Spatially tuned localization of interleukins and OX40 agonists enhances synergistic anti-tumor immunity

Summary

While advances in immunotherapy have revolutionized care for cancer patients, most approaches still fail to mount a robust anti-cancer effect. To address this challenge, the field has turned to combination immunotherapies, but systemic administration of potent combination therapies has resulted in severe immune-related adverse effects. Here, we use an injectable polymer-nanoparticle (PNP) hydrogel to rationally localize immunotherapeutic cytokines (IL-12 and IL-2) intratumorally and antibodies (OX40a) peritumorally. We hypothesize that selectively altering the biodistribution of these cargo enables tolerable and synergistic anti-cancer combinations. Using in vivo imaging, we demonstrate that administration sites are critical for redistributing cargo to either the tumor or tumor-draining lymph node (tdLN). Further, we demonstrate that this approach synergistically improved treatment efficacy in the B16F10 and MC38 tumor models and alters cellular phenotypes, resulting in significant tumor clearance in both single and double tumor models.

Graphical abstract

Highlights

• •Hydrogel technology enables targeted localization of immunotherapy
• •Intratumorally and peritumorally administered combination therapies clear tumors
• •Rationally localized therapy results in abscopal clearance and long-term survivors

Klich et al. use a hydrogel to deliver immunotherapy combinations intratumorally and peritumorally. Rational selection of the site of administration drives tumor clearance and long-term survivors.

Introduction

While immune checkpoint blockade is now a frontline cancer immunotherapy, up to 75% of patients fail to respond.^1,2,3,4,5,6 This limited response is driven in part by the highly immunosuppressive tumor microenvironment (TME), underscoring the need for combination therapies that target both innate and adaptive anti-cancer immunity.⁷

Cytokines such as interleukin-12 (IL-12) and interleukin-2 (IL-2) are promising innate immune agonists that promote interferon-γ (IFNγ)-mediated T cell cytotoxicity and enhance natural killer (NK) cell activity, potentially converting immune-excluded tumors to an anti-cancer state.⁸ However, their pleiotropic effects and short therapeutic windows often necessitate high systemic doses, resulting in immune-related adverse effects and dose-limiting toxicities.^9,10

Intratumoral (IT) delivery has emerged as an approach to improve the therapeutic index of potent immune agonists.¹¹ However, peak plasma IL-12 levels occurred 5–7 h after IT administration in one clinical trial,¹² highlighting the need for local retention. Anchoring strategies, such as alum-tethered cytokines, have reduced systemic toxicity while improving efficacy,^13,14 while tumor-targeted cytokines and agonists such as CpG have improved safety and potency.^15,16,17

Nevertheless, many of these strategies require modification of the therapeutic cargo and repeated administration, limiting translatability. Materials-based approaches, including immunostimulatory liposomes and hydrogels, offer an alternative for improving the safety and efficacy of physicochemically distinct combination immunotherapies.^18,19,20,21

Moreover, targeting the lymph nodes, not just the tumor, represents an important strategy for generating durable anti-tumor immunity. For example, OX40, a T cell co-stimulatory receptor involved in memory T cell formation, has shown therapeutic promise as both a monotherapy and in combination with other immunotherapies.^22,23,24,25

Here, we investigated how the selective locoregional delivery of distinct immune agonists, enabled by an injectable hydrogel depot technology, could impart immunological changes to the tumor microenvironment or the tumor-draining lymph node (tdLN), facilitating both cytotoxicity and immunological memory. Based on prior results demonstrating improved safety, tolerability, and potency with the sustained delivery of antibodies and immunostimulatory cytokines,^20,26 we hypothesized that injectable polymer-nanoparticle (PNP) hydrogels could provide sustained site-specific control over the localization of cytokines (IL-12 and IL-2) and agonistic anti-OX40 antibodies to create a synergistic combination immunotherapy.

We evaluated these immune agonists, delivered either IT or peritumorally (PT), in the poorly immunogenic B16F10 melanoma model, and we validated the lead combination in the MC38 adenocarcinoma model. We examined how site of administration of these therapies skewed biodistribution with in vivo imaging of excised organs and explored how this altered biodistribution could be leveraged to enable synergistic combination immunotherapies. We characterized these responses with flow cytometry, demonstrating that site of administration is crucial to mounting distinct immunophenotypes in the tumors and lymph nodes in treated mice. We demonstrated that maximizing the therapeutic efficacy required pro-inflammatory cytokines IL-2 and IL-12 to be delivered IT in combination with PT administered OX40a, where efficacy was substantially diminished if the cargo were co-localized. Notably, this platform approach was demonstrated to be both cargo and tumor agnostic, improving efficacy in multiple tumor models and mounting abscopal effects without requiring modification of the cargo, suggesting a promising path for future exploration of combination immunotherapies.

Results

PNP hydrogel depots modulate cargo exposure depending on site of administration

We hypothesized that sustained local delivery of immune agonists such as OX40a from injectable hydrogels would increase drug exposure at the site of administration. We previously showed using positron emission tomography that PNP hydrogels altered CD40a pharmacokinetics, increasing retention at the injection site and accumulation in tdLNs.²⁶ We therefore hypothesized that PNP hydrogels could be used to selectively deliver cargo to tumors through IT administration or to tdLNs through PT administration. (Figure 1A). To this end, we labeled anti-OX40 agonist antibodies (OX40a) with Alexa Fluor 647 NHS-ester and prepared OX40a loaded PNP hydrogels by physically mixing aqueous solutions of dodecyl-modified hydroxypropyl methylcellulose (HPMC-C₁₂), poly(ethylene glycol)-poly(lactic acid) (PEG-PLA) nanoparticles, and labeled OX40a. These hydrogels form rapidly upon mixing through dynamic, non-covalent interactions between the HPMC-C₁₂ polymers and the PEG-b-PLA NPs, enabling facile encapsulation of diverse cargo and easy administration by injection (Figure 1A). PNP hydrogels exhibit robust solid-like properties across a range of physiologically relevant frequencies, shear-thinning behavior necessary for injectability through fine gauge needles used in clinical settings, rapid self-healing that minimizes burst release following injection, and excellent biocompatibility.^27,28,29

Figure 1.

Polymeric nanoparticle hydrogels increase site-specific cargo accumulation

(A) Schematic of experimental hypothesis indicating that PNP hydrogels can be used to selectively distribute immunotherapies as a function of site of administration, comparing intratumoral and peritumoral injection (top). Schematic of dodecyl-modified hydroxypropyl methylcellulose (HPMC-C₁₂) and poly(ethylene glycol)-poly(lactic acid) (PEG-PLA) nanoparticles as functional building blocks of the PNP hydrogel depot technology (bottom).

(B) Experimental timeline, groups, and doses for tumor induction, treatment, and tumor or lymph node excision and imaging by IVIS, with n = 6–10 per group per time point (data pooled from two independent experiments).

(C) Representative images of excised tumors on days 1, 3, and 5 from each treatment group.

(D) Quantification of average radiance in excised tumors for each time point.

(E) Representative images of excised tumor-draining lymph nodes on days 1, 3, and 5 from each treatment group.

(F) Quantification of average radiance in excised tumor-draining lymph nodes for each time point. Data are reported as mean ± SEM. Statistics are two-tailed unpaired Welch’s t test, run in GraphPad Prism.

To evaluate differences in biodistribution and bioaccumulation of cargo from distinct sites of administration, C57BL/6 mice were inoculated with either 5 × 10⁵ MC38 cells or 3 × 10⁵ B16F10 cells and treated 7 days thereafter either IT or PT with PNP-1-5 hydrogels (n.b., 1-5 refers to the hydrogel formulation comprising 1 wt% HPMC-C₁₂ and 5 wt% PEG-PLA NP, hereafter is referred to simply as PNP hydrogel) loaded with 10 μg of fluorolabeled OX40a and adjuvanted with 20 μg of CpG, per dose (Figure 1B). To assess retention in either the tumor or tdLN, tumors and tdLNs were excised at the time points indicated and imaged with an in vivo imaging system (IVIS). At each point across a five-day time series, we found that IT administration resulted in significantly greater retention and accumulation in the tumor compared to PT administration of the labeled antibody (Figures 1C and 1D). In reversal of this trend, PT administration resulted in a meaningful increase in antibody exposure to the tdLNs when compared to IT or systemic administration (Figures 1E and 1F); this result was most notable at the critically relevant time point of D3, at which point we would expect peak OX40 receptor upregulation post-activation³⁰ (Figure 1F, p = 0.0009).

We also sought to quantify the impact of locoregional, site-specific administration relative to current clinical approaches for protein therapeutics, namely intravenous infusion. In reporting the ratios IT:Systemic and PT:Systemic of average radiance, we demonstrate that IT delivery via PNP hydrogels increases tumor exposure by up to 21.8× relative to systemic administration and that PT delivery via PNP hydrogels increases lymph node exposure by up to 3.4× relative to systemic administration (Figures S1A and S1B). Interestingly, when quantifying signal in the tdLNs, PT administration resulted in over 2× greater exposure relative to systemic administration for all four time points evaluated whereas IT administration only crossed this threshold at 24 h, demonstrating that overall, PT administration enables more targeted locoregional delivery to tdLNs compared to IT or systemic routes of administration.

Collectively, these data demonstrate that the PNP hydrogel depot technology can be leveraged to selectively skew cargo accumulation and total exposure as a function of site of administration. Interestingly, we also noted by inspection (unquantified) differences in tumor burden between groups, with those mice receiving PT OX40a having the smallest tumors (Figure S2).

IT hydrogel administration of cytokines confers survival benefit over PT administration

Having demonstrated distinct pharmacokinetics and biodistribution from IT and PT hydrogel administration, as well as systemic administration in a standard liquid vehicle, we set out to examine whether these differences resulted in changes in therapeutic efficacy when delivering proinflammatory cytokines. First, we evaluated the effect and tolerability of a single administration of low dose IL-12 in C57B/6 mice induced with 3 × 10⁵ B16F10 cells and treated 4 days thereafter (Figure 2A). Treatment groups included PNP hydrogels loaded with 5 μg of IL-12 administered either IT or PT, empty PNP hydrogel delivered IT as an untreated control, or IT soluble IL-12 as a standard liquid vehicle control.

Figure 2.

Intratumoral cytokine immunotherapy extends survival over peritumoral administration

(A) Experimental overview of tumor induction, treatment, and measurement timeline, n = 10 for all groups.

(B) Tumor growth over 25 days post-tumor inoculation for mice treated with empty gel, intratumoral gel, peritumoral gel, or intratumoral soluble containing 5 μg of IL-12, left to right, respectively.

(C) Survival curve comparing therapeutic effect of intratumoral gel to intratumoral soluble administration of cytokine.

(D) Survival curve comparing therapeutic effect of intratumoral gel to peritumoral gel of cytokine.

(E) Experimental overview of tumor induction, treatment, and measurement timeline, n = 7 for vehicle control, n = 6 for untreated, n = 10 for all others. Cytokine treatment consisted of 5 μg IL-2 and 20 μg IL-12.

(F) Tumor growth curves through day 20 post-inoculation of intratumorally administered empty gel with ICB, intratumorally administered PNP hydrogel with cytokines in combination with ICB, untreated, and peritumorally administered PNP hydrogel with cytokines in combination with ICB, from left to right, respectively.

(G) Probability of survival through 50 days post-tumor inoculation. Survival curves were compared by log rank Mantel-Cox test.

We first observed that IT administration of IL-12 with PNP hydrogel led to better control over tumor growth through 25 days post-tumor induction compared to soluble and PT administration of the same therapy (Figure 2B). Indeed, IT PNP hydrogel administration resulted in significantly longer survival of mice compared to those treated with soluble IT administration of the same therapy, with median survival extending to 32 days from 29 days for the control treatment (p = 0.045, Figure 2C). Further, the group receiving IT PNP hydrogel administration had survivors through day 49 post-tumor induction compared to day 39 for mice receiving IT soluble cytokine. We also compared the survival benefit between mice treated with PT and IT hydrogel-based cytokine formulations. Here, there was a notable extension in the median survival by almost a week, from 26 days for mice treated with PT hydrogel-based formulations to 32 days for mice treated with IT hydrogel-based formulations (p = 0.064, Figure 2D). This trend was corroborated by quantifying tumor burden on D20, the last time point at which no animals from the IT and PT treatment groups were euthanized. Here, the average tumor burden for animals treated with low-dose IT cytokine monotherapy was 34.7 mm² compared to 63.6 mm² for animals receiving the same treatment PT (Figure S3, p = 0.052).

Having demonstrated that IT PNP hydrogel delivery of cytokines could elicit anti-cancer responses, we next tested whether higher doses could be tolerated before exploring combinations to enhance efficacy without increasing toxicity. Based on doses reported for other tumor-anchoring systems,^13,14,15 we administered 5 or 20 μg of IL-12 to B16F10 tumors on day 7 (n = 8/group) and monitored weight loss as an indicator of treatment-induced toxicity for two weeks (Figure S4). No mice lost >5% body weight, and any weight loss was recovered within one week.

Given the favorable safety profile of higher-dose IL-12-loaded hydrogels, we therefore chose to examine the therapeutic efficacy of IT-injected PNP hydrogels loaded with both 20 μg of IL-12 and 5 μg of IL-2, which are complementary cytokines that activate both the JAK and STAT pathways. To move toward more clinically relevant treatment regimen, we included immune checkpoint blockade (ICB) in all treatments hereafter. Given well-documented resistance to aPD1 monotherapy,^31,32 we pursued treatment regimen in combination with ICB to determine whether our engineering approach could improve upon standard of care. Here, all mice, except for those in the untreated group, received ICB as 200 μg doses of aPD1 delivered intraperitoneally every three days for five total administrations (Figure 2E). Mice receiving IT hydrogel-based cytokine formulations, in combination with ICB, had well controlled tumor burden through day 20 post-tumor inoculation compared to those receiving other treatments (Figure 2F). Mice receiving empty hydrogels and those that were untreated had tumors that started growing exponentially around 10 days after tumor inoculation, while 3 of 10 mice receiving the PT hydrogel-based cytokine treatment had tumors that started growing exponentially by day 20. The mice were monitored for long term survival, and only the mice receiving IT hydrogel-based cytokine formulations had long term survivors (defined as surviving ∼2× longer than any animal in the untreated control group, or 50 days), extending median survival from 32 to 37 days compared to those receiving PT administration of the same treatment (Figure 2G, p = 0.0018).

IT hydrogel administration of cytokines polarizes the tumor microenvironment

Our initial results showed that IT, but not PT, hydrogel-based cytokine delivery improved anti-cancer efficacy, supporting our hypothesis that administration site alters the tumor microenvironment. To investigate the mechanism underlying the survival benefit of IT PNP cytokines, we used flow cytometry to assess tumor T cell populations seven days after IT or PT treatment (representative gating Figure S5). Interestingly, while the overall frequencies of CD8⁺ and CD4⁺ T cells among CD45⁺ cells were not different between the mice receiving IT vs. PT injections, the compositions of T cell subsets were meaningfully skewed toward an anti-tumor phenotype in mice receiving IT PNP cytokines compared to those receiving PT PNP cytokines (Figure 3A). For instance, with respect to CD8⁺ T cell subsets, mice receiving IT cytokine therapy had a greater percentage of central memory, IFNγ-expressing, and naive and stem-like memory cells (Figures 3A–3G). Similar trends were observed among the CD4⁺ T cells (Figures 3H–3M). Notably, IT PNP cytokine treatment elicited a significant increase in the fraction of CD4⁺ central memory T cells when compared to those generated by PT PNP cytokine treatment, but IT PNP cytokine also led to a modest increase of CD25⁺FOXP3⁺ regulatory T cells relative to PT PNP cytokine treatment (Figures 3J, 3N, and 3O).

Figure 3.

Intratumoral administration of combination cytokine immunotherapy alters the T cell immunophenotype of the tumor

(A) Representative population of CD8⁺ T cells gated on CD62L and CD44 for subset phenotypes from murine B16F10 tumors; treatments included IT PNP cytokine (5 μg IL-2 + 20 μg IL-12) in blue and PT PNP cytokine (5 μg IL-2 + 20 μg IL-12) in pink.

(B–E) Quantification of percentage of (B) CD8⁺ T cells, (C) CD44⁺CD62L⁺ central memory T cells, (D) CD44⁺CD62L⁻ effector memory T cells, and (E) CD44⁻CD62L⁺ T cells.

(F) Representative populations of interferon-gamma producing CD8⁺ T cells gated on CD8a and IFNγ.

(G) Quantification of percentages of IFNγ CD8⁺ T cells.

(H) Representative population of CD4⁺ T cells gated on CD62L and CD44 for subset phenotypes from murine B16F10 tumors.

(I–M) Quantification of percentages of (I) CD4⁺ T cells, (J) CD44⁺CD62L⁺ central memory T cells, (K) CD44⁺CD62L⁻ effector memory T cells, (L) CD44⁻CD62L⁺ T cells, and (M) IFNγ⁺ CD4⁺ T cells.

(N) Representative populations of FOXP3⁺ regulatory T cells gated on CD25 and FoxP3.

(O) Quantification of percentages of regulatory T cells. n = 6 for all groups. Tumor excised 7 days after treatment. Data reported as mean ± SEM. Values reported as percentages of CD45⁺ cells (CD4⁺ and CD8⁺ T cells) or of parent T cell population. Statistics determined by two-tailed Mann Whitney test in GraphPad prism.

Assessing myeloid populations, we also observed that IT PNP cytokine treatment augmented the fraction of MHCII⁺CD11c⁺-activated antigen-presenting dendritic cells (DCs), and it particularly increased the percentage of inflammatory DCs (iDCs) when compared to PT PNP cytokines. In contrast, PT PNP cytokine treatments elicited higher fractions of cDC1s and cDC2s in the tumor (Figures S6A–S6D). Other populations, including B cells and NK cells, were similar across treatment groups in the tumors (Figures S6E–S6H). We concomitantly examined lymphoid and myeloid populations in the tdLN on day 7 post-treatment. Here, there was a modest increase in the fraction of cytotoxic CD8⁺ T cells, NK cells, and MHCII⁺CD11c⁺ DCs when treated IT compared to PT (Figures S7A–S7K). Of note, IT PNP cytokine administration reduced the fraction of cDC2s compared to PT administration (Figure S7L).

To evaluate therapeutic effects across multiple timescales, we conducted a similar analysis to assess the frequency of T cell subsets three days post-treatment. At this time point, we observed fewer differences in cell phenotypes between the two treatment groups, though IT PNP treatment drove an increase in the percent of cDC1s and cDC2s in the tumor while PT administration increased the representation of iDCs and NK cells (Figures S8A–S8H). We also quantified frequencies of lymphoid and myeloid cells in the tdLNs on day three (Figures S9A–S9L) and most notably found that IT administration of hydrogel-based cytokine formulations resulted in a slightly increased frequency of effector memory CD8⁺ T cells, a decrease in naive and stem-like memory CD8⁺ T cells, an increase in cDC1s, and a decrease in cDC2s (Figures S9C, S9D, S9I, S9K, and S9L).

Collectively, these results suggest that IT PNP cytokine treatments increased the antigen specificity and proliferative capacity of T cells in the tumor microenvironment by augmenting the percentage of antigen-exposed, self-renewing, and migratory central memory T cells among both CD8⁺ and CD4⁺ populations. These subsets have been shown to cf. anti-tumor immunity through the production of higher levels of cytokines and their longer persistence than effector memory subsets.^33,34,35 These changes in T cell subsets may be generated by shifts in the dendritic cell populations, where IT PNP cytokine treatments increased the percentage of cDC1s and cDC2s in the tdLNs at just three days post-treatment, potentially priming the expansion of central memory T cells observed by day 7. Finally, IT PNP cytokine treatments resulted in an increase in IFNγ-producing CD8⁺ T cells, which has direct anti-cancer consequences.

PT hydrogel administration of OX40a confers survival benefit over IT administration

Having examined how administration site affects cytokine-induced anti-cancer immunity, we hypothesized that similar effects would occur with agonistic anti-OX40 antibodies, which promote T cell activity and memory formation. Because these processes occur largely in the tdLNs, where tumor antigens, T cells, and antigen-presenting cells interact, we hypothesized that PT delivery of OX40a would be more effective than IT delivery.

To assess the therapeutic potential of OX40a, we induced melanoma as before and treated with PNP hydrogels loaded with 10 μg of OX40a and 20 μg of CpG per dose on day 7. Mice received the same aPD1 schedule as before. (Figure 4A). We adjuvanted OX40a with CpG because OX40 is expressed on activated T cells, and prior studies suggest that CpG co-delivery enhances OX40 efficacy.³⁶ We first tested whether PT delivery of CpG alone could improve treatment (Figure S10A). PT CpG + systemic aPD1 provided no therapeutic benefit, confirming that CpG alone did not account for the effects of adjuvanted OX40. We also compared locoregional OX40a delivery with systemic administration. Intraperitoneal OX40a + aPD1 provided no benefit over controls (Figure S10B), consistent with the current limitations translating IV-infused biologics to the clinic.

Figure 4.

Peritumoral administration of OX40a controls tumor growth and extends murine survival in B16F10

(A) Experimental scheme, IT OX40a n = 6, n = 8 all other groups. One animal from the PT OX40a group was censored on day 31 due to the development of an ulcer.

(B) Tumor growth curves through day 20 post-inoculation of intratumorally administered OX40a with ICB, peritumorally administered OX40a with ICB, ICB only, and untreated, left to right, respectively.

(C) Probability of survival through 50 days post-tumor inoculation. Survival curves were compared by log rank Mantel-Cox test.

Here, we observed that tumors remained well controlled in both groups receiving hydrogel-delivered therapies, while mice receiving only aPD1 or those that were untreated experienced rapid tumor growth, by two weeks post-administration (Figure 4B). Yet, only PT administration of the hydrogel-based OX40a treatment resulted in long-term survivors, extending median survival from 26 to 31 days when compared with IT administration (Figure 4C).

PT administration of OX40a alters tumor and lymph node immunophenotype

We now sought to characterize how IT and PT administration of OX40a-loaded PNP hydrogel formulations (denoted simply IT or PT mAb) altered the immunophenotype of the tumor microenvironment and tdLNs. First, we assessed the frequency of T cell subsets in the tumor seven days following IT or PT administration of OX40a-loaded hydrogels. The frequency of CD8⁺ T cells and CD8⁺ central memory T cells were similar across both treatment groups, but we found that CD8⁺ T cells were enriched for effector memory subsets when treated PT rather than IT (Figures 5A–5D). We also observed that PT OX40a administration noticeably decreased the frequency of naive and stem-like memory CD8⁺ T cells when compared to IT treatments (Figure 5E), whereas both treatment groups resulted in comparable frequencies of IFNγ⁺ cytotoxic CD8⁺ T cells (Figures 5F and 5G).

Figure 5.

Peritumoral administration of OX40a (mAb)-hydrogels augments anti-cancer T cell phenotypes

(A) Representative population of CD8⁺ T cells gated on CD62L and CD44 for subset phenotypes from murine B16F10 tumors. IT PNP mAb (10 μg OX40a +20 μg CpG) in blue and PT PNP mAb (10 μg OX40a +20 μg CpG) in pink.

(B–E) Quantification of percentage of (B) CD8⁺ T cells, (C) CD44⁺CD62L⁺ central memory T cells, (D) CD44⁺CD62L⁻ effector memory T cells, and (E) CD44⁻CD62L⁺ T cells.

(F and G) Representative populations of interferon-gamma producing CD8⁺ T cells gated on CD8a and IFNγ. Quantification of percentage of (G) IFNγ⁺ CD8⁺ T cells.

(H) Representative population of CD4⁺ T cells gated on CD62L and CD44 for subset phenotypes from murine B16F10 tumors.

(I–M) Quantification of percentages of (I) CD4⁺ T cells, (J) CD44⁺CD62L⁺ central memory T cells, (K) CD44⁺CD62L⁻ effector memory T cells, (L) CD44⁻CD62L⁺ T cells, and (M) IFNγ⁺ CD4⁺ T cells.

(N) Representative populations of FOXP3⁺ regulatory T cells gated on CD25 and FoxP3.

(O) Quantification of percentages of regulatory T cells. n = 6 for all groups. Data reported as mean ± SEM. Values reported as percentages of CD45⁺ cells (CD4⁺ and CD8⁺ T cells) or of parent T cell population. Statistics determined by two-tailed Mann-Whitney test in GraphPad prism.

Impressively, we found that PT administration of hydrogel-based OX40a formulations dramatically increased the proportion of CD4⁺ T cells and, among them, significantly boosted the fraction with an effector memory phenotype (with a corresponding reduction in the percent of central- or naive and stem-like memory phenotypes) and those expressing IFNγ (Figures 5H–5M). Critically, PT administration reduced the fraction of regulatory T cells when compared to IT administration of the same therapy (Figure 5K).

To complement these data, we assessed myeloid populations in the tumor on day 7 post-treatment and found modest increases in frequencies of MHCII⁺ CD11c⁺ DCs, cDC2s, cDC1s, B cells, NK cells, and CD4⁺ T cells in the tumors from mice treated with PT OX40a treatments while IT treatments led to an increase in cell frequency for only iDCs and CD8⁺ T cells (Figures S11A–S11H).

Concurrently, we examined T cell populations in the tdLNs on day 7 post-treatment. Here, we observed similar trends, where PT administration of OX40a augmented the frequency of central memory CD8⁺ T cells and effector memory CD8⁺ T cells while decreasing the corresponding frequency of naive and stem-like memory CD8⁺ T cells compared to IT administration (Figures S12A–S12D). We found a marginal increase in the frequency of helper CD4⁺ T cells but similar frequencies of CD4⁺ T cell subsets between treatment groups in the lymph nodes and similar percentages of NK cells and MHCII⁺CD11c⁺ DCs (Figures S12E–S12J). Further, PT administration of the OX40a therapy resulted in decreased fractions of cDC1s and cDC2s in the tdLNs compared to IT administration (Figures S12K and S12L).

Recognizing that a critical parameter of anti-cancer immunity is the generation of memory T cell subsets, we wanted to characterize the tumor microenvironment and tdLNs at a later stage post-treatment. We again performed flow cytometric analysis on tumors and tdLNs from mice treated with OX40a-loaded hydrogels 14 days after treatment (day 21 post-tumor induction). In the tumors, we noted that PT OX40a treatments increased the contribution of MHCII⁺ CD11c⁺ DCs and cDC2s (Figures S13A and S13B) while decreasing the proportion of iDCs and cDC1s when compared to IT administration (Figures S13C and S13D). We observed similar fractions of B cells and a small decrease in the proportion of NK cells from PT treatments compared to IT treatments (Figures S13E and S13F). Interestingly, we found a meaningful decrease in the percentage of CD8⁺ T cells and an increase in the frequency of CD4⁺ T cells with PT OX40a treatments (Figures S13G and S13H). Crucially, meaningful differences in frequency of lymphoid and myeloid cells persisted in the tdLNs through day 14, evidence of a sustained response (Figures S14A–S14L).

Consistent with cytokine studies, flow cytometry of tumors and tdLNs after OX40a therapy suggests that site of administration, not just sustained locoregional delivery, is crucial to mounting a robust anti-cancer immune response. Here, PT hydrogel-based OX40a treatments elicited superior anti-tumor cellular responses compared to IT treatments, which is the exact opposite trend from cytokine immunotherapy. These results support our hypothesis that PT OX40 treatments would increase cargo exposure in the T cell zones of lymphoid organs (i.e., tdLNs) where OX40⁺ cells localize after antigen activation. In contrast to cytokine immunotherapies (which increased central memory T cell subsets), these data suggest that the enhanced anti-tumor therapeutic efficacy of PT OX40a treatments is driven by an expansion of effector memory subsets of both CD4⁺ and CD8⁺ T cells. These effector subsets acquire increased immune functionality such as rapid antigen recall and the ability to produce both inflammatory cytokines and cell lytic molecules.³⁷ Indeed, these effector subsets have been shown to correlate with positive prognosis in several cancers.³⁸ Our data indicate that PT hydrogel-based OX40a treatment also resulted in dramatic expansion of CD4⁺ T cells among all T cells, decreased immunosuppressive regulatory T cells in the tumors, and prolonged T cell and APC responses, consistent with established mechanisms of OX40 agonism.^39,40,41

Combination of IT administered cytokines and PT delivered OX40a mounts robust anti-cancer response

Noting the individual efficacies of cytokines delivered IT and OX40a delivered PT, we next sought to explore possible synergy between these therapeutic approaches. We hypothesized that the combination of IT delivered cytokines and PT administered OX40a would capitalize on tumor polarization and mount a robust memory response to cf. superior anti-cancer efficacy.

To assess the potency of this combination, we first induced B16F10 melanoma tumors in C57B/6 mice and treated IT with PNP hydrogels loaded with cytokines as before (5 μg IL-2 and 20 μg IL-12) with and without concomitant PT administration of PNP hydrogels loaded with 10 μg OX40a and 20 μg CpG (Figures 6A–6C). All groups except for the untreated control group received 250 μg of aPD1 delivered every five days for three total administrations. Reporting tumor burden through day 20 post-tumor induction, we found that mice receiving hydrogel treatments had better controlled tumor growth than those that did not, consistent with our previous findings (Figure 6B). Interestingly, tumor growth was better controlled in the group receiving the combination therapy of IT cytokines + PT OX40a than in the group receiving IT cytokines alone (Figure 6B). This trend persisted through the duration of the study, where we found that 50% of mice treated with the IT cytokines + PT OX40a combination survived past day 50 post-tumor inoculation compared, whereas no mice survived past day 41 when treated with IT cytokines and ICB alone (Figure 6C). The combination therapy extended median survival almost two weeks to 46 days compared to 32.5 days for mice receiving only IT cytokine and ICB (Figure 6C).

Figure 6.

Therapeutic efficacy of intratumorally and peritumorally administered combination immunotherapy

(A) Experimental scheme. In the B16F10 study, n = 6 for IT cytokine + PT OX40a, n = 8 all other groups. In the MC38 study, n = 8 for all groups.

(B) Tumor growth curves for mice induced with B16F10 through day 20 post-inoculation of intratumorally administered IL-12 and IL-2 with ICB, intratumorally administered IL-12 and IL-2 and peritumorally administered OX40a with ICB, ICB only, and untreated clockwise from top left, respectively.

(C) Probability of survival for mice induced with B16F10 through 50 days post-tumor inoculation.

(D) Tumor growth curves for mice induced with MC38 through day 20 post-inoculation of intratumorally administered IL-12 and IL-2 with ICB, intratumorally administered IL-12 and IL-2 and peritumorally administered OX40a with ICB, ICB only, and untreated clockwise from top left, respectively.

(E) Probability of survival for mice induced with MC38 through 50 days post-tumor inoculation.

(F) Experimental scheme, n = 8 for all groups.

(G) Probability of survival for mice induced with B16F10 through 50 days post-tumor inoculation.

(H) Tumor growth curves for mice induced with B16F10 through day 20 post-inoculation of intratumorally delivered cytokine-antibody combination, peritumorally administered cytokine-antibody combination, ICB only, and untreated, left to right, respectively. Survival curves were compared by log rank Mantel-Cox test.

To verify that these effects were tumor agnostic, we repeated the study in C57B/6 mice inoculated with 5 × 10⁵ MC38 murine adenocarcinoma cells and treated with the same regimens (Figures 6A–6D, and 6E). Tumor burden remained well controlled over the duration of the study in both groups receiving hydrogel delivered immunotherapy, with 100% of animals being tumor free by day 50 post-tumor induction in those two cohorts (Figures 6D and 6E).

To validate that the observed synergy is a consequence of rational localization of the respective cargo, we wanted to see how efficacious the combination immunotherapy was when co-delivered at a single site. Here, we induced mice with B16F10 melanoma tumors and treated either IT or PT with a single hydrogel-based formulation comprising the entire immunotherapy combination, including IL-12, IL-2, OX40a, and CpG at the same doses as before (Figure 6F). Mice receiving the IT total combination therapy had a median overall survival of 36 days while those receiving the same therapy PT had a median overall survival of 30.5 days (Figure 6G). Interestingly, mice receiving the IT total combination therapy had well controlled tumor burden through day 20 post-tumor induction whereas 25% of those receiving the PT total combination therapy had tumors that reached exponential growth by day 23 (Figure 6H). Despite the difference in immediate tumor growth, which resulted in earlier mouse euthanasia, the long-term survival of both cohorts converged by day 40 with 25% long term survivors, which is 25% fewer than when the therapies were appropriately localized in target tissues (Figure 6G).

As before, we sought to characterize these responses at the cellular level. We induced B16F10 tumors, treated, and excised tumors and tdLNs for analysis by flow cytometry on day 7 post-treatment. We began by assessing T cell populations. First, we observed that the spatially distinct combination therapy comprising IT cytokines + PT OX40a resulted in the greatest frequency of CD8⁺ T cells compared to either of the monotherapies or the IT total combination. Interestingly, the IT total combination therapy resulted in a lower frequency of CD8⁺ T cells than any other treatment, indicating a deleterious effect of having all cargo delivered simultaneously into the tumor (Figures 7A and 7B). When examining CD8⁺ T cell subsets, we previously noted that IT cytokines drive an expansion of central memory cells, and this same trend was especially noticeable here when comparing to PT OX40a monotherapy. The central memory subsets are nevertheless recovered when the therapies are delivered together, either in a co-localized or spatially distinct fashion (Figure 7C). Similarly, the IT cytokine + PT OX40a combination therapy increases the frequency of CD8⁺ effector memory cells, which was consistent with our observations for the PT OX40a monotherapy relative to IT cytokines alone, an improvement that is not realized when the therapies are co-localized (Figure 7D). The expansion of these critical cell populations is reinforced by the dramatic reduction in naïve- and stem-like central memory CD8⁺ T cells in tumors from mice receiving IT cytokine + PT OX40a combination therapy, especially compared to the IT cytokine monotherapy (Figure 7E). Fascinatingly, we observed that the IT cytokine + PT OX40a combination therapy elicited the greatest frequency of IFNγ-producing CD8⁺ T cells compared to all other groups, salvaging the otherwise low frequency of this population observed for the PT OX40a monotherapy (Figures 7F and 7G). These quantifications were repeated for CD4⁺ T cell subsets, where we saw that PT OX40a drove an expansion of CD4⁺ T cells relative to other treatment groups (which is mostly preserved in the combination therapy regimes and still a noticeable increase from IT cytokine therapy alone; Figures 7H and 7I). While the combination therapies decreased CD4⁺ central memory populations, this behavior was complimented by a subtle increase in CD4⁺ effector memory fractions (Figures 7J and 7K). Once again, these differences paralleled decreases in the naïve- and stem-like central memory CD4⁺ T cells (Figure 7L). Most strikingly, however, was the synergistic effect the IT cytokine + PT OX40a combination immunotherapy had on IFNγ⁺ CD4⁺ T cells and regulatory T cells (Figures 7M–7P). The IT cytokine + PT OX40a combination therapy redeemed the IFNγ⁺ CD4⁺ T cell population relative to the PT OX40a monotherapy (and still meaningfully drove the population up from the IT cytokine therapy) while significantly decreasing the frequency of Tregs, a property not directly observed for either of the monotherapies alone. Importantly, this feature is specific to the IT cytokines + PT OX40a combination therapy and not the co-localized IT combination therapy.

Figure 7.

Flow cytometry of tumors on day 7 after treatment with combination immunotherapies

(A) Representative population of CD8⁺ T cells gated on CD62L and CD44 for subset phenotypes from murine B16F10 tumors. IT PNP cytokines (5 μg IL-2 + 20 μg IL-12) in light blue, PT PNP mAb (10 μg OX40a +20 μg CpG) in pink, IT PNP cytokines + PT PNP mAb in dark blue, and IT PNP co-localized combination in purple.

(B–E) Quantification of percentage of (B) CD8⁺ T cells of CD3⁺ T cells, (C) CD44⁺CD62L⁺ central memory T cells of CD8⁺ T cells, (D) CD44⁺CD62L⁻ effector memory T cells of CD8⁺ T cells, and (E) CD44⁻CD62L⁺ naive and stem-like T cells of CD8⁺ T cells.

(F) Representative population of CD8⁺ T cells gated on CD8a and IFNγ for quantification of IFNγ expressing subsets from murine B16F10 tumors.

(G) Quantification of IFNγ expressing T cells among CD8⁺ T cells.

(H) Representative population of CD4⁺ T cells gated on CD62L and CD44 for subset phenotypes from murine B16F10 tumors.

(I–N) Quantification of percentages of (I) CD4⁺ T cells of CD3⁺ T cells, (J) CD44⁺CD62L⁺ central memory T cells of CD4⁺ T cells, (K) CD44⁺CD62L⁻ effector memory T cells of CD4⁺ T cells, (L) CD44⁻CD62L⁺ naive and stem-like T cells of CD4⁺ T cells, (M) IFNγ expressing T cells among CD4⁺ T cells, and (N) CD25⁺ FoxP3⁺ regulatory T cells.

(O) Representative population of CD4⁺ T cells gated on CD4 and IFNγ for quantification of IFNγ expressing subsets from murine B16F10 tumors.

(P) Representative population of CD4⁺ regulatory T cells gated on CD25 and FoxP3. n = 6 for all groups. Values reported as percentages of CD45⁺ cells (CD4⁺ and CD8⁺ T cells) or of parent T cell population. Data reported as mean ± SEM. Statistics determined by Kruskal-Wallis test with Dunn’s multiple comparisons test in GraphPad prism.

We continued our characterization of the tumor by examining the frequencies of various myeloid populations (Figures S15A–S15F). Notably, the IT cytokine + PT OX40a combination therapy resulted in a moderate decrease in the proportion of MHCII⁺ CD11c⁺ DCs relative to the monotherapies, a moderate decrease in the percentage of cDC2cs relative to PT OX40a monotherapy, an increase in iDCs compared to PT OX40a monotherapy, and a decrease in cDC1s compared to both monotherapies (Figures S15A–S15D).

Next, we characterized T cell populations and their subsets in the tdLNs on day 7 post-treatment, as before. Here, we found that the IT cytokine + PT OX40a combination therapy significantly increased the percentage of CD8⁺ T cells compared to PT OX40a monotherapy while the co-localized IT combination therapy did not (Figures S16A and S16B). The IT cytokine + PT OX40a combination therapy resulted in a decrease in the frequency of central memory CD8⁺ T cells but dramatically increased the frequency of effector memory CD8⁺ T cells, a population which was not particularly augmented by any therapy alone (Figures S16A, S16C, and S16D). Both combination groups resulted in a consistent decrease in the naive- and stem-like memory CD8⁺ T cells, though the IT cytokine + PT OX40a combination therapy did so more significantly when compared to IT cytokines alone (Figure S16E). Further, in the case of the CD4⁺ T cell population and its subsets (Figures S16F–S16J), we noted the most impressive skew in the effector memory CD4⁺ population in the IT cytokine + PT OX40a combination therapy, which elicited significantly increased frequency compared to the monotherapies.

IT cytokines and PT antibody combination elicits abscopal effects

We next wanted to determine whether our lead combination therapy could drive systemic anti-tumor immunity. To model aggressive metastases, we co-induced two tumors by injecting 300K B16F10 cells into both the left and right flank of mice, treating only the right tumor on D7 post-inoculation (Figure S17A). Both treated and untreated tumors in animals receiving only aPD1 and in untreated animals grew rapidly in the first two weeks post-inoculation while those receiving cytokine or cytokine + antibody combination were well controlled (Figure S17B). Survival was remarkably similar to that seen for the single-tumor efficacy study (Figure S17C). We found that only the combination of IT cytokines and PT antibodies resulted in long-term, progression free survivors (33%), and that the site-specific combination immunotherapy resulted in significantly better overall survival compared to IT cytokines (p = 0.0207). Collectively, these results demonstrate the capacity for site-specific, locoregional delivery of immune agonists to impart robust, systemic anti-cancer immunity by clearing abscopal tumors.

Discussion

Cancer immunotherapies, particularly immune checkpoint inhibitors, have transformed cancer treatment and improved outcomes for many patients. However, many patients still fail to respond or suffer from dose-limiting immune-related adverse events that ultimately restrict efficacy.^42,43

In seminal work by Moynihan et al., it was demonstrated that successful combination approaches must engage both innate and adaptive immunity to eradicate fully established tumors and mount a robust memory anti-cancer response.³² Here, we demonstrated the capacity of a PNP hydrogel depot technology to selectively deliver distinct immunomodulating cargo to the tumor and tdLNs to promote innate and adaptive anti-cancer immunity; we showed that site-specific therapeutic administration could be rationally employed to increase cytotoxic IFNγ-producing T cells, a pool of T cells that have acquired effector function, and a reservoir of highly proliferative T cells to maintain the longevity of the immune response, all while simultaneously decreasing the frequency of immunosuppressive regulatory T cells, in the tumor microenvironment and tdLN.

Many reports have characterized the benefits of IT anchoring of cytokines to improve efficacy while reducing toxicity.¹⁵ Our studies corroborate these previous reports, where IT administration, but not PT administration, of combination cytokine immunotherapy polarized T cell phenotypes and increased IFNγ⁺ CD8⁺ T cell populations in the tumor microenvironment, which underlies cytotoxic tumor clearance⁴⁴ (Figure 3G). This effector response may be driven IL-2 upregulating the IL-12 receptor and preparing effector cells for enhanced Th1 signaling, increasing production of anti-tumor cytokines like IFNγ and the proliferation of effector cells. Interestingly, IT administration of these cytokines resulted in significant expansion of central memory subsets of both CD4⁺ and CD8⁺ T cells compared to PT administration of the same therapy (Figures 3C and 3J). Central memory T cells have enhanced survival and proliferative potential, providing a reservoir of antigen-exposed T cells ready to expand in response to challenges⁴⁵; the augmented central memory T cell contributions in animals receiving IT cytokine treatments we reported therefore may cf. an advantage in protecting against metastasis or relapse.

Importantly, we reported opposite trends with OX40a treatments. OX40 agonism has been shown to critically promote T effector cell activation and inhibit T regulatory cell function, encouraging especially the expansion and survival of CD4⁺ T cells, properties which have been shown crucial for eradicating local, distal, and spontaneous malignancies.²⁴ In this work, we show PT administration of OX40a most notably augmented T effector memory cell subsets (both CD8⁺ and CD4⁺), CD4⁺ T cells (as a fraction of all CD45⁺ cells), and downregulated immunosuppressive regulatory T cells compared to IT administration of the same therapy (Figures 5D–5I, 5K, and 5O). While CD8⁺ T cells have canonically been the focus for cancer immunotherapy due to their cytotoxic effector functions, recent reports suggest that CD4⁺ T cells play a pivotal role in directing the anti-tumor immune response and that local enrichment of their frequency correlates with survival in high-risk patients.^46,47,48

The process of T cell differentiation and drivers of T cell commitment to specific cell subsets is complex and tightly regulated in both space and time, and as suggested by the results of adoptive immunotherapy studies, a heterogeneous population of central memory and effector memory T cell populations may potentiate the best clinical outcomes.^49,50 Here, we demonstrated that site-specific therapeutic administration could be rationally employed to influence that composition. For example, IT cytokines alone augmented the central memory T cell response and CD8⁺ T cells while PT OX40a expanded the effector memory and CD4⁺ T cell responses. Independently, neither therapeutic approach mounted a sufficiently efficacious response to result in a meaningful number of long-term survivors. But, the combination therapy of IT cytokines + PT OX40a elicited robust central and effector memory T cell responses while significantly reducing immunosuppressive regulatory T cell responses, resulting in synergistic efficacy of 50% long term survivors in the highly immunosuppressive B16F10 melanoma model and 100% tumor regression in the MC38 adenocarcinoma model (Figures 6C and 6E). Importantly, these results could not be reproduced by simply combining all these agents in a single vehicle, as seen by the reduction in survival and phenotypic impact when the therapies were all co-localized (Figures 6G and 7; Figure S16). Impressively, the site-specific therapeutic approach resulted in 33% progression free survivors in a two-tumor B16F10 model designed to investigate abscopal effects (Figure S17C).

These results underscore the need to target therapy in a rational, spatiotemporal fashion. We hypothesize that cytokine driven expansion of central memory T cells is complementary to OX40 agonized lymph node interactions that drive a more robust effector memory T cell population. The combination therapy resulted in an entirely unique tumor immunophenotype consisting of both expanded central memory and effector memory T cell subsets, increased IFNγ-producing CD4⁺ and CD8⁺ T cells, and nearly complete regression of immunosuppressive regulatory T cells, results that realized superior efficacy compared to either of the leading monotherapies alone.

Overall, we have demonstrated a strategy for site-specific immunotherapy administration that bridges innate and adaptive immune pathways toward a wholly endogenous, antigen-agnostic anti-cancer response. This work demonstrated that our injectable hydrogel depot technology is an emerging immunoengineering system that can potentiate synergistic combination immunotherapy and address unique immunological questions regarding the precise spatiotemporal delivery of potent immune agonists.

Limitations of the study

The reported study uses murine cancer models such as the B16F10 model of melanoma and the MC38 model of colon adenocarcinoma induced in C57BL/6 mice. Future work would evaluate the immunotherapy delivery approach in other cancer lines and/or animal models to enable more generalizable claims toward clinical translation. Further, the study evaluated cellular phenotypes in tdLNs and tumor microenvironments to suggest mechanisms of efficacy, but future work would examine knockout models to investigate the immunological mechanism of action, evaluating whether immune checkpoint inhibitors are necessary to complement this approach and which cell types most significantly contribute to therapeutic benefit. Finally, this study reports the use of cytokines (IL-12 and IL-2) and antibodies (OX40a); additional therapeutic combinations could be explored.

Resource availability

Lead contact

Requests for further information, resources, and reagents should be directed to and will be fulfilled by the lead contact, Prof. Eric A. Appel (eappel@stanford.edu).

Materials availability

Materials generated in this study are available upon reasonable request from the lead contact with a completed materials transfer agreement through Stanford University’s Industrial Contracts Office.

Data and code availability

• •Data reported in this paper will be shared by the lead contact upon reasonable request for research purposes.
• •This paper does not report code.
• •Any additional information required to reanalyze the data reported in this paper is available upon reasonable request from the lead contact.

Acknowledgments

This work was supported financially in part by the Goldman Sachs Foundation (administered by the Stanford Cancer Institute, SPO# 162509). J.H.K. is thankful for the generous support of an Agilent Fellowship. A.N. is grateful for funding from the Paul and Mildred Berg Fellowship. E.L.M. was supported by the NIH Biotechnology Training Program (T32-GM008412). B.S.O. was supported by the Eastman Kodak Fellowship. J.B. is thankful for a Marie-Curie fellowship from the European Union (H2020; no. 101030481). Flow cytometry was performed on an instrument in the Stanford Shared FACS Facility obtained using the NIH S10 Shared Instrument Grant (1S10OD026831-01).

Author contributions

Conceptualization, J.H.K. and E.A.A.; methodology, J.H.K., A.N., E.L.M., B.S.O., J.B., and E.A.A.; investigation, J.H.K., A.N., E.L.M., B.S.O., Z.D.K., and J.B.; visualization, J.H.K., A.N., and E.A.A.; supervision, J.H.K. and E.A.A.; funding acquisition, E.A.A.; project administration, J.H.K. and E.A.A.; writing – original draft, J.H.K.; writing – review & editing, J.H.K., A.N., E.L.M., B.S.O., J.B., and E.A.A.

Declaration of interests

E.A.A., J.H.K., E.L.M., and A.N. are listed as inventors on a pending patent application describing the technology reported in this manuscript. E.A.A. is a co-founder, equity holder, and advisor to Appel Sauce Studios LLC, which holds a global exclusive license to this technology from Stanford University.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
OX40a	BioX Cell	Cat# BE0031; RRID: AB_1107592
aPD1	BioX Cell	Clone RMP1-14, Cat# BE0146; RRID: AB_10949053
anti-mouse CD16/CD32	BD	Cat# 553142; RRID: AB_394657
anti-mouse CD44	BioLegend	Cat# 103079; RRID: AB_2941373
anti-mouse/rat XCR1	BioLegend	Cat# 148213; RRID: AB_2564368
anti-mouse CD11c	BioLegend	Cat# 117308; RRID: AB_313777
anti-mouse CD19	BioLegend	Cat# 115520; RRID: AB_313655
anti-mouse CD161	BioLegend	Cat# 108740; RRID: AB_2562274
anti-mouse I-A/I-E	BioLegend	Cat# 107606; RRID: AB_313321
anti-mouse CD45	BioLegend	Cat# 103128; RRID: AB_493715
anti-mouse CD8a	BioLegend	Cat# 100750; RRID: AB_2562610
anti-mouse CD3	Invitrogen	Cat# 46-0032-82; RRID: AB_1834427
anti-mouse CD11b	BD	Cat# 563553; RRID: AB_2738276
anti-mouse F4/80	BioLegend	Cat# 123132; RRID: AB_11203717
anti-mouse CD4	BD	Cat# 612900; RRID: AB_2827960
anti-mouse CD62L	BD	Cat# 612833; RRID: AB_2870155
anti-mouse CD25	BioLegend	Cat# 101908; RRID: AB_961212
anti-mouse CD206	BioLegend	Cat# 141723; RRID: AB_2562445
anti-mouse Ly-6C	BioLegend	Cat# 128030; RRID: AB_2562617
anti-mouse Ly-6G	BioLegend	Cat# 127643; RRID: AB_2565971
anti-mouse IFN-γ	BioLegend	Cat# 505810; RRID: AB_315404
anti-mouse FoxP3	BioLegend	Cat# 126403; RRID: AB_1089118
Chemicals, peptides, and recombinant proteins
Hydroxypropyl methyl cellulose	Sigma-Aldrich	Cat# 423203
N,N-diisopropylethylamine (Hunig’s base)	Sigma-Aldrich	Cat# D125806
N-methyl-2-pyrrolidone (NMP)	Sigma-Aldrich	Cat# 328634
dichloromethane (DCM)	Sigma-Aldrich	Cat# 270997
Hexanes	Sigma-Aldrich	Cat# 208752
diethyl ether	Sigma-Aldrich	Cat# 296082
3,6-Dimethyl-1,4-dioxane-2,5-dione (lactide, LA)	Sigma-Aldrich	Cat# 303143
1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU)	Sigma-Aldrich	Cat# 139009
1-dodecyl isocyanate	Sigma-Aldrich	Cat# 389064
Poly(ethylene glycol) methyl ether (5 kDa)	Sigma-Aldrich	Cat# 81323
murine IL-12	Sino Biological	CT022-M08H
murine IL-2	Sino Biological	51061-MNAE
Alexa Fluor 647 NHS-ester	ThermoFisher	A20006
acridine orange/propidium iodide cell viability stain	Vitascientific	LGBD10012
collagenase/hyaluronidase	STEMCELL Technologies	# 07912
DNAse 1	STEMCELL Technologies	#100–0762
Experimental models: Cell lines
murine B16F10 melanoma	ATCC	CRL-6475
murine MC38 colon adenocarcinoma	Sigma Aldrich	SCC172
Experimental models: Organisms/strains
Mouse: C57BL\6	Charles River	N/A
Oligonucleotides
CpG ODN	Invivogen	tlrl-2395
Software and algorithms
Prism	Graphpad	https://www.graphpad.com
Adobe Illustrator	Adobe	https://www.adobe.com/products/illustrator.html
Adobe Illustrator	Adobe	https://www.adobe.com/products/illustrator.html

Experimental model and study participant details

All animal studies were performed in accordance with the National Institutes of Health (NIH) guidelines, with the approval of the Stanford Administrative Panel on Laboratory Animal Care (protocol APLAC-32947). Seven to eight week old female C57BL/6 were purchased from Charles River and housed in the animal facility at Stanford University. Mice were allowed to acclimate in the Stanford facilities for 1 week prior to beginning experimental procedures. The effect of sex on the results of the study were not evaluated.

Murine B16F10 melanoma was purchased from ATCC (CRL-6475). Aliquots of MC38 murine colon adenocarcinoma were provided as a gift from the Davis Lab at Stanford and subsequently purchased from Sigma Aldrich (SCC172). Cell lines were routinely tested for mycoplasma. Cells were maintained in RPMI media (Cytiva, SH30027.FS), supplemented with 10% fetal bovine serum (R&D Systems, S11550) and 1% penicillin-streptomycin (Themo Fisher, 15140122) prior to tumor inoculation.

Method details

Materials

Hypromellose (HPMC, meets USP testing specifications), N-methyl-2-pyrrolidone (NMP), 1-dodecylisocynate, N,N-diisopropylethylamine (DIPEA), acetone, monomethoxy-PEG (5 kDa), diazobicylcoundecene (DBU), acetic acid, diethyl ether, hexanes, dimethyl sulfoxide (DMSO), acetonitrile were purchased from Sigma-Aldrich and used as received. Dichloromethane (DCM) was purchased from Sigma-Aldrich and further dried via cryodistillation. Lactide was purchased from Sigma-Aldrich and recrystallized from ethyl acetate (dried over sodium sulfate) three times.

HPMC-C₁₂ synthesis

HPMC-C₁₂ was prepared according to previously reported procedures.^51,52 HPMC (1.0 g) was dissolved in NMP (40 mL) by stirring at 80°C for 1 h. Once the polymer had completely dissolved and the solution cooled to room temperature a solution of 1-dodecylisocynate (0.5 mmol) in NMP (5 mL) was added dropwise, followed by DIPEA (catalyst, 125 μL). The reaction was maintained at 50°C for 30 min, then heat was shut off and mixture was left stirring overnight at room temp. The solution was then precipitated from acetone and HPMC-C12 was purified by dialysis against MilliQ water for 3–4 days (MWCO 3.5 kDa) and lyophilized, yielding HPMC-C12 as a white amorphous powder. The polymer was dissolved at 60 mg mL-1 in sterile PBS, pH 7.4, prior to use in hydrogels.

PEG-PLA synthesis

PEG-PLA was prepared and analyzed as previously reported.^51,52 Recrystallized lactide (10 g) was fully dissolved in cryo-distilled DCM (45 mL) under N2 (g) with mild heating. Methoxy poly(ethylene glycol) (5 kDa; 2.5 g) was heated to 100°C under vacuum for 1–2 h, allowed to cool under N2, and then dissolved in cryodistilled DCM (5 mL). Once dissolved, the full PEG solution was added to the lactide solution under N2 and mixed with hand swirling. A solution of DBU (150 μL cryodistilled DBU per 1 mL cryodistilled DCM) was prepared and 500 μL added to the lactide-PEG solution under N2. The reaction was swirled by hand and allowed to react for 8 min before quenching with acetic acid (∼2 drops in 500 μL acetone). The PEG-PLA copolymer was precipitated from excess 50:50 mixture ethyl ether and hexanes, collected, and dried under vacuum to yield a white amorphous powder.

PEG-PLA nanoparticle (NP) preparation

NPs were prepared and analyzed as previously reported. Briefly, a solution (1 mL) of PEG-PLA in 25:75 DMSO:acetonitrile (50 mg mL−1) was added dropwise to water (10 mL) at a stir rate of 600 rpm. NPs were purified by ultracentrifugation over a filter (MWCO 10 kDa; Millipore Amicon Ultra-15) followed by resuspension in PBS to a final concentration of 200 mg mL−1. NP size and dispersity were characterized by DLS (Wyatt DynaPro PlateReader-II; average diameter = 34.1 nm, PDI ≤ 0.05).

PNP hydrogel formulation

To prepare PNP hydrogels, HPMC-C₁₂ was dissolved at 6 wt % in PBS and loaded into a 1 mL luer-lock syringe. A 20 wt % solution of PEG-PLA NPs in PBS was added to a solution of PBS with adjuvant, cytokine, and/or antibodies, depending on formulation, and loaded into a second 1 mL syringe. The two syringes were connected with a female-female luer lock elbow, with care to avoid air at the interface of the HPMC-C₁₂ and nanoparticle solution, and gently mixed until a homogenous PNP hydrogel was formed. Hydrogels were formulated with final concentrations of 1 wt % HPMC-C₁₂ and 5 wt % NPs. Each dose consisted of 20 μL hydrogel injection either intratumorally or peritumorally through a 27-gauge needle, as indicated. Intratumoral soluble/bolus controls were administered as a 20 μL injection in PBS through a 27-guage needle, while systemic treatments of anti-PD1 were delivered as 100 μL soluble intraperitoneal injections.

For our studies, 20 μL intratumoral injections were chosen intentionally to minimize tumor disruption. Tumor interstitial volumes have been characterized extensively, with averages of 30–55% of tumor volume, depending on the model.^53,54 Tumors were treated when they reached volumes between 60–100 mm³. Applying conservative constraints, a 20 mm³ injection volume is thus the acceptable volume while maintaining intratumoral injections.

Immunotherapy formulations

Immunotherapy injections were formulated with varying concentrations of cytokines, antibodies and adjuvants in either soluble form (in phosphate buffered saline, PBS) or in Polymer Nanoparticle hydrogels (PNP hydrogels). IL-12 was dosed at 5 or 20 μg per injection, as indicated (Sino Biological CT022-M08H). IL-2 was dosed at 5 μg per injection (Sino Biological, 51061-MNAE). Anti-OX40 antibody was dosed at 10 μg per injection (BioX Cell, BE0031). ODN 2395 (CpG) was dosed at 20 μg per injection (Invivogen, tlrl-2395). Anti-PD1 antibody was dosed at 200 or 250 μg per injection, as indicated (BioX Cell, Clone RMP1-14, BE0146).

Mammalian cell culture

Murine B16F10 melanoma was purchased from ATCC (CRL-6475). Aliquots of MC38 murine colon adenocarcinoma were provided as a gift from the Davis Lab at Stanford and subsequently purchased from Sigma Aldrich (SCC172). Cells were maintained in RPMI media (Cytiva, SH30027.FS), supplemented with 10% fetal bovine serum (R&D Systems, S11550) and 1% penicillin-streptomycin (Themo Fisher, 15140122) prior to tumor inoculation.

Murine tumor induction and monitoring

All animal studies were performed in accordance with the National Institutes of Health (NIH) guidelines, with the approval of the Stanford Administrative Panel on Laboratory Animal Care (protocol APLAC-32947). Seven to eight week old female C57BL/6 were purchased from Charles River and housed in the animal facility at Stanford University. Mice were allowed to acclimate in the Stanford facilities for 1 week prior to beginning experimental procedures. Flanks were shaven prior to injections.

MC38 flank xenografts were generated on 9-week-old female C57BL/6 mice by injecting a 50 μL of MC38 cells (10 × 10⁶ cells/mL) encapsulated in a 1 wt % alginate hydrogel subcutaneously above the right hind leg. Cell loaded alginate hydrogels were prepared using a dual syringe mixing technique as described previously. A stock solution of sterile alginate (Pronova UP LVG) was dissolved at 5wt % by adding saline to the polymers and allowing them to dissolve over 1 day at 4C. A stock solution of calcium sulfate (100 mM) was prepared in water in a large container in the form of a slurry, mixed vigorously. Calcium sulfate and cells were combined at the volume needed to reach the desired final concentration and loaded into a 1 mL syringe, while alginate (at the volume needed to reach the desired final concentration) was loaded into a separate 1 mL syringe. The two syringes were connected with a female-female luer lock elbow with care to avoid air at the polymer-solution interface, and the solutions were mixed gently until a homogenous alginate hydrogel (1 wt % alginate, 10 mM calcium sulfate) was formed.

B16F10 melanoma models were generated similarly, whereby 9-week-old female C57BL/6 mice were injected with a 50 μL alginate hydrogel loaded with B16F10 cells (6 × 10⁶ cells/mL) subcutaneously above the right hind leg (and, for the two-tumor abscopal study, above the left hind leg).

Following tumor inoculation, mice were monitored for palpable tumors. Treatments began on day 7 after tumor inoculation, unless otherwise noted, at which point tumors grew to approximately 5 mm in diameter. Treatment groups were formed to maintain consistent tumor burden at the time of treatment, resulting in groups of N = 6–12, depending on the study design. Mice were cage blocked or randomized depending on the study. Tumor burden was measured on D-1 before treatment to ensure fair distribution of tumors across treatment groups. Mouse weight was tracked to assess acute toxicity, with euthanasia criteria for sustained losses of more than 20% of body mass relative to the start of treatment. Tumor burden was tracked three times a week using caliper (Mitutoyo) measurements to calculate total tumor area (L × W); mice were euthanized once total tumor burden grew to 150 mm² or larger.

Alexa Fluor 647 conjugation of anti-OX40 antibody

Anti-OX40 antibody was conjugated with Alexa Fluor 647 NHS-ester (ThermoFisher, A20006) according to the manufacturer’s specifications. Briefly, Alexa Fluor 647 NHS-ester was prepared at 5 mg/mL in anhydrous DMSO and combined with the antibody in 8 molar excess. The reaction was left to stir and incubate for 1 h at room temperature followed by 24 h at 4°C before purification of the conjugated antibody via repeated washes through Amicon ultra centrifugal filters, 10 kDA MWCO (Millipore, UFC501008).

In vivo biodistribution of anti-OX40 antibody

C57BL/6 mice were injected either peritumorally or intratumorally with 1:5 PNP hydrogels loaded with 10 μg of AF647 labeled anti-OX40 antibody and 20 μg of CpG. At the indicated time points, mice were euthanized with CO₂ and lymph node and tumors were excised and imaged using an in vivo imaging system (IVIS Lago). Excised tissues of interest were imaged at 640/670 excitation/emission with medium binning, fstop = 8 for 2 s exposure (tumors) or 640/710 excitation/emission with medium binning, fstop = 4 for 5 s exposure (lymph nodes). Average radiance is reported.

Sample preparation for flow cytometry of lymph nodes and tumors

Tumor draining inguinal lymph nodes were harvested from mice at days 3, 7, and 14 post intratumoral or peritumoral immunotherapy administration, as indicated. LNs were mechanically disrupted to single cell suspensions using Corning single-frosted micro slides (2948–75 × 25), passed through 70 micron filters (Celltreat, 229484), spun at 500 rcf for 5 min, resuspended in PBS, and counted using acridine orange/propidium iodide cell viability stain (Vitascientific, LGBD10012) and a Luna-FL dual Fluorescence cell counter (Logos biosystems). 1 million live cells per sample were transferred to a 96-well conical bottom plate (Thermo Scientific, 249570) and stained.

Tumors were similarly excised, weighed and mechanically disrupted via surgical scissors. Bulk homogenate was enzymatically digested with an enzyme mixture consisting of collagenase/hyaluronidase (STEMCELL Technologies, # 07912) and DNAse 1 (STEMCELL Technologies, #100–0762) at 0.1×:0.15× ratios, respectively, in RPMI media (Cytiva, SH30027.FS). Samples and enzymatic mixture were transferred to 5.0 mL protein lobind tubes (Eppendorf, #0030108302) and left to digest at 37C while shaking at 300 rpm for 25 min. Digestion was stopped by adding complete RPMI media, previously described, supplemented with 1% 0.5 M EDTA (BioVision, 2103–100). Samples were spun at 300 rcf for 10 min with brake on low, resuspended in PBS, filtered, counted and plated as described for LNs.

Staining protocol and flow cytometry

LN and tumor samples were first stained with 200 μL Ghost Dye Violet 510 live/dead (Tonbo Biosciences, 13–0870-T100) for 5 min on ice, quenched with 100 μL FACS buffer (PBS, 3% heat inactivated FBS, 1 mM EDTA), and spun at 935xg for 2 min (with brake on low for tumor samples). Samples were then incubated with 50 μL anti-mouse CD16/CD32 (1:50 dilution, BD, 553142) for 5 min on ice before staining with 50 μL surface antibody stain for 30 min on ice. Samples were spun as before. For intracellular cytokine staining, samples were incubated in 200 μL 1× IC Fixation Buffer for 30 min at room temperature in the dark. Cells were spun at 500xg for 5 min, washed in 200 μL of 1× Permeabilization Buffer, and spun as before. Samples were then incubated in 100 μL of full intracellular antibody stain diluted in 1× Permeabilization buffer for 30 min at room temperature in the dark. Cells were quenched in 200 μL of 1× Permeabilization buffer, centrifuged at 500xg for 5 min, then washed and spun once more. Samples were resuspended in 200 μL FACS buffer and run on the Agilent NovoCyte Penteon Flow Cytometer in the Stanford Shares FACS Facility. Data was analyzed in FlowJo. FoxP3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific, 00–5523–00) were used for fixation and permeabilization according to the manufacturer’s instructions.

LN full antibody stain included anti-mouse CD44 (1:80 dilution; BV750; BioLegend, 103079), anti-mouse/rat XCR1 (1:160 dilution; AF647; BioLegend, 148213), anti-mouse CD11c (1:200 dilution; PE; BioLegend, 117308), anti-mouse CD19 (1:200 dilution; PE-Cy7; BioLegend, 115520), anti-mouse CD161 (NK1.1) (1:100 dilution; BV605; BioLegend, 108740), anti-mouse I-A/I-E (MHCII) (1:400 dilution; FITC; BioLegend, 107606), anti-mouse CD45 (1:400 dilution; AF700; BioLegend, 103128), anti-mouse CD8a (1:200 dilution; BV785; BioLegend, 100750), anti-mouse CD3 (1:200 dilution; PerCP/eFluor710; Invitrogen, 46003283), anti-mouse CD11b (1:100 dilution; BUV395; BD, 563553), anti-mouse F4/80 (1:125 dilution; BV421; BioLegend, 123132), anti-mouse CD4 (1:200 dilution; BUV805; BD, 612900), and anti-mouse CD62L (1:80 dilution; BUV737; BD, 612833).

Tumor surface antibody stain included anti-mouse CD25 (1:50 dilution; FITC; BioLegend, 101908), anti-mouse/rat XCR1 (1:100 dilution; AF647; BioLegend, 148213), anti-mouse CD206 (1:50 dilution; BV650; BioLegend, 141723), anti-mouse Ly-6C (1:100 dilution; BV570; BioLegend, 128030), anti-mouse Ly-6G (1:100 dilution; BV711; BioLegend, 127643), anti-mouse CD11c (1:100 dilution; PE; BioLegend, 117308), anti-mouse CD19 (1:100 dilution; PE-Cy7; BioLegend, 115520), anti-mouse CD161 (NK1.1) (1:50 dilution; BV605; BioLegend, 108740), anti-mouse I-A/I-E (MHCII) (1:200 dilution; FITC; BioLegend, 107606), anti-mouse CD45 (1:200 dilution; AF700; BioLegend, 103128), anti-mouse CD8a (1:100 dilution; BV785; BioLegend, 100750), anti-mouse CD3 (1:100 dilution; PerCP/eFluor710; Invitrogen, 46003283), anti-mouse CD11b (1:100 dilution; BUV395; BD, 563553), anti-mouse F4/80 (1:80 dilution; BV421; BioLegend, 123132), and anti-mouse CD4 (1:100 dilution; BUV805; BD, 612900). Tumor intracellular antibody stain included anti-mouse IFN-γ (1:20 dilution; APC; BioLegend, 505810) and anti-mouse FoxP3 (1:20 dilution; PE; BioLegend, 126403).

Quantification and statistical analysis

For imaging analysis, n = 6–10 samples were imaged per timepoint per group (pooled across two independent experiments), and data are presented as mean ± standard error of mean (SEM) as specified in the corresponding Fig. captions. Statistics are two-tailed Welch’s t test. Ratios of means were taken and SEM was propagated.

For survival studies and flow cytometry, a sample size of n = 6–12 was used and data are presented as mean ± standard error of mean (SEM) as specified in the corresponding Fig. captions. Comparisons between two treatment groups (i.e., intratumoral to peritumoral) were determined by two-tailed Mann Whitney test in GraphPad prism. Survival curves were compared by log rank Mantel-Cox test in GraphPad prism. Comparisons between multiple groups were determined by Kruskal-Wallis test with Dunn’s multiple comparisons test in GraphPad prism.

Throughout this text, p values less than 0.2 are shown.

Published: October 9, 2026