STING agonist delivery by tumour-penetrating PEG-lipid nanodiscs primes robust anticancer immunity

Immunotherapies like checkpoint blockade and CAR T-cell therapy have transformed cancer care, but many patients do not respond. Potent innate immune stimulators, notably STING agonists such as cyclic dinucleotides (CDNs), can induce in situ vaccination by activating dendritic cells (DCs) to prime antitumour T-cell responses. However, CDNs are hydrophilic, membrane-impermeable, and rapidly degraded, limiting systemic administration. Liposomal and polymeric nanocarriers can systemically deliver CDNs but historically expose only a small fraction (~2–10%) of tumour or infiltrating immune cells, likely due to poor penetration in dense tumour extracellular matrix and reticuloendothelial clearance. The authors hypothesized that a deformable, sub-100 nm, high-aspect-ratio nanocarrier—PEG-lipid nanodiscs (LNDs)—would penetrate tumours more effectively than traditional liposomes, increasing co-localization of STING agonists with dying tumour cells to optimally activate DCs and enhance T cell-mediated tumour immunity.

Prior work established the importance of STING in antitumour immunity and the therapeutic potential of STING agonists (e.g., Woo 2014; Corrales 2015; Sivick 2018). Small-molecule non-nucleotide STING agonists are in development (Ramanjulu 2018; Pan 2020; Chin 2020). Nanocarriers such as liposomes and polymersomes have improved CDN pharmacokinetics and delivery but typically achieve limited cellular exposure within tumours (~2–10%) due to transport barriers (Koshy 2017; Shae 2019; Wehbe 2020; Cheng 2018; Luo 2017). Nanoparticle design parameters (size <100 nm, shape, deformability) influence tumour penetration (Albanese 2012; Cabral 2011; Chauhan 2011; Tang 2014; Ding 2019; Niora 2020). PEGylated lipid nanodiscs (lipodisks) formed from high-Tm phospholipids and PEG-lipids have been characterized (Johnsson & Edwards 2003; Zetterberg 2016) and explored for drug delivery (Ahlgren 2017; Lin 2017; Zhang 2018), but their in vivo tumour-penetrating capacity had not been thoroughly analysed.

Design and formulation: A CDN prodrug (dialanine peptide linker) was synthesized and conjugated to a thiol-terminated PEG-phospholipid via thiol–maleimide coupling to generate a CDN-PEG-lipid designed for endosomal peptidase cleavage and self-immolative release of the active CDN. LNDs were formed by self-assembly of high-Tm phospholipids with PEGylated lipids including 5 mol% CDN-PEG-lipid via ethanol precipitation. Benchmark PEGylated liposomes were formulated with similar clinical compositions; due to stability constraints, CDN-PEG-lipid was incorporated at 1 mol% in liposomes (5 mol% caused aggregation). Particles were characterized by TEM and DLS. Fluorescent labels (sulfo-Cy5 or IR800cw) were incorporated for tracking. In vitro assays: Uptake of fluorescent LND-CDN and liposome-CDN by RAW-ISG macrophages and MC38 tumour cells was measured by flow cytometry and confocal microscopy, including co-localization with membrane and lysosomal markers. Release of free CDN after cellular uptake was quantified by LC–MS/MS. STING activation potency was assessed in RAW-Lucia ISG and human THP-1 reporter cells. Transport modeling and in vitro penetration: Coarse-grained molecular dynamics (Martini 2 force field, Gromacs) simulated translocation of 40 nm LNDs vs PEGylated or bare liposomes through a 20 nm rigid pore under a constant pulling force, analyzing deformation and translocation. Passive diffusion across track-etched membranes (50 and 200 nm pores) was tested in vitro. Penetration into MC38 tumour spheroids was evaluated by 3D confocal microscopy after incubation with fluorescent nanocarriers. Stability: Bioactivity retention of LND-CDN vs free CDN-PEG-lipid in 10% serum was measured using RAW-ISG reporter assays over time. In vivo pharmacokinetics and biodistribution: C57BL/6 mice bearing MC38 flank tumours received intravenous cGAMP-Cy5, LND-CDN, or liposome-CDN (equal CDN and dye molar doses). Plasma fluorescence was serially measured to derive two-phase decay half-lives. Tissue biodistribution at 24 h was quantified as %ID/g. Whole-animal cryofluorescence tomography (CFT) at 4 h and tumour histology at 24 h (with vascular dextran) assessed spatial distribution and extravascular penetration. Therapeutic efficacy and safety: Single-dose intravenous therapy was tested in MC38 (day 7 or 10 post-inoculation), 4T1 orthotopic breast cancer, and TC-1 HPV-driven tumour models. Comparators included PBS, parent CDNs at equal or 20× higher dose, ADU-S100 at 20× dose, and liposome-CDN at equal dose. Tumour growth, survival, rechallenge immunity, body weight, serum enzymes (ALT/AST/BUN), cytokine kinetics, and liver/spleen histology were assessed. Repeat dosing (weekly ×3) evaluated anti-PEG IgG by ELISA. Mechanistic studies: Efficacy in STING−/− mice and cytokine blocking (anti-IFNAR1, anti-TNFα, anti-IFNγ; alone or combined) tested pathway dependence. Flow cytometry of tumours treated with cargo-free fluorescent LNDs or liposomes quantified uptake by endothelial cells, myeloid subsets, DCs, and CD45− tumour cells. Depletions (anti-CD8, anti-NK1.1) and Batf3−/− mice assessed roles of adaptive and cross-presenting DCs. Using MC38-ZsGreen tumours, tumour-draining lymph node (TDLN) DCs were analysed for co-uptake of nanoparticle (Cy5) and tumour antigen (ZsGreen), activation (CD86), and kinetics (AUC over 3 days). IFN-γ ELISPOT measured tumour-specific T-cell responses. Intratumoral administration compared LND-CDN vs liposome-CDN when transport barriers were minimized.

- Particle properties and uptake: LNDs formed discoid nanoparticles with mean diameters ~26 nm (TEM) and ~33 nm (DLS), with 5–6 nm thickness consistent with a single bilayer. The smallest stable liposome-CDN formulation achievable was ~60 nm. Both carriers were taken up by RAW-ISG and MC38 cells; LND-CDN was a few-fold more potent than liposome-CDN and free CDN in activating RAW-ISG reporter cells, with all reaching maximal activation at micromolar concentrations. - Transport and penetration: Simulations showed LNDs deformed and entered a 20 nm pore under ~330 pN force, whereas liposomes could not. Analytical modeling suggested near-thermal forces could enable LND translocation when disc size slightly exceeds constriction size. Experimentally, LND-CDNs passively diffused across 50 and 200 nm pores while liposome-CDNs did not. In MC38 spheroids, LNDs penetrated cores significantly better than liposomes; LNDs associated with membranes and endosomes, while liposomes localized predominantly to endolysosomes. LND-CDNs retained ~75% bioactivity after 48 h in 10% serum, whereas free CDN-PEG-lipid decayed to near baseline by 24 h. - Pharmacokinetics and biodistribution: Plasma terminal half-lives were ~1 h for cGAMP-Cy5, 12.6 h for LND-CDN, and 7.6 h for liposome-CDN. At 24 h, tumour uptake was 7.4% ID/g for LND-CDN vs 1.1% ID/g for liposome-CDN; free cGAMP was ≤0.5% ID/g or undetectable in tissues. LNDs accumulated in tumour-draining and contralateral lymph nodes with low uptake in most other tissues; liposomes concentrated in spleen, LNDs more in liver. CFT at 4 h showed broad tumour distribution for LNDs versus patchy, low tumour accumulation for liposomes. Histology confirmed substantially greater extravascular tumour area coverage and fluorescence intensity for LNDs. - Therapeutic efficacy: A single 5 nmol IV dose of LND-CDN in MC38 induced tumour regression and complete responses in ~75% of mice; parent CDNs at 5 or 100 nmol and ADU-S100 at 100 nmol were ineffective. Liposome-CDN (5 nmol) transiently delayed growth but most tumours rebounded. LND-CDN remained effective when delayed to day 10 (~140 mm³ tumours), curing a majority; 8/9 cured mice rejected tumour rechallenge at day 90, indicating immune memory. In 4T1 orthotopic tumours, LND-CDN delayed growth and significantly increased median survival; 20× higher dose parent CDN had no effect. In TC-1 tumours, LND-CDN achieved ~43% cures. - Safety: Transient weight loss and transient elevations in serum ALT/AST and inflammatory cytokines normalized within 24–48 h; liver and spleen histology at 48 h appeared normal. Lower single or split doses (2.5 nmol) reduced efficacy while maintaining similar transient toxicity. Three weekly doses did not elicit detectable anti-PEG IgG. - Mechanism: LND-CDN efficacy required host STING; blocking IFNAR1 or TNFα reduced efficacy, and blocking IFNAR1+TNFα+IFNγ abrogated therapy. Early (4 h) intratumoural IL-6 and TNFα were elevated after both LND and liposome treatment. Without CDN cargo, both carriers reached tumour endothelium and myeloid cells; LNDs accumulated ~2× more in CD11c+ DCs and more in CD45− tumour (cancer) cells, indicating superior tumour cell exposure. CD8+ T-cell depletion abolished efficacy, NK depletion did not; Batf3−/− mice failed therapy, implicating cross-presenting DCs. In TDLNs of MC38-ZsGreen tumours, LND treatment yielded significantly more DCs co-positive for nanoparticle and tumour antigen over the first 3 days (~3× higher AUC than liposomes) and more activated Ag+CD86+ DCs. ELISPOT showed stronger tumour-specific T-cell responses with LND-CDN than liposome-CDN. When administered intratumorally to bypass transport, both LND- and liposome-CDNs were highly effective (5/5 vs 4/5 cures), indicating superior systemic efficacy of LNDs arises from improved tumour penetration and antigen–agonist co-localization.

Formulating STING agonist CDNs in deformable, sub-50 nm PEG-lipid nanodiscs overcomes key transport barriers that limit traditional liposomes. Simulations and diffusion experiments reveal that LNDs can readily deform to traverse narrow constrictions, enabling broad penetration throughout tumour parenchyma in vivo. This widespread distribution exposes a majority of tumour cells to the agonist, promoting optimal co-localization of dying tumour antigens with STING agonist within dendritic cells. Consequently, LNDs more effectively activate and license tumour antigen-bearing DCs in draining lymph nodes, driving robust CD8 T-cell priming and durable tumour control and memory. While both LNDs and liposomes reach tumour endothelium and trigger acute cytokine responses and necrosis, only LNDs achieve deep extravascular penetration and widespread cancer-cell exposure necessary for sustained adaptive immunity. The prodrug linker strategy ensures endosomal release in both tumour and antigen-presenting cells, consistent with enhanced efficacy across multiple tumour models. These data, together with low and transient systemic toxicity and lack of anti-PEG responses upon limited repeat dosing, suggest LNDs are a promising platform for systemic STING agonist delivery and potentially other immunostimulatory agents. The findings also imply that other flexible, high–aspect ratio nanomaterials (e.g., apolipoprotein-based nanodiscs) may provide similar advantages for intratumoural transport and immune activation.

A single intravenous dose of STING agonist delivered via tumour-penetrating PEG-lipid nanodiscs induces potent and durable antitumour immunity with immune memory, outperforming free CDNs and state-of-the-art PEGylated liposomes. LNDs achieve higher circulation persistence, markedly increased tumour accumulation, and deep extravascular penetration, enabling co-localization of STING agonist with tumour antigens in dendritic cells and robust CD8 T-cell priming. This work establishes flexible nanodiscs as an effective systemic delivery platform for STING agonists and highlights the importance of nanoparticle deformability and tumour penetration for immunotherapy. Future studies should explore repeated dosing regimens and long-term immunogenicity, alternative non-PEG coronas to mitigate potential anti-PEG responses, optimization across diverse tumour microenvironments, and translation to non-human primates and humans.

- The liposome comparator required a lower CDN-PEG-lipid mol% (1%) than LNDs (5%) due to aggregation, potentially confounding per-particle CDN payload and potency comparisons. - Maximum tolerated single-dose for LND-CDN was 5 nmol under these conditions; dose–response and multi-dose regimens were not extensively optimized. - Although no anti-PEG IgG was detected after three weekly doses, broader assessment of anti-PEG responses and hypersensitivity over extended or intensive dosing schedules remains necessary for clinical translation. - Transport and biodistribution assessments relied on fluorescent labels and surrogate cGAMP-Cy5, which may not identically reflect native drug behaviour; quantitative drug release in vivo was not directly measured in tumours. - Efficacy was demonstrated in murine syngeneic models; generalizability to human tumours and safety in larger species require further validation. - Mechanistic conclusions are based on cytokine blocking, genetic models, and correlative DC analyses; direct visualization of agonist–antigen co-localization dynamics in vivo remains inferential.