Nanoparticle Encapsulation of Synergistic Immune Agonists Enables Systemic Codelivery to Tumor Sites and IFNβ-Driven Antitumor Immunity

Generation of effective tumor site–specific immunity is pivotal to the success of an immunotherapy and depends on the robust activation of local antigen-presenting cells (APC), such as dendritic cells (DC) and macrophages (1–3). Immunostimulatory small molecules, cytokines, and antibodies delivered to the tumor site can generate powerful local antitumor immune responses individually or in specific synergistic combination (4–7). Systemic delivery enables ready access to the APC-rich perivascular areas of the tumor and has significant advantages over local delivery. Local administration depends on accurate prior knowledge of tumor sites, which is often clinically infeasible, and has limitations in the spreading of therapeutics throughout the tumor microvasculature (8–10). Systemic delivery further enables access to sites of metastasis, which are nearly impossible to clinically detect and treat early and therefore a significant contributor to mortality (11). The challenges to systemic delivery, however, are minimizing off-target toxicity and, in the case of delivering multiple synergistic therapeutics, guaranteeing codelivery to the same target cells (12, 13). Toward this goal, we present a nanotechnology approach to “reengineer” tumor site–specific immunity by coencapsulating two synergistic immune-potentiating agents within a single liposomal nanoparticle (NP) that is delivered systemically to accumulate preferentially within the APC-rich perivascular areas of the tumor and drive an IFNβ-mediated antitumor immune response. Coencapsulation within a NP not only prevents toxic systemic dissemination of these therapeutics, but also guarantees their codelivery to the same target cell (3, 14). We selected the 4T1 murine model of triple-negative breast cancer (TNBC) as an optimal test-bed for this therapy because it is poorly immunogenic with spontaneous metastasis and profound immunosuppression, as is reminiscent of clinical TNBC (15–17). A total of 12%–17% of newly diagnosed early breast cancers are TNBCs, which form the most aggressive subset of breast cancer with a very high risk of recurrence and metastasis, leading to disproportionate mortality (15, 18–24). Our strategy to drive local tumor site–specific immunity can have significant impact on treatment, which is otherwise severely limited.

Effective activation of local APCs plays a central role in antitumor immunity because these cells bridge the innate and adaptive arms of the immune system. Activated APCs help recruit and activate T and B cells that mediate tumor clearance and generate immunological memory (1, 25). APC activation requires both the internalization of tumor antigens, which are readily available within the tumor microenvironment (TME) itself, as well as a strong immune-potentiating stimulus. As such, our goal was to systemically deliver a robust immune-potentiating stimulus via a tumor-draining NP that is readily available and amenable for uptake by APCs. NPs delivered systemically have been shown to be highly endocytosed by perivascular macrophages in tumor sites and remain collected within these cells for prolonged periods of time (26). We hypothesized that harnessing multiple overlapping innate immune pathways could trigger a more potent, synergistic cytokine response. We elected to codeliver two strong inducers of type I IFNs: Stimulator of Interferon Genes (STING) agonist cyclic diguanylate monophosphate (cdGMP; ref. 27), which has gained significant attention in recent years, and Toll-like receptor 4 (TLR4) agonist monophosphoryl lipid A (MPLA; ref. 28), which was clinically approved for use as the first molecular vaccine adjuvant in humans. Both cdGMP and MPLA are widely used as vaccine adjuvants and target host pattern recognition receptors, which recognize conserved, immunogenic molecules from viruses and bacteria (i.e., specific nucleic acids and cell membrane components) to trigger the appropriate immune response. STING agonists are cyclic dinucleotides and small-molecule second messengers that, when free in the cytosol, bind STING machinery to trigger the robust production of type I IFNs (27, 29–32). The engagement of STING machinery has also been shown to activate natural killer (NK) cells that can directly attack tumor cells (2, 33). MPLA is sensed by transmembrane TLR4 and has been shown to trigger a strong proinflammatory Th1 cytokine response (28, 34). Studies have shown that direct intratumoral injection of NPs loaded with STING agonists resulted in sufficient therapeutic efficacy in the B16F10 melanoma model in mice (33, 35). Furthermore, NPs loaded with STING agonists or MPLA have shown promise as vaccine adjuvants that drain to lymph nodes (36).

Specifically, we exploited the leaky endothelium (37, 38) of advanced TNBC tumors to achieve efficient draining of systemically delivered 60-nm cdGMP/MPLA-encapsulated immuno-nanoparticles (immuno-NPs; Fig. 1). For our primary investigations, we selected the 4T1 murine model of metastatic TNBC as a clinically relevant test model, where orthotopic inoculation of cells in the mammary fat pad has been shown to lead to metastasis predominantly in the lungs and lymph nodes (Fig. 1A; ref. 17). To illustrate the broad application of immuno-NPs, we selected an all-purpose, versatile liposome and coencapsulated both hydrophilic cdGMP (within the core) and hydrophobic MPLA (within the lipid bilayer) on the same NP (Fig 1B). We elected to use relatively smaller liposomes in the 60-nm size range because of their established benefits in draining to and retention within tumors (8, 37, 38) and we incorporated a poly(ethylene glycol) (PEG) coating for improved solubility and circulation. The systemic administration of our immuno-NPs resulted in significant intratumoral deposition in both the primary tumor and sites of metastasis, predominantly in APC-rich perivascular regions of the TME. Upon internalization of the NPs, released cdGMP and MPLA triggered a robust, synergistic site-specific cytokine gradient driven largely by IFNβ that resulted in APC- and NK cell–driven local and systemic immune recruitment (Fig. 1C).

To synthesize NPs, lipid films were prepared consisting of 48.5 mol% DOPC (1,2-dioleoyl-sn-glycero-3-phosphocholine, Avanti), 48.5 mol% DPPC (1,2-dipalmitoyl-sn-glycero-3-phosphocholine, Avanti), and 3 mol% mPEG2000-DSPE [methoxy-poly(ethylene glycol)-2000 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N, Laysan Bio]. MPLA (100 μg per 42 μmol lipid) was added prior to film formation (Sigma-Aldrich). Lipid films were hydrated in PBS containing 200 μg cyclic di-GMP (InvivoGen), heated to 60°C for 1 hour, and vortexed for 30 seconds every 10 minutes. Samples were ultra-sonicated on ice for 5 minutes at alternating power settings (7W and 5W every 30 seconds). NPs were dialyzed in PBS for 2–4 hours and stored at 4°C. Size and surface charge were measured via dynamic light scattering (DLS) and zeta potential measurements, respectively (Beckman Coulter). Encapsulated cdGMP was measured using size exclusion high-permeation liquid chromatography (Shimadzu). Stability studies were conducted at 25°C and 37°C. For immuno-NP fluorescence studies, NPs were synthesized by incorporating 0.2 mol% of DiI, DiD, or DiR purchased from Thermo Fisher Scientific.

All animal studies were carried out following institutional, local, state, and federal guidelines under an Institutional Animal Care and Use Committee–approved protocol.

Murine B16F10 melanoma cells and RAW 264.7 macrophages (ATCC) were cultured in DMEM (Gibco) containing 10% FBS (HyClone). Murine 4T1 TNBC cells expressing GFP and luciferase (a gift from the Schiemann Laboratory at the Case Comprehensive Cancer Center) were cultured in RPMI medium (Gibco) containing 10% FBS. Short tandem repeat authentication was used for cell line authenticity. Cell lines were routinely tested and confirmed to be free of Mycoplasma contamination.

For 4T1 studies, BALB/c mice (Jackson Laboratories) were inoculated by orthotopic injection of 5 × 10⁵ 4T1 cells in mammary fat pad #9. Tumors were monitored by bioluminescence imaging (IVIS Spectrum, Perkin Elmer) and caliper measurements. Tumor-bearing animals were treated on day 10 after inoculation. NPs and free immuno-agents were administered by intravenous or intratumoral injection as noted. Anti-PD-1 (250 μg, clone RMP1-14) and anti-CTLA-4 (100 μg, clone 9D9; BioXCell) were administered by intraperitoneal injection. For alanine aminotransferase (ALT)/aspartate aminotransferase (AST) measurements of hepatotoxicity, mice were bled retro-orbitally and serum was analyzed by the University Hospitals Pathology Core (Cleveland, OH).

For B16F10 studies, C57/BL6 mice (Jackson Laboratories) were inoculated by orthotopic subcutaneous injection of 1 × 10⁶ B16F10 cells on the dorsal flank. Tumors were monitored by caliper measurements. Tumor-bearing animals were treated on day 7 after inoculation. NPs were administered by intravenous injection.

For Spectrum studies, fluorescent NPs were prepared with 0.2 mol% DiD (Invitrogen). 4T1 tumor–bearing mice were injected with fluorescent NPs and either live-animal imaging or imaging of whole-excised organs was performed using an IVIS Spectrum. Quantitative biodistribution of NPs in the blood plasma was calculated using a value of 77 mL/kg for average total blood volume of a mouse (Jackson Laboratories).

For fluorescence molecular tomography (FMT) studies, fluorescent NPs were prepared with 1 mol% DSPE-VivoTag-800 (Perkin Elmer). 4T1 tumor–bearing mice were injected with fluorescent NPs, euthanized at noted timepoints, and FMT (Perkin Elmer) was used to image whole-excised organs.

Anti-mouse CD45 (30-F11), Ly-6G (1A8), Ly-6C (AL-21), CD11b (M1/70), CD11c (HL3), F4/80 (BM8), CD49b (DX5), CD3ϵ (145-2C11), CD4 (GK1.5), CD8α (Ly-2, 53-6.7), and CD19 (1D3) flow cytometry antibodies were purchased from BD Biosciences. Flow cytometry analysis was performed 48 hours after treatment. Blood samples were obtained via retro-orbital bleeding and mice were euthanized immediately afterwards and tumors and spleens were harvested. Organs were gently disrupted into single-cell suspensions in progressive steps. Cells were stained to identify immune cell populations with a blocking step using anti-mouse CD18/CD32 and analyzed using a BD FACS LSR II Flow Cytometer (Becton Dickinson). FlowJo software was used to analyze data. For intracellular IFNβ staining, surface-stained cells were fixed and permeabilized (Fix/Perm Wash Buffer, BioLegend), stained with anti-IFNβ (Abcam), and followed by secondary antibody staining.

Six million RAW 264.7 cells were plated in triplicate per well of a 24-well plate and treated with immuno-NPs containing 20 μg/mL cdGMP, 17 μg/mL MPLA, or equivalent amounts of both cdGMP and MPLA. Cell culture supernatants were harvested 24 hours after seeding, clarified by centrifugation at 4°C, and analyzed per the manufacturer's protocols for the presence of IFNα and IFNβ using LumiKine Xpress Bioluminescent Cytokine ELISA Kits from InvivoGen.

4T1 tumor–bearing mice were injected with fluorescent NPs and perfused 24 hours later with PBS and PBS containing 4% paraformaldehyde (Alfa Aesar). Following euthanization, organs were harvested, fixed in 4% PFA/PBS, dehydrated in 30% sucrose/PBS, and embedded and frozen in optimum cutting temperature (Thermo Fisher Scientific). Primary (anti-CD31, anti-CD11c, anti-F4/80, and anti-CD49b) and secondary antibodies were purchased from Thermo Fisher Scientific, Abcam, and BioLegend. Frozen sections, 10 μm in thickness were stained with 1:100–1:150 primary antibodies overnight at 4°C, followed by staining with 1:100–1:150 secondary antibodies for 1 hour at 25°C. Stained sections were mounted with No. 1.5 glass coverslips using Vectashield DAPI aqueous mounting medium (Vector Laboratories) and imaged using a Leica TCS SP8 gated STED Confocal Microscope (Leica Microsystems).

Carbon film grids were treated by glow discharge for 3 minutes. Liposomes were prepared in dilute suspensions (1:5–1:10 dilutions from treatment concentrations) and applied to treated grids and incubated for 30 seconds. Grids were washed 5 times with 5 mmol/L Tris buffer and directly afterwards stained 5 times with 2% uranyl acetate. Imaging was performed using a Tecnai G2 SpiritBT Electron Microscope (FEI) operated at 80 kV.

All statistical analyses are detailed in figure legends. Prism 7 (GraphPad Software) was used to analyze data by one- or two-way ANOVA with Tukey or Sidak post-test. P values less than 0.05 were considered statistically significant. Unless otherwise noted, all values are reported as the mean ± SE of at least three independent biological replicates. In animal studies, at least 5 mice were included in each group.

All data generated and analyzed in this study are included in this article (and its Supplementary Data).

To first define a quantitative rationale for our proposed systemic NP-mediated delivery platform, we investigated the mechanistic efficacy of delivering free immuno-agents both intravenously and intratumorally (Fig. 2). Importantly, while doses of up to 50 μg of free cdGMP administered intratumorally are needed to regress early 4T1 primary tumors (27), to highlight a key advantage of NP encapsulation and systemic delivery, we elected to coadminister just 7 μg of free cdGMP and 6 μg of free MPLA i.v. or i.t., which are equivalent to doses that we aimed to deliver via immuno-NPs for therapeutic efficacy. Mice bearing orthotopic 4T1 primary tumors were treated with either free cdGMP and MPLA systemically (intravenous) or locally in the tumor (intratumor). Flow cytometry analysis 48 hours later indicated that there were no significant changes of major immune cell subsets in treated groups compared with controls in the blood (Fig. 2A), tumor (Fig. 2B), or spleen (Fig. 2C). In certain cases, immune cell populations decreased in treated mice compared with controls (i.e., blood neutrophils, tumor T and B cells, and splenic macrophages, CD4⁺ T cells, and B cells). Notably, a reduction in tumor-infiltrating T cells is directly correlated to advanced disease (39) and may be reflected in these free agonist treatments as early as 48 hours. These results strongly suggested that free cdGMP and MPLA delivered systemically or directly into the tumor had no therapeutic efficacy and clearly motivated the need for engineering a NP-mediated delivery strategy.

Toward this goal, we prepared 60-nm liposomes containing equimolar quantities of DOPC and DPPC and 3 mol% mPEG2000-DSPE via ultrasonication. Transmission electron microscopy was performed on negative-stained samples for analysis of lipid ultrastructure (Fig. 3A, white arrows indicate multilamellar NPs). DLS measurements indicated that NPs were 61.2 ± 13.4 nm in diameter (Fig. 3B) and zeta potential measurements indicated a neutral surface charge (−5.4 ± 1.0 mV; Fig. 3C). Encapsulation efficiency of cdGMP was approximately 42% (Fig. 3D) and stability measurements indicated that hydrodynamic size remained the same over 24 hours (Fig. 3E). Release measurements of cdGMP at 37°C demonstrated that 14% of cdGMP was released after 3 hours and just 19% was released over 21 hours, suggesting that high cdGMP payloads could be stably delivered to tumor sites (Fig. 3F). To validate the function of these NPs, we treated RAW 264.7 macrophages in vitro with equivalent amounts of NPs containing cdGMP only, MPLA only, and both cdGMP and MPLA on the same particle (Fig. 3G). Cell culture supernatants were assayed for levels of secreted IFNα (Supplementary Fig. S1) and IFNβ (Fig. 3H) by ELISA after 24 hours. Both individual formulations mediated IFNβ production 19–25 times above background levels (Fig. 3H). Most significantly, however, immuno-NPs loaded with both cdGMP and MPLA induced a synergistic production of IFNβ that was 7–10 times the levels induced by NPs carrying a single agonist alone (Fig. 3H). We therefore selected this coloaded immuno-NP formulation for treatment studies in vivo.

We next quantified tumor-homing efficiency of immuno-NPs. Mice bearing orthotopic 4T1 primary tumors were injected with fluorescent NPs and live-animal Spectrum imaging was performed 24 hours later (Fig. 3I). Particle fluorescence appeared brightest per pixel basis in the region of the primary tumor and was predominantly colocalized with luminescence from luciferase-expressing tumor cells, suggesting that immuno-NPs drained with high efficiency to the primary tumor site (Fig. 3i). FMT imaging on excised tumors indicated that particle signal appeared to collect in distinct nodes (white and yellow arrows, Supplementary Fig. S2). To quantify tumor draining, 4T1 tumor–bearing mice were injected with fluorescent NPs and perfused 4 and 28 hours later. Quantification of NP fluorescence obtained from Spectrum imaging indicated that 4.7% of immuno-NPs accumulated in tumor masses within 4 hours and this accumulation increased to 6.0% 28 hours postinjection (Fig. 3j). These data suggested that most of the tumor drainage occurred at a short-time scale within hours after injection to the finite perivascular niche (Fig. 3J). Biodistribution analysis at 28 hours indicated that accumulation of immuno-NPs within the tumor compared with the liver, which is expected to have high retention due to its intrinsic clearance function, was similar per organ basis (Fig. 3K) and higher per gram tissue basis (Fig. 3L). Confocal microscopy analysis indicated that particles were found throughout the tissue from the tumor periphery to the center, often localizing in perivascular regions near the tumor vasculature as noted by CD31 staining for tumor endothelium (Supplementary Fig. S3A and S3B). Notably, NP signal often appeared punctate within individual cells, reminiscent of endosomal localization. Taken together, these data suggested that immuno-NPs are stable and drained with high efficiency to primary mammary tumors within 4 hours. Because safety considerations are paramount for systemically administered therapies, we injected healthy mice that did not bear tumors with immuno-NPs and assayed serum levels of ALT (Fig. 3M) and AST (Fig. 3N) as a metric of hepatotoxicity. While AST levels were significantly elevated compared with untreated controls 1 day posttreatment, they dropped back to baseline within just 4 days. While mice injected with immuno-NPs lost approximately 10% of their weight within 24 hours, they regained this weight within 7 days with an additional approximately 8% weight gain within 3 weeks compared with controls (Fig. 3O). We highlight, therefore, that any immuno-NP–mediated toxicity is minimal and transient, with recovery observed within just 4–7 days.

We next performed flow cytometry analysis for the types of immune cells found both locally within the tumor, as well as systemically in the blood and spleen 48 hours after treatment (Fig. 4). We used treatment with immune checkpoint inhibitors anti-programmed cell death protein-1 (anti-PD-1) and anti-cytotoxic T-lymphocyte–associated antigen-4 (anti-CTLA-4) as a metric of a clinically approved immunotherapy for comparison. Immuno-NP treatment very significantly increased DCs (11–13-fold), macrophages (5–10-fold), and NK cells (13–15-fold) in the blood compared with treatment with inhibitors and untreated controls (Fig. 4A). A representative dot plot for blood immune cells is shown in Fig. 4B. Similarly, immuno-NP treatment increased the number of DCs in the tumor 2.0-fold relative to untreated controls and also increased the numbers of tumor macrophages and NK cells 2.6–4.7-fold and 4.0–8.3-fold, respectively, compared with treatment with inhibitors and untreated controls (Fig. 4C). A representative dot plot for tumor immune cells is shown in Fig. 4D. Immuno-NPs also increased numbers of splenic macrophages 1.5–1.8-fold compared with mice treated with inhibitors and control mice (Fig. 4E). Figure 4F shows a representative dot plot for splenic immune cells. The upregulation of APCs and NK cells in all three blood/tissue compartments suggested that immuno-NPs could mediate both an innate and adaptive immune response, which is important for long-term therapy. Notably, we also compared immune cell recruitment and activation with immuno-NPs that were actively targeted to unique molecules on the tumor-associated vasculature such as α_νβ₃ integrins and P-selectin (8, 40), but these formulations did not have significant upregulation of immune cells compared with controls 48 hours post-treatment. We deduced from these early findings that targeted NPs are likely to be sequestered at target sites that may not be enriched in APCs, while untargeted NPs are free to drain to the perivascular areas for ready uptake by APCs.

To determine the efficacy of immuno-NPs in reducing primary tumor burden, we treated mice bearing advanced 4T1 mammary tumors twice intravenously with immuno-NPs, as noted (Fig. 5A and B, black arrows indicate treatment days). Within a day, tumor signal in mice treated with immuno-NPs decreased by 85%, while tumor signal in both control groups increased by 55%–57%. Tumors were excised posttreatment and mice treated with immuno-NPs had tumors that were significantly 50%–60% reduced in mass compared with either control group (Fig. 5C; Supplementary Fig. S4). Notably, there were no statistical differences in tumor masses between control groups, indicating that the vehicle alone had no effect in mediating treatment efficacy (Fig. 5C). We then treated mice bearing 4T1 orthotopic tumors with immuno-NPs either intravenously or intratumorally alone or supplemented with anti-PD-1 and anti-CTLA-4. We compared these treatment groups with control groups where mice were given either free cdGMP and MPLA intravenously or intratumorally, inhibitors only, or left untreated (Supplementary Fig. S5A, black arrows indicate treatment days; weight monitoring data in Supplementary Fig. S5B). A week after the start of treatment, mice given free immune agonists intravenously had to be euthanized because of large and ulcerating primary tumors. We noted that in the untreated control group, smaller tumor cell bioluminescence signals, especially during the latter stages of therapy (Supplementary Fig. S5A), did not correlate to smaller tumor sizes as measured by physical caliper measurements (Fig. 5D), indicating, as has been widely shown in the literature (15–17, 19, 22, 24), that the 4T1 tumor mass is composed of a highly heterogeneous mixture of cells where nontumor cells infiltrating the TME are often involved in advancing the cancer. From the start of treatment and up to 1 week afterwards, tumors of mice treated with immuno-NPs intravenously alone or with inhibitors significantly remained the smallest with volumes of approximately 170 mm³ (data for day 5 posttreatment in Fig. 5D).

To establish a mechanistic framework for immuno-NP uptake and function within the primary tumor, we next treated 4T1 tumor–bearing mice with two consecutive daily doses of fluorescent immuno-NPs and analyzed for cellular uptake by flow cytometry. Compared with mice injected with empty NPs, mice treated with immuno-NPs had significantly elevated numbers of CD45⁺ immune cells and NK cells per gram of tumor tissue (Fig. 5E). Furthermore, there was a significantly increased number of cells that had taken up NPs in these treated mice, while levels of uptake within tumor cells, CD45⁺ immune cells, DCs, macrophages, and NK cells were statistically similar (Fig. 5F). Notably, while approximately 53% of NPs were taken up by CD45⁺ immune cells (∼3% by DCs, ∼9% by macrophages, and ∼33% by NK cells), only 0.9% of NPs were taken up by tumor cells, indicating that perivascular accumulation was optimal for targeting of APCs and NK cells in lieu of tumor cells, because this niche is enriched for these immune cells (Fig. 5F). Further analysis also demonstrated that DCs in particular endocytosed significantly more immuno-NPs within the primary tumor (Fig. 5G), and that these NP⁺ DCs produced significantly more IFNβ (Fig. 5H). To investigate longer-term efficacy, blood and tumors were harvested from these mice 7 and 14 days posttreatment and assayed for immune cell content by flow cytometry. Compared with untreated controls, DCs were significantly elevated in the blood 7 days posttreatment (Fig. 5I), with no statistical differences in CD4⁺ T cells (Fig. 5LJ). Notably, CD8⁺ T cells were significantly elevated in the blood 14 days posttreatment (Fig. 5K). Intratumoral DCs (Fig. 5L) and NK cells (Fig. 5M) were similarly elevated per gram of tumor tissue 7 days posttreatment, with no statistical differences in numbers of intratumoral T cells (Fig. 5N and O).

To determine the efficacy of immuno-NPs in treating metastasis, we treated 4T1 tumor–bearing mice with immuno-NPs with and without checkpoint inhibitors intravenously or intratumorally and free agonists intravenously or intratumorally (Fig. 6A, black arrows indicate treatment days). From all the formulations, it was only immuno-NPs administered intravenously that completely prevented metastasis in the lungs and lymph nodes (Fig. 6A). Twelve days after the start of treatment, metastatic signal in mice treated with immuno-NPs intravenously remained near baseline and significantly lower than signal in mice treated with free agonists intratumorally, immuno-NPs intratumorally with inhibitors, and inhibitors only (Fig. 6A; representative bioluminescence images of metastasis shown in Fig. 6B). Notably 12 days posttreatment, while none of the mice treated with immuno-NPs intravenously had a metastatic signal above threshold (dashed grey line in Fig. 6A), 50%–100% of mice in all other groups had increased signal. As with the primary tumor, we then sought to establish a mechanistic framework for immuno-NP uptake and function within lungs that contained metastases. To this end, we treated 4T1 tumor–bearing mice with two consecutive daily doses of fluorescent immuno-NPs and analyzed for cellular uptake by flow cytometry. Compared with mice injected with empty NPs, mice treated with immuno-NPs had significantly elevated numbers of CD45⁺ immune cells, macrophages, and NK cells per gram of lung tissue (Fig. 6C). There were significantly more cells in treated mice that had taken up NPs (Fig 6D) and while approximately 53% of these NP⁺ cells were CD45⁺ immune cells (and significantly so), only 0.2% of cells that had taken up NPs in the lungs were tumor cells (Fig. 6D). Representative flow cytometry dot plots for NP⁺ cells are shown in Fig. 6E. Strikingly, in mice that were treated with immuno-NPs, lung DCs and macrophages had endocytosed significantly more NPs (Fig. 6F) and these NP⁺ DCs and macrophages produced significantly more IFNβ (Fig. 6G). Confocal microscopy studies indicated that immuno-NPs were found largely in the vicinity of tumor cells in the lungs and in close proximity to DCs (Fig. 6H), macrophages (Fig. 6I), and NK cells (Fig. 6J), indicating that this immune response was tumor associated and not nonspecific. In areas of the lungs that did not contain metastases, immuno-NPs were also largely absent (Supplementary Fig. S6).

Finally, to establish the broader impact of the efficacy of this immuno-NP treatment to other aggressive cancers, we investigated their therapeutic efficacy in the treatment of mice bearing orthotopic B16F10 melanoma tumors (Fig. 7). We treated tumor-bearing mice with two consecutive daily doses of immuno-NPs and observed that tumor volume dropped by 50% in treated mice compared with controls within a single day and continued to remain a significant 50%–65% reduced over the course of 2 weeks (Fig. 7A, black arrows indicated treatment days). Survival in treated mice was also significantly increased compared with controls (Fig. 7B; weight monitoring data in Supplementary Fig. S7). Blood flow cytometry analysis indicated significantly elevated levels of DCs (Fig. 7C). There were no statistical differences in levels of blood NK cells (Fig. 7D) and CD4⁺ T cells (Fig. 7E). Importantly, CD8⁺ T cells were significantly elevated in blood even 7 days posttreatment and continued to increase 14 days posttherapy (Fig. 7F). Taken together, these results indicate that the therapeutic efficacy of immuno-NPs can be translated across multiple models of aggressive cancer.

While immune agonists can generate powerful antitumor immunity, their successful delivery to the TME relies on efficiently accessing the tumor microvasculature, minimizing off-target toxicity, and, in the case of delivery of multiple synergistic therapeutics, targeting the same cell (6–10, 12, 14). Here, we elected to coencapsulate STING agonist cdGMP (27, 29–32) and TLR4 agonist MPLA (28, 34) within a single liposomal NP and exploited systemic delivery to access the leaky tumor microvasculature in entirety. Our delivery strategy mediated preferential collection within the perivascular areas of the TME that are rich in APCs. Coencapsulation within a NP not only prevents cdGMP/MPLA-induced off-target systemic toxicity, but also guarantees delivery to the same target APCs. The success of this large class of immunotherapies relies on these absolute fundamental advantages of NP-mediated systemic delivery. In the case of the primary tumor, systemic delivery enables immuno-NP accumulation within the APC-rich perivascular areas of the tumor, which may drive preferential accumulation of immuno-NPs within CD45⁺ immune cells compared with tumor cells. DCs, macrophages, and NK cells are highly efficient phagocytes compared with tumor cells that are also notoriously poor antigen presenters. In the case of metastasis, it can be argued that systemic delivery may be the singular route of efficient access. Given that both cdGMP and MPLA are strong inducers of type I IFNs and share common downstream effectors such as NF-κB and IRF3 (41, 42), the IFNβ synergy we measured as a result of coencapsulation of both agonists is likely due to amplification of these downstream pathways. IFNβ drives robust innate immunity, NK cell activity, and APC-mediated activation of CD8⁺ T cells (2, 24). It is also pivotal in preventing malignant transformation and dedifferentiating cancer stem cells (43–45).

Our mechanistic analysis has demonstrated that treatment efficacy depends on the accumulation of immuno-NPs at the site of the primary tumor or metastases, their uptake by local APCs, and the production of IFNβ by these cells. Notably, we have shown these APCs endocytose a significantly greater number of immuno-NPs. We also observed significantly increased levels of blood CD8⁺ T cells in mice treated with immuno-NPs, indicating a functioning vaccination mechanism, where activated tumor-resident DCs can travel to lymph nodes to prime systemic T cells and mediate their recruitment back to the site of the tumor. These data further demonstrate that NK cells are a significant contributor to efficacy, and their role is highly advantageous because they are not only direct-killing initial responders but can also be involved in the formation of memory NK cells.

In conclusion, we have constructed a versatile lipid-based immuno-NP platform that serves to mount a robust antitumor immune response by accumulating with high efficiency in the APC-rich perivascular regions of primary tumors and metastases and inducing the strong production of IFNβ. We have shown that synergy between cdGMP and MPLA coencapsulated in the same NP drives the production of high levels of IFNβ and mediates a robust APC- and NK cell–driven antitumor immune response. Finally, we have shown that immuno-particles delivered systemically via the vasculature outperform even intratumoral delivery and serve to reduce and limit the growth of the primary tumor and prevent metastasis without the need for additional supplementation with clinically approved checkpoint inhibitors. This nanotechnology platform serves to effectively reverse significant TME immunosuppression and this approach shows much promise for NP platforms that similarly aim to “reengineer” TME immunity.

C.J. Hoimes reports receiving a commercial research grant from Merck Sharpe and Dohme, has received speakers bureau honoraria from Genentech, Bristol Myers Squibb, and has an unpaid consultant/advisory board relationship with 2bPrecise, Seattle Genetics, Bristol Myers Squibb, Genentech, Array Biopharma, and AstraZeneca. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P.U. Atukorale, C.J. Hoimes, E. Karathanasis

Development of methodology: P.U. Atukorale, P.A. Bielecki, C.J. Hoimes, E. Karathanasis

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.U. Atukorale, S.P. Raghunathan, V. Raguveer, T.J. Moon, C. Zheng, M.L. Wiese, A.L. Goldberg, G. Covarrubias

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.U. Atukorale, P.A. Bielecki, C.J. Hoimes, E. Karathanasis

Writing, review, and/or revision of the manuscript: P.U. Atukorale, P.A. Bielecki, G. Covarrubias, C.J. Hoimes, E. Karathanasis

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.U. Atukorale, T.J. Moon

Study supervision: P.U. Atukorale, E. Karathanasis

This work was supported by grants from the NCI (R01CA177716 and U01CA198892), the Alex's Lemonade Stand Foundation, and the Angie Fowler AYA Cancer Research Initiative of the Case Comprehensive Cancer Center (to E. Karathanasis). P.A. Bielecki was supported by a graduate research fellowship from the National Science Foundation. G. Covarrubias was supported by a fellowship from the NIH Interdisciplinary Biomedical Imaging Training Program (T32EB007509) administered by the Department of Biomedical Engineering, Case Western Reserve University. We acknowledge the Case Center for Imaging Research, Case Comprehensive Cancer Center Flow Cytometry Core, and the Case School of Medicine Light Microscopy Core Facility and support from NIH S10-RR031845. We further acknowledge Mei Yin and the Cleveland Clinic Lerner Research Institute Imaging Core for assistance with negative staining and electron microscopy. We also thank the University Hospitals Pathology Core for assistance with ALT/AST measurements.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.