Injectable Nanoparticle-Based Hydrogels Enable the Safe and Effective Deployment of Immunostimulatory CD40 Agonist Antibodies

Abstract

When properly deployed, the immune system can eliminate deadly pathogens, eradicate metastatic cancers, and provide long-lasting protection from diverse diseases. Unfortunately, realizing these remarkable capabilities is inherently risky as disruption to immune homeostasis can elicit dangerous complications or autoimmune disorders. While current research is continuously expanding the arsenal of potent immunotherapeutics, there is a technological gap when it comes to controlling when, where, and how long these drugs act on the body. Here, this study explored the ability of a slow-releasing injectable hydrogel depot to reduce dose-limiting toxicities of immunostimulatory CD40 agonist (CD40a) while maintaining its potent anticancer efficacy. A previously described polymer-nanoparticle (PNP) hydrogel system is leveraged that exhibits shear-thinning and yield-stress properties that are hypothesized to improve locoregional delivery of CD40a immunotherapy. Using positron emission tomography, it is demonstrated that prolonged hydrogel-based delivery redistributes CD40a exposure to the tumor and the tumor draining lymph node (TdLN), thereby reducing weight loss, hepatotoxicity, and cytokine storm associated with standard treatment. Moreover, CD40a-loaded hydrogels mediate improved local cytokine induction in the TdLN and improve treatment efficacy in the B16F10 melanoma model. PNP hydrogels, therefore, represent a facile, drug-agnostic method to ameliorate immune-related adverse effects and explore locoregional delivery of immunostimulatory drugs.

Keywords: biomaterials; cancer; drug delivery; immunotherapy; pharmacokinetics.

Scheme 1

Hydrogels can localize the effects of CD40 agonists to reap their anti‐tumor benefits while abating their immune‐related adverse effects. A) CD40 agonist antibodies engage CD40 receptors on antigen presenting cells to potently upregulate antigen presentation, co‐stimulatory receptor expression, and secretion of immunostimulatory cytokines. Altogether, these effects potentiate T cell priming and create a more supportive environment for effector T cell function. B) Injectable, supramolecular polymer‐nanoparticle (PNP) hydrogels composed of dodecyl‐modified hydroxypropyl methylcellulose and poly(ethylene glycol)‐b‐poly(lactic acid) nanoparticles can be used to encapsulate CD40 agonists for local drug delivery. C) Peritumoral hydrogel administration leads to slow release of CD40 agonists into the local microenvironment, focusing the immunostimulatory effects on the tumor and draining lymphatics. In contrast, traditional systemic approaches lead to widespread exposure throughout the body and the occurrence of immune‐related adverse effects.

Figure 1

PNP hydrogels exhibit solid‐like rheological properties with robust shear‐thinning and self‐healing capabilities. A) Oscillatory shear rheology of PNP hydrogels indicating a storage modulus (G′) greater than the loss modulus (G″) throughout the tested frequencies. B) Steady shear rheology measurements indicating a decrease of viscosity at increasing shear. C) Ashby style plot demonstrating the injectability of PNP hydrogels with or without cargo under clinically relevant constraints. Based on methodology from Lopez Hernandez et al.^{[ 18 ]} D) Injection of PNP hydrogel through a 21‐gauge needle: (i) pre‐injection, (ii) injection of the hydrogel through the needle demonstrating liquid‐like rheological properties, (iii) when no longer under high shear conditions PNP hydrogels regain solid‐like properties, such as (iv) the resistance of flow due to gravity.

Figure 2

Sustained locoregional hydrogel‐based delivery of CD40a slows tumor growth and improves safety in a metastatic model of B16F10 melanoma. A) C57B/6 mice were inoculated with two flank B16F10 tumors and treated with an injectable hydrogel containing either anti‐PD1 or CD40a, or were treated with weekly systemic administration of anti‐PD1. Empty hydrogels (vehicle) were used as a negative control. Local delivery of a high dose (100 µg) of CD40a B) inhibited tumor growth and C) extended overall survival of mice. D) Mice treated with CD40a hydrogels exhibited reduced tumor growth in both treated and untreated tumors 10 days after treatment. E) Comparison of treatment‐induced weight loss due to high dose local CD40a administered as either a local bolus or as a PNP hydrogel. Longitudinal percent change in body mass (left panel) and the corresponding area under the curve (AUC) analysis (right panel). Efficacy (n = 10) and toxicity (n = 4 for CD40a treated mice, n = 2 for untreated) datasets represent two independent experiments. Data represented by means and SEM. Differences in tumor growth and survival were assessed using general linear models with SAS statistical software. Comparison of body weights and tumor growth inhibition was performed using one‐way ANOVA. False discovery rate (FDR) was controlled using the Benjamini and Hochberg method. Untreated controls indicate that mice were not administered any compound, hydrogel, or solutions. Vehicle‐treated mice were treated with the hydrogel containing no therapeutic compound.

Figure 3

PNP hydrogels retain CD40a at the injection site and reduce exposure to distant tissues. A) B16F10 bearing mice were treated with 100 µg of Zr⁸⁹‐labeled CD40a administered either as a PNP hydrogel or local bolus. Mice were imaged daily using positron emission tomography (PET) for 12 days. Representative images of CD40a biodistribution 24 h after treatment for B) local bolus or C) PNP hydrogel administration routes. Representative images of CD40a biodistribution 72 h after treatment for D) local bolus or E) PNP hydrogel administration routes. Scale bar is shared between (B) and (C), and between (D) and (E).

Figure 4

PNP hydrogels increase drug exposure to target tissues and reduce exposure in off‐target tissues compared to local bolus. A) Percent change in area under the curve (AUC) values for each tissue relative to local bolus administration. AUCs were derived from the PET pharmacokinetic data. Statistical comparisons indicate a significant difference from local bolus AUC, performed as multiple unpaired t‐tests. Corrections for multiple comparisons were performed using the FDR approach (Q = 1%). B) CD40a retention half‐life at the injection site, estimated from a one‐phase decay fit of PET data. Comparison performed using the extra sum‐of‐squares F test. PET pharmacokinetic curves tracking CD40a concentration in C) off‐target tissues, D) ipsilateral lymph nodes (including the tumor‐draining lymph node), and E) off‐target contralateral lymph nodes. N = 4 for both groups, data shown as mean and SEM.

Figure 5

PNP hydrogels mitigate gross toxicity caused by CD40a therapy. CD40a dose was titrated and treatment‐induced weight loss was observed. Area under the curve analysis of body mass curves during the acute weight loss phase (5 days post treatment) for mice bearing A) MC38 and B) B16F10 tumors. Statistical comparisons performed with a one‐way ANOVA, multiple testing error controlled using the FDR approach (Q = 5%). Histopathology of the liver and spleen were assessed in B16F10 mice, 3 days after treatment. C) Quantification of degree of liver necrosis. D) Representative images of livers and spleens from treated mice. Scale bars indicate 100 microns. Results are from three independent experiments. N = 8–10 for (A), n = 4–5 for (B), and n = 5 for (C–E).

Figure 6

PNP hydrogels attenuate the induction of pro‐inflammatory serum cytokines compared to dose‐matched local bolus administration. Serum cytokine levels were assessed by Luminex 24 h after treatment. Systemic levels of cytokine‐storm associated cytokines. A) IFNγ, with linear regression of dose‐response data indicating a significant change in the dose‐response slope associated with administration route. B) CXCL10, with linear regression of dose‐response data indicating a significant change in the elevation of the dose‐response curve associated with administration route. Hydrogel delivery suppressed systemic induction of C) IL12, D) IL5, and E) MCP1. F) Linear regression revealed significant differences between administration methods in the dose‐response slope for these cytokines. Hydrogel delivery also suppressed systemic induction of G) TNFα, H) RANTES, and I) IL1b. J) The dose‐response slope is not altered in these cytokines, however the elevation of the dose‐response curve is significantly altered depending on administration route (see Figure S19, Supporting Information). Dotted lines indicate mean value corresponding to the 5 µg systemic dose (maximum tolerated systemic dose). N = 5 for all groups, data represented as mean and SEM. Shaded area in linear regressions indicate 95% CI. Statistically significant differences from the MTD systemic dose were assessed using a regression model, and multiple testing error was controlled using the FDR approach (Q = 5%). * denotes a FDR‐adjusted significant difference (p < 0.05). Statistical comparison of slope and intercept performed using built‐in analysis in the linear regression functionality of GraphPad Prism.

Figure 7

Sustained delivery of CD40a via PNP hydrogel increases effector cytokine levels in the tumor draining lymph node. Lymph nodes were collected from mice 3 days after treatment for cytokine analysis by Luminex. Hydrogel delivery of CD40a led to elevated levels of the pro‐inflammatory cytokine IFNγ and the associated CXCL10 chemokine. A) IFNγ levels with linear regression indicating a significant change in the dose‐response slope associated with administration method. B) CXCL10 levels with linear regression indicating a significant change in the elevation of the dose‐response curve associated with administration method. Compared to systemic routes, locoregional delivery elevated levels of immunostimulatory cytokines C) TNFα, D) IL15, E) IL18, and F) IL22. Similarly, locoregional delivery enhanced levels of the chemokines G) MCP3 and H) GMCSF. Dotted line indicates mean value corresponding to the 5 µg systemic dose (maximum tolerated systemic dose). N = 5 for all groups, data represented as mean and SEM. Statistically significant differences from the MTD systemic dose were assessed using a regression model, and multiple testing error was controlled using the FDR approach (Q = 5%). * denotes a FDR‐adjusted significant difference (p < 0.05). Statistical comparison of slope and intercept performed using built‐in analysis in the linear regression functionality of GraphPad Prism.

Figure 8

Locoregional CD40a therapy transforms the tumor immune microenvironment into a more immunogenic state. B16 tumors were explanted for immunohistochemistry 3 days after treatment, and stained for the pan‐T cell marker CD3 (cyan) and the macrophage marker CD68 (magenta). Blue indicates DAPI nuclear stain. A) Tumor tissue treated with isotype control antibody‐loaded hydrogels. B) Tumor tissue treated with 50 µg dose of CD40a in a hydrogel. C) Tumor tissue treated with 50 µg dose of CD40a as a local bolus. Tumor borders are indicated by the yellow dotted line. i–iii) Magnification of CD3+ cell infiltrated zones as indicated by the white dotted lines in (A)–(C). Single channel grayscale images of the CD3 and CD68 stains in the detailed view are provided below. Scale bars: (A)–(C) denote 500 microns; (i)–(iii) denote 100 microns.

Figure 9

Hydrogel delivery yields effective low dose CD40a monotherapy and synergizes with PD‐L1 blockade. A) Experimental scheme for low dose monotherapy on established B16F10 tumors. Hydrogels containing 10 µg CD40a slowed the B) growth of B16F10 tumors and C) extended survival. D) Experimental scheme for evaluating the impact of additional systemic anti‐PD‐L1 therapy in the treatment of established B16F10 tumors. E) Low dose locoregional CD40a therapy synergizes with systemic PD‐L1 blockade and yields long term survivors in the B16F10 model. Results are from two independent experiments. N = 10 for all groups. Data in (B) indicates mean and SEM. * indicates adjusted p values <0.005. Differences in tumor growth and survival were assessed using general linear models with SAS statistical software. False discovery rate (FDR) was controlled using the Benjamini and Hochberg method (Q = 1%).