to investigators involved in this issue (James Allison and Tasuku Honjo). The PD‐1/PD‐L1 axis is one of the most important immunosuppressive mechanisms in the tumor microenvironment. Expression of PD‐L1 on cancer cells and antigen‐presenting cells, as well as the expression of PD‐1 on tumor‐infiltrating T cells, and their interaction may result in suppression of both cells and lead to cancer progression. PD‐1/PD‐L1 expression was also one of the most important impediments for the efficacy of DC vaccines in cancer immunotherapy. Consistently, it is demonstrated that PD‐L1 expression levels can predict the response to treatment with the DC vaccine in patients with glioblastoma.28 Several investigators have tried to block PD‐L1 in DCs or PD‐1 in T cells to increase the efficacy of anti‐cancer DC vaccines. However, none of them has investigated the impact of concomitant suppression of both PD‐1 and PD‐L1 at the same time. We decided to suppress the expression of both PD‐1 and PD‐L1 molecules by using a novel therapeutic strategy via siRNA‐loaded CDS NPs. Synthesized NPs had acceptable physicochemical characteristics including the size, zeta potential, siRNA loading capacity, low toxicity, and high cellular uptake (Fig. 1). It should be noted that the small size of our synthesized NPs (67·5 nm) has made them potent tools for using in vivo, because NPs >100 nm may be phagocytosed by macrophages. The gene silencing function of siRNA‐loaded NPs was also efficient (Fig. 2). Although after treatment with siRNA‐loaded NPs, 15% of DCs were PD‐L1⁺ (Fig. 2g), analysis of PD‐L1 intensity on DCs showed that PD‐L1 expression intensity has significantly decreased (more than 10‐fold) compared with untreated cells (Fig. 2h). Similar patterns of frequency and intensity were observed following treatment of T cells with PD‐1 siRNA‐loaded NPs (Fig. 2j, k). Therefore, despite the DCs and T cells still expressing PD‐L1 or PD‐1, the low intensity of these molecules makes them unable to trigger robust signaling.

Silencing of PD‐L1 in DCs potently improved antigen‐presenting characteristics as determined by assessment of DC viability, expression pattern of surface molecules, cytokine production profile, and migration capacity. Results showed that PD‐L1 could improve the survival of DCs slightly. Survival is an important challenge in DC therapy because it is demonstrated that only small part of injected DCs can reach the tumor site. Therefore, it will be critical to improve DC survival before injection. On the other hand, capability of DCs to present antigen and deliver stimulatory signal to T cells is also an important issue on this topic. Accordingly, our results showed that down‐regulation of PD‐L1 in DCs significantly increased expression of HLA‐DR and CD86 (Fig. 3c) implying the potential of DCs to prime T cells. Cytokines are other important second signals provided by DCs for T cells. The balance between IL‐12 and IL‐10 is an important index for stimulation of T cells. Whereas IL‐12 is an indicator of potent DC vaccine, IL‐10 exhibits the tolerogenic phenotype of DCs. The results showed that PD‐L1 silencing enhances the IL‐12/IL‐10 production ratio, demonstrating DC vaccines with high T‐cell priming potential (Fig. 3h).

Dendritic cells migrate to draining lymph nodes through interaction of CCR7 expressed on DCs with CCL21 secreted in the paracortex of lymph nodes to present antigens to T cells. The ability of DC vaccines to respond to lymph node‐derived CCL21 is also critical to have efficient DC‐based immunotherapy. Interestingly, our results showed that silencing of PD‐L1 robustly enhanced the migratory behavior of DCs toward CCL‐21 chemokine, in vitro (Fig. 3i).

One important fact observed in experiments evaluating the impact of PD‐1/PD‐L1 silencing on T‐cell priming function of DCs (Fig. 4) was the greater importance of PD‐1 suppression compared with PD‐L1. Increased expansion of PD‐1‐silenced T cells even in co‐culture with control DC may indicate greater importance of PD‐1 silencing on T cells in comparison with PD‐L1 silencing on DCs for expansion of T cells (Fig. 4a). A similar phenomenon was observed following the evaluation of apoptosis, cytokine production, Treg cell differentiation, and CD107 cytotoxic T lymphocyte induction (Fig. 4). These observations may be related to expression of PD‐L2 on DCs, which triggers PD‐1 signaling to some extent in T cells, whereas PD‐1 silencing permanently suppresses its signaling in T cells even in the presence of PD‐L1 or PD‐L2.

As mentioned, several investigators have tried to separately suppress PD‐L1 or PD‐1 to have an efficient DC‐based cancer immunotherapy. Accordingly, Ge et  al. have shown that blockage of PD‐L1 by mAbs enhanced the proliferation, maturation, and IL‐12 secretion by DCs and increased T‐cell responses in a breast tumor‐bearing human‐SCID model, which was consistent with our results. Similar to our findings, they showed that T cells secrete higher levels of IFN‐γ, have lower apoptosis, and a higher proliferation rate when they are primed by DCs in the presence of anti‐PD‐L1 antibodies.29 Similarly, Van der Waart et al.12 demonstrated that siRNA‐mediated down‐regulation of PD‐L1/2 in DC vaccines increased the proliferation of CD8⁺ T‐cell responses. It has also been shown that PD‐L1/PD‐L2‐silenced DCs by the siRNA‐loaded Cationic Lipid SAINT‐18 NPs had higher potential to increase T‐cell proliferation and secretion of IFN‐γ, IL‐2, and TNF than normal DCs, which was consistent with our results.11 In another study, it has been shown that silencing of PD‐L1/PD‐L2 in human monocyte‐derived DCs by DLin‐KC2‐DMA‐containing NPs markedly increased ex vivo antigen‐specific CD8⁺ T‐cell response.10 Van den Bergh et al.30 have also shown that transpresentation of IL‐15 can increase the efficacy of siRNA‐mediated PD‐L1/PD‐L2‐silenced human DCs to increase the proliferation and secretion of IFN‐γ and TNF. On the other hand, it has been shown that blockade of PD‐1 by anti PD‐1 mAb increased T‐cell responses to autologous dendritic/myeloma fusion vaccine, ex vivo.14 Other investigators have also suggested that concomitant blockage of PD‐1 and TIM‐3 in tumor‐infiltrating CD8⁺ T cells promotes their response to autologous DC vaccines, ex vivo.13 As discussed in the above‐mentioned studies, silencing of PD‐L1 in DCs or PD‐1 in T cells could increase the T‐cell priming potential of DCs and enhance T‐cell responses. However, there is no study regarding the concomitant silencing or blocking of both PD‐L1 and PD‐1 in order to increase T‐cell responses by DC vaccines. Our results showed that co‐silencing of PD‐1/PD‐L1 could synergistically enhance T‐cell responses, suggesting an appropriate therapeutic strategy for future pre‐clinical and clinical studies (Fig. 5). Our results imply that down‐regulation of PD‐L1 in DC vaccine before its administration can significantly enhance its maturation and stimulatory function. On the other hand, down‐regulation of PD‐1 on tumor‐infiltrating T cells can enhance anti‐tumor responses induced by PD‐L1‐silenced DC vaccines. Targeted silencing of PD‐1 in tumor‐resident T cells may be possible by using potent nano‐based delivery systems such as the CDS NPs used in our study. These results propose the evaluation of this anti‐cancer therapeutic approach in pre‐clinical studies, which is the goal of our subsequent study.

Figure 5.

Silencing of programmed cell death protein 1 (PD‐1)/programmed death ligand 1 (PD‐L1) molecules enhances T‐cell responses to DCs. Down‐regulation of PD‐L1 in DCs by using small interfering RNA (siRNA) ‐loaded nanoparticles (NPs) increases their maturation and survival, and stimulatory molecules. Silencing of PD‐1 in T cells also enhances their responses and survival. Co‐culture of tumor‐lysate‐loaded PD‐L1‐silenced DCs with PD‐1‐silenced T cells synergistically enhances T‐cell responses.

Disclosures

There is no conflict of interest.