A Dual Immunotherapy Nanoparticle Improves T-Cell Activation and Cancer Immunotherapy

Abstract

Combination immunotherapy has recently emerged as a powerful cancer treatment strategy. A promising treatment approach utilizes coadministration of antagonistic antibodies to block checkpoint inhibitor receptors, such as antiprogrammed cell death-1 (aPD1), alongside agonistic antibodies to activate costimulatory receptors, such as antitumor necrosis factor receptor superfamily member 4 (aOX40). Optimal T-cell activation is achieved when both immunomodulatory agents simultaneously engage T-cells and promote synergistic proactivation signaling. However, standard administration of these therapeutics as free antibodies results in suboptimal T-cell binding events, with only a subset of the T-cells binding to both aPD1 and aOX40. Here, it is shown that precise spatiotemporal codelivery of aPD1 and aOX40 using nanoparticles (NP) (dual immunotherapy nanoparticles, DINP) results in improved T-cell activation, enhanced therapeutic efficacy, and increased immunological memory. It is demonstrated that DINP elicits higher rates of T-cell activation in vitro than free antibodies. Importantly, it is demonstrated in two tumor models that combination immunotherapy administered in the form of DINP is more effective than the same regimen administered as free antibodies. This work demonstrates a novel strategy to improve combination immunotherapy using nanotechnology.

Combination immunotherapy has emerged as a powerful new strategy in cancer treatment.^[1] Clinical data suggest that combination immunotherapy regimens that enhance T-cell activation are effective in treating metastatic disease.^[2] Currently, the most effective combination immunotherapeutic regimens consist of combining multiple antagonistic antibodies that target checkpoint inhibition receptors.^[3] However, some combination checkpoint blockade agents demonstrate significant autoimmune-mediated toxicity.^[4] Consequently, recent efforts have focused on combining immune checkpoint blockade agents with T-cell agonists, because these combination immunotherapy strategies elicit less immune-mediated toxicity in the clinical setting.^[5,6]

A particularly promising combination immunotherapeutic regimen is the coadministration of aPD1 and aOX40 to block T-cell inhibition and induce T-cell activation, respectively. With this strategy, maximum T-cell activation would be expected when the T-cells are able to bind both agents (aPD1 and aOX40) simultaneously.^[5,7,8] However, standard administration of these therapeutics as free antibodies results in only a subset of the T-cells binding to both aPD1 and aOX40 (Figure 1a). Moreover, it is likely that the T-cells bind to each agent sequentially rather than simultaneously. We hypothesized that single binding events/sequential binding resulted in suboptimal T-cell activation, treatment efficacy, and immune memory formation when compared to simultaneous binding of both aPD1 and aOX40. We further theorized that we could increase the spatiotemporal precision of aOX40 and aPD1 codelivery to T-cells using a dual-immunotherapy NP (DINP) platform, thereby promoting simultaneous dual-therapeutic binding events.^[9]

Figure 1.

Dual immunotherapy nanoparticle (DINP) conjugated with aPD-1 and aOX40 bind to its target proteins simultaneously. a) Schematic depicting DINP facilitated enhancement of combination immunotherapy. PD-1 is a coinhibitor of T-cell activation. OX40 is costimulator of T-cell activation. DINP blocks the negative signal (red arrow) and stimulates the positive signal (green arrow) simultaneously, leading to the best T-cell activation for tumor killing; while the mixture of free antibodies results in single binding events or a subset of the T-cells binding to both aPD1 and aOX40. b) Representative images depicting nanoparticles before and after dual antibody conjugation. Scale bar: 100 nm. c) Quantification of nanoparticle size and zeta potential change following DINP fabrication. Data represent mean ± standard deviation (SD) (n = 3). d) Flow cytometric analysis assessing the ability of DINP to bind to OX40 and PD1 proteins.

DINP was formulated by conjugating aPD1 and aOX40 to maleimide-terminated poly(ethylene glycol)-b-poly(lactide-co-glycolide) (PEG-PLGA) NP using thiol-maleimide chemistry. We aimed to achieve 1:1 aPD1 to aOX40 ratio due to our lack of knowledge about the best ratio determined by several factors such as actual amount and distribution of PD1 and OX40 receptors on T-cells. We also hypothesized that increased T-cell activation would be associated with high density of binding sites per NP. We were able to achieve a sizeable number of binding sites per NP by incubation of 200 µg mL⁻¹ aPD1 and 100 µg mL⁻¹ aOX40 to 1 mg mL⁻¹ NP, resulting in 49.1 ± 5.5 µg of aPD1 and 44.0 ± 6.0 µg of aOX40 conjugated to per mg NP (Table S1, Supporting Information). Physical characterization of DINP demonstrated a spherical morphology with an average hydrodynamic diameter of 166.9 ± 6.5 nm and a negatively charged surface (Figure 1b,c). As the hydrodynamic diameter of NP was affected by the surface bound species and electrical outer layer, we also measured DINP size using transmission electron microscopy (TEM) images (Figure S1, Supporting Information). The average size was 67.7 ± 11.2 for naked NP and 100.1 ± 18.4 for DINP, indicating almost a single layer of antibodies was conjugated to DINP. To confirm that at least some of the aPD1 and aOX40 antibodies were properly oriented and capable of binding to their respective ligands, we incubated DINP with fluorescently labeled recombinant murine PD1 or OX40 Fc chimeric proteins that bind aPD1 and aOX40, respectively. Using flow cytometry, we showed that aOX40-conjugated NP (aOX40-NP) and aPD1-conjugated NP (aPD1-NP) were able to bind to their corresponding proteins, while DINP was able to simultaneously bind to both proteins, confirming proper orientation (Figure 1d; Figure S2, Supporting Information).

Next, we assessed the ability of DINP to engage immunoregulatory receptors on T-cells and elicit activation in vitro. Antigen-expanded OT1 CD8+ T-cells were cultured in conditions previously demonstrated to mimic exhaustion, then coincubated with B16-OVA tumor cells in media containing different concentrations of DINP or a mixture of free aPD1 and aOX40 antibodies. Following incubation, T-cell activation was assessed by IFN-γ Enzyme-Linked ImmunoSpot (ELISpot).^[10] We found that DINP-treated T-cells demonstrated both greater number of IFN-γ producing cells and higher overall activity of IFN-γ production compared to cells treated with equivalent amounts of free antibodies across all treatment concentrations (Figure 2a,b). To assess whether DINP-treated T-cells also demonstrated enhanced antitumor activity in vitro, we evaluated the killing of B16-OVA tumor cells by DINP-treated OT1 CD8+ T-cells. DINP-treated T-cells were significantly more effective at killing B16-OVA than T-cells treated with free antibodies (Figure 2c). Our data suggest that DINP is more effective in inducing T-cell activation and cytotoxicity than conventional dual-antibody therapy in vitro.

Figure 2.

DINP improves combination immunotherapy facilitated CD8+ T-cell activation and tumor cell killing in vitro. a,b) Activation of OT1 CD8+ T-cells is enhanced following combination immunotherapy treatment as assessed by IFN-γ ELISpot (a), and IFN-γ activity (b). c) B16 ova viability following coincubation with combination immunotherapy treated OT1 CD8+ T-cells. Statistical significance was assessed using the Mann Whitney test. Data represent mean ± standard deviation (SD) (n = 8). *P < 0.05, **P < 0.01, ***P < 0.001.

To investigate whether DINP can improve combination immunotherapy in vivo, C57BL/6 mice bearing bilateral flank B16–F10 melanoma tumors were treated with various immunotherapy regimens. Since OX40 is exclusively expressed by activated T-cells,^[7,11] mice were first immune-primed with a single dose of aPD1^[12] (200 µg, intraperitoneally) and radiotherapy (10 Gy) to one of the flank tumors.^[13] Animals were then given aOX40-NP, aPD1-NP, DINP, or a mixture of free aPD1 and aOX40 antibodies intravenously (Figure S3, Supporting Information). The therapeutic efficacy of each treatment arm was assessed by measuring the growth rate of non-irradiated tumors. Animals treated with DINP demonstrated the highest immunotherapeutic response rates across all treatment groups with a cure rate of 30% (Figure 3a,b; Figure S4, Supporting Information). Importantly, 5/6 of cured mice successfully resisted tumor rechallenge, indicating that the treatment strategy is capable of inducing durable antitumor immunological memory formation (Figure 3c).

Figure 3.

DINP improves the efficacy of combination immunotherapy in vivo. a) Individual tumor growth curves of B16F10 tumors present in animals treated with nanoparticle mono-immunotherapy or combination immunotherapy administered as free antibodies or DINP (100 µg anti-PD1 + 100 µg anti-OX40 per dose, two doses in total). b) Average tumor growth curves and survival curves of animals shown in (a). Tumor growth over time was compared by two-way ANOVA (P < 0.0001) followed by Bonferroni’s multiple comparison tests. Data represent mean ± standard error of the mean (SEM) (n = 8–10). Differences in survival were determined for each group by the Kaplan–Meier method and the P value between mixture of free antibodies and DINP was calculated by the log-rank test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. c) Survival curves of DINP-treated cured animals following tumor rechallenge. d) Individual growth curves of orthotopic 4T1 tumors present in animals treated with combination immunotherapy administered as free antibodies, a mixture of aOX40-NP and aPD1-NP, or DINP (100 µg anti-PD1 + 100 µg anti-OX40 per dose, two doses in total). e) Average tumor growth curves and survival curves of animals shown in (d). Tumor growth over time was compared by two-way ANOVA (P < 0.0001) followed by Bonferroni’s multiple comparison tests. Data represent mean ± standard error of the mean (SEM) (n = 10). Differences in survival were determined for each group by the Kaplan–Meier method and the P value between mixture of free antibodies and DINP was calculated by the log-rank test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To validate our in vivo results, we evaluated the therapeutic efficacy of DINP treatment using an orthotopic 4T1 breast cancer model. Animals bearing bilateral mammary fat pad 4T1 tumors were immune-primed with aPD1 and radiotherapy and subsequently treated with different immunotherapeutic regimens. As seen in the B16–F10 model, we observed that DINP treatment resulted in the greatest control of tumor burden and increased the survival time of tumor-bearing animals by over 20% when compared to all other treatment arms (Figure 3d,e). Importantly, we found that DINP treatment was significantly more effective than the combination of aPD1-NP and aOX40-NP, demonstrating that the enhanced immunotherapeutic response observed in our DINP treatment arm cannot simply be attributable to properties specific to NP.

To confirm that the improved immunotherapy is attributed to the simultaneous binding by DINP rather than to NP delivery effects, we assessed the therapeutic efficacy of aPD1-NP plus free aOX40, aOX40-NP plus free aPD1, and DINP in the B16F10 melanoma tumors model. As seen in Figure S5 of the Supporting Information, DINP demonstrated the highest response rate (100%) and significantly better than aOX40-NP plus free aPD1 in tumor inhibition. The survival curve showed that the tumor-free survival rate after DINPs treatment was 30%, compared to 10% after the treatment by aPD1-NP plus free aOX40 or by aOX40-NP plus free aPD1. Taken together, these data suggest that DINP enhances combination immunotherapy.

We next sought to determine the mechanism by which DINP augments the antitumor immune response. First, we compared the relative number of T-cells simultaneously receiving aPD1 and aOX40 therapy following DINP versus free antibody therapy. Immune-primed animals bearing bilateral B16F10 tumors were given fluorescently labeled DINPs or fluorescently labeled free antibodies. T-cells in spleen and tumor were harvest 2 h post therapeutics administration. Using flow cytometry, we compared the number of T-cells that possess both fluorescence label between the DINP and the free antibody treatment arms (Figure S6, Supporting Information). We found that DINP treatment provided a significantly higher percentage of T-cells with both aPD1 and aOX40 binding in total T-cell population when compared to free antibody treatment, either in spleen (25.5% ± 0.7% vs 7.7% ± 0.9%) or in tumor (20.1% ± 3.0% vs 4.9% ± 0.4%).

To determine if DINP-mediated codelivery of aPD1 and aOX40 translates into increased T-cell activation and expansion in vivo, we quantified the overall number of tumor infiltrating T-cells in B16F10-bearing animals treated with various immunotherapeutic regimens. Mice treated with DINP had a significantly higher number of CD8+ T-cells (median = 85.2%) compared to other treatments, including mixture of free aPD1 and aOX40 (median = 68.5%) (Figure 4a; Figure S7, Supporting Information). This finding was further validated by immunoflourescence microscopy imaging of excised tumors (Figure 4c). Furthermore, DINP treatment increased the ratio of CD8+ to regulatory T-cells infiltrating the tumor (median = 19.0) compared to free antibody therapy (median = 6.9), which has been shown to be an important prognostic marker for survival in human melanoma^[14] (Figure 4b). Of total CD8+ T-cells, the median frequency of effector memory T-cells among DINP-treated animals was 97.5%, compared to 96.0% in dual antibody-treated animals. Additionally, the ratio of effector memory to central memory T-cells was significantly higher among DINP-treated animals (median = 54.4) compared to free antibody-treated animals (median = 23.0). Taken together, this higher effector memory frequency and increased effector-to central-memory ratio observed in DINP-treated animals is suggestive of an antigen-driven T-cell response with greater ongoing antitumor effector activity among the tumor infiltrating T-cell population.

Figure 4.

DINP improves the efficacy of combination immunotherapy in vivo by improving CD8+ T-cell expansion and tumor infiltration. a) Scatter dot plot with median line assessing the relative abundance of CD8+ T-cells, and CD4+FOXP3+ regulatory T-cells (Treg) subpopulations in un-irradiated tumors of animals undergoing different immunotherapy treatment regimens by flow cytometric analysis. T-cells were defined as CD3+. Statistical significance was assessed using the Mann Whitney test (n = 5–7). *P < 0.05, **P < 0.01. b) Scatter dot plot with median line assessing the relative abundance of central memory (T_cm) and effector memory (T_EM) in CD8+ T-cells subpopulations in un-irradiated tumors of animals undergoing different immunotherapy treatment regimens by flow cytometric analysis. Statistical significance was assessed using the Mann Whitney test (n = 5–7). *P < 0.05, **P < 0.01. c) Representative immunofluorescent images of tumors analyzed in (c). Scale bar in first row = 100 µm; scale bar in second row = 25 µm. d) Individual growth curves of tumors in animals treated with DINP combination immunotherapy with or without CD8+ T-cell or NK cell depletion. e) Average tumor growth curves and survival curves for each treatment arm shown in (d). Tumor growth over time was compared by two-way ANOVA (P < 0.0001) followed by Bonferroni’s multiple comparison tests. Data represent mean ± standard error of the mean (SEM) (n = 9–10). Differences in survival were determined for each group by the Kaplan–Meier method. P value was calculated by the log-rank test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To confirm that DINP-mediated enhancement of combination immunotherapy is attributable to increased activation of CD8+ T-cells, we evaluated DINP treatment in CD8+ T-cell depleted B16–F10-bearing animals (Figure 4d,e). Anti-CD8 depleting antibodies were administered to immune-primed B16F10-bearing animals prior to DINP treatment. We found that CD8+ T-cell depletion resulted in near-complete abrogation of DINP treatment efficacy. By contrast, the effect of NK cell depletion using anti-NK1.1 on DINP efficacy was significantly less pronounced, compared to the CD8 depletion. These data suggest that DINPs effect is primarily through CD8+ T-cells and the cytotoxic capacity of the adaptive immune system. Taken together, our data show that combination immunotherapy given in the form of DINP improves treatment response and antitumor immunity by increasing antigen-driven T-cell activation and effector function, resulting in a more immunoreactive tumor microenvironment.

In summary, we demonstrate that codelivery of synergistic immunotherapeutics using NPs can improve the treatment response of combination immunotherapy. We showed that DINP induces higher T-cell activation than free antibody immunotherapeutics. Importantly, we demonstrated that DINP was significantly more effective than free antibody therapeutics or single therapeutic NPs. While current research is focused on the development of new immunotherapeutics, our work shows that we can significantly improve treatment efficacy through NP delivery. This work carries important implications for cancer immunotherapy as it details a novel strategy and can result in a new class of highly effective immunotherapeutics.

Experimental Section

Materials

mPEG-PLGA (AK029; LA:GA = 50:50 (w:w);M_W: ≈3000: 36 000 Da), PLGA-PEG-Mal (Maleimide) (AI110; LA:GA = 50:50; M_W:: ≈30 000–5000 Da) were obtained from Polyscitech. Anti-PD-1 (clone: RMP1-14), anti-OX-40 (CD134) (clone: OX-86), anti-CD8a (clone: 2.43), anti NK1.1 (clone: PK136) were from BioXcell. Recombinant mouse PD-1 chimera protein, recombinant mouse OX40/TNFRSF4 Fc chimera protein, and goat antihuman IgG (H+L) affinity purified PAb were from R&D Systems. Goat antirat IgG (H+L) secondary antibody horseradish peroxidase (HRP), Alexa Fluor 488 (AF488) and Alexa Fluor 647 (AF647) protein labeling kit, 1-Step ultra TMB-ELISA substrate solution, clear flat-bottom immuno 96-well plates were from ThermoFisher Scintific. All antibodies used for flow cytometric assays were from BD Biosciences and are listed in the Table S2 of the Supporting Information. All other chemicals were obtained from Sigma-Aldrich unless otherwise noted.

Cell Lines

The B16–F10, B16-OVA, and 4T1 cell lines were acquired from ATCC, where these lines were authenticated using morphology, karyotyping, and polymerase chain reaction (PCR)-based approaches and tested for mycoplasma. B16–F10 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco) supplemented with 10% fetal bovine serum (Mediatech), 100 U mL⁻¹ penicillin and 100 µg mL⁻¹ streptomycin (Mediatech), and 2 × 10⁻³ m l-glutamine (Gibco). 4T1 cells were cultured in RPMI Medium 1640 (Gibco) supplemented with 10% fetal bovine serum (Mediatech), 100 U mL⁻¹ penicillin, and 100 µg mL⁻¹ streptomycin (Mediatech), and 2 × 10⁻³ m l-glutamine (Gibco). The cell cultures were maintained below 50% confluence and early-passage cultures (between 4 and 9) were utilized for experiments.

Preparation of DINP

The DINP was synthesized by a two-step method. First, PLGA-PEG-Mal nanparticles were synthesized through nanoprecipitation technique.^[15] mPEG-PLGA and PLGA-PEG-Mal (7:3 weight ratio) were dissolved into acetone with a final polymer concentration of 10 mg mL⁻¹. The organic phase was added dropwise into the aqueous phase (endotoxin free H₂O) through a syringe under the oil to water ratio of 1:2. The solution was stirred at room temperature under a vacuum until the acetone completely evaporated. The solution were centrifuged and washed with endotoxin free H₂O. The PLGA-PEG-Mal NPs were then conjugated with PD1 and/or OX40 antibodies through maleimide-thiol click chemistry.^[16] Different feeding ratios of the antibodies (Table S1, Supporting Information) were dissolved into ethylenediaminetetraacetic acid (EDTA) (5 × 10⁻³ m) containing phosphate-buffered saline (PBS) (pH = 7.4) buffer and 0.75 × 10⁻⁶ m tris(2-carboxyethyl)phosphine was added. The solution was gently shaken for 10 min at room temperature and PLGA-PEG-Mal NPs were added with a final NP concentration of 1 mg mL⁻¹. The reaction was lasted in different incubation conditions (Table S1, Supporting Information). The final product was centrifuged and washed with endotoxin free H₂O or PBS and supernatant was collected.

Quantification of Conjugated PD1 and/or OX40 Antibodies on Nanoparticles

The conjugated antibodies were quantified by calculating the difference from feeding amount and supernatant amount. The quantification of antibodies in the supernatant was analyzed by a standard sandwich ELISA assay.^[17] Specially, immuno 96-well plates were coated with 2 µg mL⁻¹ of goat antihuman IgG antibody, followed by the addition of 100 ng mL⁻¹ recombinant mouse PD-1 chimera protein or recombinant mouse OX40/TNFRSF4 Fc chimera protein, which were used as the capture agent for the bioactive anti-PD1 or anti-OX40 in the supernatant. 200 ng mL⁻¹ of HRP-conjugated goat antirat IgG was then added as the detection antibody, followed by an HRP-sensitive colorimetric substrate.

Characterization of DINP

Intensity-average diameter (D_h, also known as hydrodynamic diameter) and mean zeta potential (mean ζ) of PLGA-PEG-Mal nanoparticles before (NP) and after the conjugation of antibodies (DINP) were analyzed by dynamic light scattering and an aqueous electrophoresis method using a Zetasizer Nano ZS Instrument (Malvern, Inc.). All measurements were based on the average of three separate measurements. The morphology of the nanoparticles was recorded by TEM (Zeiss EM 910).

Binding Activity of DINP

The recombinant mouse PD-1 chimera protein or recombinant mouse OX40/TNFRSF4 Fc chimera protein were first labeled by AF488 or AF647 protein labeling kit, respectively. Nanoparticles conjugated with anti-PD1 and/or anti-OX40 were blocked by 1% bovine serum albumin (BSA) PBS (pH = 7.4) buffer for 1 h. 200 µg NPs were incubated with 1 µg fluorescencent PD-1 and/or OX40 proteins in PBS with 1% BSA for 2 h. The solution was washed with washing buffer (PBS (pH = 7.4) containing 0.05% Tween-20) four times. The binding between NPs and proteins were tested by flow cytometry.

Animal Study

For all animal studies, eight-week-old female C57BL/6 mice (The Jackson Laboratory) were used. All animal work was approved and monitored by the University of North Carolina Animal Care and Use Committee.

Sample sizes were calculated based on the preliminary data, with a calculated effect size of 1.821. The nonparametric analog of this effect size can be stated in terms of p₁ = Pr (X < Y), or an observation in Group X is less than an observation in Group Y when H₁ is true. The null hypothesis being tested is p₁ = 0.5. For effect size 1.821, p₁ = 0.099. A sample size of at least 8 in each group will have 80% power to detect a probability of 0.099 that an observation in Group X is less than an observation in Group Y, using a Wilcoxon (Mann-Whitney) rank-sum test, with a 0.05 two-sided significance level. Mice were assigned to treatment groups based on cage numbers.

Two perpendicular diameters were measured with a caliper and tumor volumes were calculated using the formula V = 0.52 × a × b^2, where a and b are the larger and smaller diameters, respectively. The tumor volumes were assessed every 2 d. Two independent researchers assessed tumor volume over time with one researcher blinded to the treatment group assignments. Statistical differences in average tumor growth curves were determined by two-way ANOVA using variables of time and volume. Differences in survival in each group were determined using the Kaplan–Meier method and the overall P value was calculated by the log-rank test using the GraphPad Prism 6.0. P value: *, P < 0.05; **, P < 0.01; ***, P < 0.001,****, P < 0.001.

Efficacy of DINP in Improving Tumor Immunotherapy

In the melanoma tumor model, 75 000 B16–F10 cells were suspended in DMEM, mixed with an equal volume of Matrigel (BD Biosciences), and subcutaneously injected on the left flank of C57BL/6 mice on day 0 and the right flank on day 2. 200 µg aPD-1 was intraperitoneally injected into animals on day 4. The left flank tumors were irradiated with 10 Gy on day 5 using an X-RAD 320. A lead shield protected the rest of the animal. DINP, the mixture of antibodies and other control NPs (100 µg anti-PD1 and/or 100 µg anti-OX40 in 200 µL PBS) were injected intravenously on day 6 and 9. For the survival mice, at one or two month(s) postprimary inoculation, secondary challenge of 200 000 B16–F10 cells was inoculated into the right flank and monitored without additional therapy. In the breast tumor model, 100 000 4T1 cells were suspended in RPMI Medium 1640, mixed with an equal volume of Matrigel (BD Biosciences), and injected on the left fourth mammary fat pad of BALB/c mice on day 0 and the right fourth mammary fat pad on day 2. The other steps were kept the same. In the depletion study, mice were treated by DINP with the same procedure. 400 µg per dose of anti-CD8a or anti-NK1.1 were injected intraperitoneally on day 10.^[18]

T -Cell Phenotype Analysis

In the study of T-cell phenotype, 100 000 B16–F10 cells were suspended in DMEM, mixed with an equal volume of Matrigel (BD Biosciences), and subcutaneously injected on the left flank of C57BL/6 mice on day 0 and the right flank on day 1. 200 µg aPD-1 was intraperitoneally injected into animals on day 6. The left flank tumors were irradiated with 10 Gy on day 7 using an X-RAD 320. A lead shield protected the rest of the animal. DINP, the mixture of antibodies and other control NPs (100 µg anti-PD1 and/or 100 µg anti-OX40 in 200 µL PBS) were injected intravenously on day 8 and 11. Mice were sacrificed on day 15. Tissues were homogenized using the GentleMACs Dissociator and the samples were passed through a 70 × 10⁻⁶ m cell strainer, followed by homogenization by using a 5 mL syringe plunger. The samples were centrifuged for 7 min at 1200 rpm, 4 °C, decanting the supernatant. Samples were washed and resuspended in cold DPBS and transferred onto a 96 well V-bottom plate. Cells were resuspended in FVS510 viability stain (1:1000 dilution in 200 µL DPBS) for 40 min on ice. Wells not receiving viability staining were resuspended in DPBS. Cells were washed twice in staining buffer (0.02% NaN3, 2% BSA in DPBS), resuspended in 100 µL Fc block (1:50 dilution in staining buffer), and incubated on ice for 15 min. Antibody master mix was added to samples at 100 µL per sample with final antibody concentrations of

• CD4 FITC (1:75)

• FoxP3 PE (1:75)

• CD44 PerCPCy5.5 (1:75)

• CD62L BV421 (1:75)

• CD3 APC (1:75)

• CD8 APCH7 (1:75)

Cells were incubated on ice for 45 min and washed twice with staining buffer. Cells were fixed and permeabilized using 250 µL fix/perm buffer overnight or for 50 min (eBioscience FoxP3 buffer set). The following morning, cells were stained in 100 µL FoxP3 PE ab diluted 1:75 in perm wash buffer for 45 min on ice, wash 2× with staining buffer, and read out on a BD LSRFortessa flow cytometer. FlowJo flow cytometry software Version 10 was used for analyses.

Fluorescent Immunohistochemistry Study

Mice were treated the same as in T-cell phenotype analysis. Tissues were fixed in 10% formalin for 72 h and then transferred to 70% ethanol. The slides were deparaffinized, placed in 10% Hydrogen peroxide in methanol for 30 min, and then gently rinsed in deionized water. Immunofluorescence analysis of CD4–CD3–CD8 was performed on paraffin specimens using Mouse anti-CD4 (14-9766 eBioscience), anti-CD3 (A0452 Dako), and anti-CD8 (14-0808 eBioscience). Antigen retrieval was performed on tissue slides with a tris-based buffer (pH 8.5) for 72 min at 100 °C and blocked with a protein block for 20 min at room temperature. The slides were given a hydrogen peroxidase block for 8 min at room temperature and first incubated in the anti-CD4 dilution (1:10) (using Discovery Ab Diluent, 760-108) for 4 h at room temperature, followed by the secondary antibody incubation (Alexa Fluor 647, A21247, goat antirat IgG, 1:100, using Discovery Ab Diluent, 760-108) at room temperature for 44 min. The slides were then given an antibody denaturation step of 95 °C incubation for 12 min.

Following the denaturation, the slides were given another hydrogen peroxidase block for the anti-CD3 dual antibody. The slides were incubated in the anti-CD3 dilution (1:200, using Discovery Ab Diluent, 760-108) at room temperature for 1 h, followed by the secondary antibody (Alexa Fluor 555, A21429, goat antirabbit IgG, 1:100, using Discovery Ab Diluent, 760-108) at room temperature for 44 min.

The anti-CD8 triple antibody was prepared by another hydrogen peroxidase block. The slides were then incubated in the anti-CD8 dilution (1:100, using PSS Discovery Diluent, 760-212) at room temperature for 2 h, followed by the secondary antibody (Alexa Fluor 488, A11006, goat antirat IgG, 1:100, using Discovery Ab Diluent, 760-108) at room temperature for 44 min.

The slides were gently rinsed and placed in Hoescht 33258 Invitrogen solution, 2 µg mL⁻¹ dilution at room temperature for 7 min for 4′,6-diamidino-2-phenylindole (DAPI) staining. The slides were finally coverslipped using Prolong Gold Antifade reagent, P36934 from Life Technologies.

In Vitro Cytotoxicity Assay

OT1 CD8+ T-cells were activated and expanded using the H-2Kb-restricted OVA MHC class I epitope (OVA_257–264; SIINFEKL), according to previously published methods.^[19] Antigen-experienced T-cells were viably frozen and subsequently thawed for experiments, allowing cells to recover overnight. Cells were incubated for 6 d with 100 IU mL⁻¹ recombinant murine IL-2, anti-CD3/28 stimulation beads according to manufacturer protocol (Miltenyi, 130-095-925), and 1 µg mL⁻¹ recombinant murine PD-L1 protein (R&D systems, 1019-B7), with media and reagents changed out every 48 h. On day 7, T-cells were coincubated with B16-OVA tumor cells at a 0.25:1 effector to target ratio in 100 µL media for 48 h, along with 250, 50, 5, 0.5, or 0 µg of free aPD-1/aOX40 or equivalent dose of DINP. After coincubation, nonadherent cells were washed thrice from the plate, and remaining cell viability was measured with a CellTiter-Glo Luminescence kit (Promega, G7570), according to manufacturer protocol.

In vitro ELISpot

OT1 CD8+ T-cells were activated and expanded using the H-2Kb-restricted OVA MHC class I epitope (OVA_257–264; SIINFEKL), according to previously published methods.^[19] Antigen-experienced T-cells were viably frozen and subsequently thawed for experiments, allowing cells to recover overnight. Cells were incubated for 6 d with 100 IU mL⁻¹ recombinant murine IL-2, anti-CD3/28 stimulation beads according to manufacturer protocol (Miltenyi, 130-095-925), and 1 µg mL⁻¹ recombinant murine PD-L1 protein (R&D systems, 1019-B7), with media and reagents changed out every 48 h. On day 7, T-cells were coincubated with B16-OVA tumor cells at a 10:1 effector to target ratio in 100 µL media for 3–6 h, along with 250, 50, 5, 0.5, or 0 µg of free aPD-1/aOX40 or equivalent dose of DINP. After coincubation, nonadherent T-cells were transferred onto an anti-IFN-γ coated ELISpot plate (BD, 551083) and incubated for 72 h before read-out, according to manufacturer protocol. Briefly, ELISpot plates were read out on an AID ELISpot reader (min. size 15, min. gradient > 1), reporting both absolute count and well activity. Well activity is an AID manufacturer-defined measurement of total cytokine release, dependent upon both absolute count and spot signal intensity. Activity values were normalized to the untreated control group.

In Vivo Binding Study

In the binding study, 100 000 B16–F10 cells were suspended in DMEM, mixed with an equal volume of Matrigel (BD Biosciences), and subcutaneously injected on the left flank of C57BL/6 mice on day 0 and the right flank on day 1. 200 µg αPD-1 was intraperitoneally injected into animals on day 7. The left flank tumors were irradiated with 10 Gy on day 8 using an X-RAD 320. A lead shield protected the rest of the animal. Antibodies were first labeled by AF488 (anti-OX40) or AF647 (anti-PD1) protein labeling kit and then conjugated to the NPs by the same procedure in preparation of DINPs. Fluorescence-labeled DINPs or the mixture of fluorescence-labeled antibodies (200 µg anti-PD1 and 200 µg anti-OX40 in 200 µL PBS) were injected intravenously on day 9. Mice were sacrificed 2 h later. Tissues were homogenized using the GentleMACs Dissociator and the samples were passed through a 70 × 10⁻⁶ m cell strainer, followed by homogenization by using a 5 mL syringe plunger. The samples were centrifuged for 7 min at 1200 rpm, 4 °C, decanting the supernatant. Samples were washed and resuspended in cold DPBS and transferred onto a 96-well V-bottom plate. Cells were resuspended in FVS510 viability stain (1:1000 dilution in 200 µL DPBS) for 40 min on ice. Wells not receiving viability staining were resuspended in DPBS. Cells were washed twice in staining buffer (0.02% NaN3, 2% BSA in DPBS), resuspended in 100 µL Fc block (1:50 dilution in staining buffer), and incubated on ice for 15 min. Antibody master mix was added to samples at 100 µL per sample with final antibody concentrations of:

• CD3e BV421 (1:100)

• CD45 BV786 (1:100)

Cells were incubated on ice for 45 min and washed twice with staining buffer and read out on a BD LSRFortessa flow cytometer. FlowJo flow cytometry software Version 10 was used for analyses.