Biomaterial-facilitated Immunotherapy for Established Oral Cancers

Abstract

We evaluated a peptide-based immunotherapy termed SynerGel: an injectable, biomaterial-based platform for intratumoral drug delivery. A drug-mimicking peptide hydrogel named L-NIL-MDP was loaded with an anti-tumor cyclic dinucleotide (CDN) immunotherapy agonist. The biomaterial combines inducible Nitric Oxide Synthase (iNOS) inhibition with controlled delivery of CDNs, demonstrating between 4- and 20-fold slower drug release than commercially available hydrogels. SynerGel allowed for immune-mediated elimination of established treatment-resistant oral tumors in a murine model, with a median survival of 67.5 days compared to 44 days in no-treatment control. This report details findings for a promising therapy showing improved efficacy over previous hydrogel systems.

Graphical Abstract

INTRODUCTION

Biomaterial-based drug delivery is a rapidly growing area of research, demonstrating the ability of biocompatible materials to control the release and presentation of bioactive signals and improve therapeutic efficacy.^1-4 Many recent reports have shown the unique utility of biomaterials in cancer immunotherapy, where immunotherapy is a method of treatment that uses a patient’s own immune system to combat cancer, or any other immune-susceptible disease.⁵ Properly designed biomaterials that are biocompatible and biodegradable can be used to interface with living systems and focus payload delivery to stimulate a desired immune response. Common biomaterials that have been used for immunotherapeutic applications include naturally-derived materials such as alginate,^6-8 hyaluronic acid,⁹ Matrigel,¹⁰ or collagen/gelatin^11-12 (among many others), which each have certain advantages and disadvantages in their use.⁵

We have previously reported on the use of synthetic peptide-based biomaterials for extended intratumoral immunotherapy, which are highly customizable due to the modular nature of the amino acid building blocks of peptides.¹³ In a murine model of head and neck squamous cell carcinoma (HNSCC), we reported on STINGel as a cationic lysine-based multidomain peptide (MDP) hydrogel that could control the delivery of anionic cyclic dinucleotides (CDNs), which are immunotherapy agents that act as Stimulator of Interferon Genes (STING) agonists to activate a powerful anti-tumor immune response.¹³ STING agonists (such as CDNs) have shown efficacy against solid tumors in the literature and in ongoing clinical trials, but often require repeated high concentration dosing.^14-16 Excitingly, STINGel resulted in significantly enhanced intratumoral treatment efficacy with only a single injection into early oral tumors (6-fold higher survival compared to treatment with CDN drug alone).

It is well known that many current immunotherapy strategies are hindered by the toxicity of systemically-delivered immunomodulators which often require high doses and frequent re-dosing, increasing the possibility of immune-related adverse events.^17-18 Injectable, drug-carrying biomaterials offer one method of overcoming these limitations, allowing for focused dose delivery, localized response, and reduced off-target effects.⁵ This study expands our work on the use of multidomain peptide (MDP) hydrogels for advanced immunotherapeutic applications. MDPs are a class of easily synthesized biomaterials designed to self-assemble into nanofibrous networks via supramolecular interactions, including hydrophobic packing, β-sheet hydrogen bonding, electrostatic charge repulsion, and ionic crosslinking.^19-20 MDPs can incorporate a large variety of residues and functional groups while still forming nanofibers and hydrogels,²¹ and can be made using the general formula of X_m(Hp)_nX_m, where ‘X’ can be any charged or sterically repulsive residue (e.g. E, D, R, K, or even hydroxyproline [O]), ‘H’ is a hydrophobic residue (e.g. L, I, V, A), ‘p’ is a polar residue (e.g. S, T, N, Q), ‘m’ can be 1-5 residues in length, and ‘n’ 4-7 residues. Part of the study described herein was to compare the drug delivery properties of a number of MDP designs with a focus on charge chemistry (testing the effects of different substitutions at the X positions, see Table 1).

Table 1.

Information on all multidomain peptides (MDPs) used in this report.

Peptide Sequence	Peptide monomer net charge	Abbreviation	Drug-loaded formulation names
KLNIL2(SL)6KLNIL2	+ 4	L-NIL-MDP	L-NIL-MDP+CDN (SynerGel)
K2(SL)6K2	+ 4	K2-MDP	K2-MDP+CDN (STINGel)
E2(SL)6E2	− 4	E2-MDP	E2-MDP+CDN
O5(SL)6O5	0	O5-MDP	O5-MDP+CDN

We have previously demonstrated that the pro-tumorigenic enzyme inducible nitric oxide synthase (iNOS) drives the induction and functional activation of immunosuppressive tumor-infiltrating myeloid-derived suppressor cells; polarizes macrophages towards the tumor-promoting M2 phenotype; and regulates CD4⁺ T cell differentiation.^22-24 We further described the anti-tumor immunomodulatory activity of iNOS inhibition as a component of immunotherapy approaches in combination with chemoradiotherapy²⁵ and checkpoint inhibitor-based radio-immunotherapy.²⁶ These findings suggest that the ability to modulate iNOS activity directly at the tumor site would be beneficial for cancer immunotherapy. We therefore reported the synthesis and characterization of a novel drug-mimicking biomaterial termed the L-NIL-MDP.²⁷ L-NIL, or N6-(1-iminoethyl)-L-lysine, is a potent small molecule inhibitor of iNOS.²⁸ We established that an L-NIL-mimicking MDP hydrogel bearing charged acetamidine functional groups could be readily synthesized and show durable L-NIL activity, generating an iNOS-inhibiting gel that would remain in vivo for weeks after only a single intratumoral injection.^27-28 Furthermore, the L-NIL-MDP showed an extended ability to reduce tumor vascular endothelial growth factor (VEGF) levels, a key modulator of immune function.²² These unique properties suggested the L-NIL-MDP biomaterial would be an excellent vehicle candidate for immunotherapies.

Our primary goal in this study was to investigate the efficacy of the iNOS-inhibiting L-NIL-MDP biomaterial for extended and focused immunotherapy delivery in a challenging model of HNSCC. Herein we describe the results from our first experiments with cyclic dinucleotide (CDN) STING agonist-loaded L-NIL-MDP, termed SynerGel. We demonstrate the ability to extend the release of anionic CDNs from the cationic L-NIL-MDP (see supporting Figure S1 for chemical structures) compared to different MDP designs and other common biomaterials used in this field (e.g. alginate, hyaluronic acid, collagen, and Matrigel), show an influence on the local immune response in vivo, and treat established solid tumors with only a single intratumoral injection of this dual-function biomaterial system.

RESULTS AND DISCUSSION

CDN release kinetics from MDPs and commercially available hydrogels

We first investigated if the drug-mimicking L-NIL-MDP hydrogel could effectively load and control the delivery of CDN immune agonists compared to other hydrogel systems. As shown in Figure 1, we performed a study comparing the CDN release properties of various types of in-house synthesized MDP hydrogels with different side-chain chemistries (see Supporting Information for synthesis methods) and commercially available hydrogels commonly used in immunotherapy applications. Table 1 contains sequence and abbreviation information on all synthetized MDPs used in this report. The following commercial hydrogels selected for testing have been used widely in the literature for localized immunotherapy drug delivery: sodium alginate (a highly anionic polysaccharide derived from seaweed), hyaluronic acid (an anionic polysaccharide derived from the extracellular matrix [ECM]), collagen (a key ECM component composed of peptide triple helices, in this case sourced from rat-tail collagen), and Corning® Matrigel (primarily composed of collagen and laminin). CDN small molecules possess negatively charged thiophosphates that link the molecules’ nucleotides (Figure S1), which we hypothesized could favorably interact with positively charged materials via intermolecular hydrogen bonding and electrostatic charge pairing, resulting in significantly different CDN release rates between chemically diverse materials. Different hydrogel systems were prepared at similar drug loading concentrations (0.9 mM) and similar material concentrations by weight according to manufacture instructions for each purchased hydrogel system (see Table S1 for detailed information). We expected that the differences observed in release kinetics could be primarily attributed to the presence or absence of favorable chemical interactions between CDN drug molecules and the biomaterial vehicles.

Figure 1.

(A, C) Drug release profiles of cyclic dinucleotide (CDN) ML RR-S2 CDA from various hydrogels systems over 24 hours, comparing commercially available alginate, hyaluronic acid (HyAcid), collagen, and Matrigel systems to synthesized multidomain peptides (MDPs) K2-MDP, E2-MDP, O5-MDP, and L-NIL-MDP, with quantification of release profiles shown in (B). All release data are shown as means ± S.D. with n = 3. Data for K2-MDP and collagen hydrogels are reproduced from our previous study for comparison.¹³ (D) Rheological storage modulus (G’) of studied MDP hydrogels, showing bulk material strength was not predictive of drug release properties.

Our results show that the highly anionic polysaccharides alginate and hyaluronic acid hydrogels showed the fastest CDN release rates, and released the greatest percentage of loaded CDN (~95%), suggesting a lack of favorable interactions and most likely charge repulsion with the negative CDN payload (Figure 1A). Collagen and Matrigel hydrogels (derived from protein-based extracellular matrix components) also showed relatively fast CDN release, and all four control hydrogels showed significantly faster release compared to L-NIL-MDP. L-NIL-MDP (SynerGel) demonstrated between 4- and 20-fold slower release in the initial linear phase, taking 18 hours to reach equilibrium in this in vitro model (Figure 1C). Furthermore, as has been observed previously for K2-MDP, L-NIL-MDP reached a biased equilibrium in this model (only ever reaching ~60% observed CDN release), providing further evidence of favorable material-to-drug interactions that hold drug payload within the hydrogel for extended periods of time. It also is interesting to note that all tested MDPs (the anionic E2-MDP, neutral O5-MDP, and cationic K2-MDP and L-NIL-MDP) showed slower (i.e., better, more controlled) CDN release profiles than any of the commercially available hydrogel systems (Figure 1C), with the tested MDPs ranging from 10-18 h to reach equilibrium, and the commercial systems ranging from 0.5-8 h. We hypothesize this may be due to known physical material differences, such as the smaller average peptide fiber size and pore size of synthetic MDPs compared to the large fiber and pore sizes of the tested biologically-derived materials.^{4, 20}

In comparing only the synthetic MDPs (Figure 1C), L-NIL-MDP showed a similar yet slightly slower release profile compared to the lysine-based K2-MDP used in our previous STINGel study,¹³ which was expected due to its similarly cationic acetamidine side-chains. L-NIL-MDP also had an approximately 3-fold slower release rate than hydroxyproline-based O5-MDP and glutamate-based E2-MDP. Interestingly the anionic E2-MDP and charge-neutral O5-MDP showed nearly identical release profiles to each other, despite differences in side-chain chemistry and significant differences in bulk material rheological strength (Figure 1D). Indeed, despite O5-MDP being a relatively weak and compliant hydrogel, and E2-MDP showing similar strength to the other cationic MDPs, clearly bulk material strength did not influence small molecule CDN diffusion, but rather the presence (or absence) of favorable intermolecular interactions governs the observed release kinetic profiles.

In vitro and in vivo characterization of L-NIL-MDP hydrogel

We next characterized how unloaded L-NIL-MDP (as the biomaterial component of SynerGel) interacted with biological systems both in vitro and in vivo, in order to directly compare it to the K2-MDP previously used in STINGel formulations. Murine oral cancer cells (MOC1) cultured on L-NIL-MDP showed high viability according to live-dead assays and were able to proliferate on the hydrogel over 7 days (Figure 2A-C), growing from the low density seeding of only a few cells on the hydrogel surface (Figure 2A) to larger clumps (Figure 2B) and finally large communities that covered the entire hydrogel surface (Figure 2C). This demonstrated that the L-NIL drug-mimicking scaffold was biocompatible and not directly cytotoxic to cancer cells, but facilitated cell attachment and growth which is consistent with other cationic hydrogel materials we have studied.⁴

Figure 2.

(A-C) Viability of mouse oral cancer (MOC1) cells cultured on L-NIL-MDP hydrogels. All cell-gel experiments were cultured under 200 μL of media and processed for live-dead viability assays at time-points of days 1, 3, and 7 (green live cell-Calcein AM; red dead cells-Ethidium-homodimer I; blue nuclei-Hoechst 33342). (D) Masson’s trichrome stained subcutaneous histology of K2-MDP hydrogel implant three days post injection in vivo at 4X magnification. (E-F) 40X magnification images of the black boxes in panel D, with image E on implant edge, and image F in implant core. (G) L-NIL-MDP hydrogel implant three days post injection at 4X magnification. (H-I) 40X magnification images of the black boxes in panel G, with image H on implant edge, and image I in implant core. Example areas of undegraded or non-infiltrated hydrogel have been marked by asterisks for clarity.

Subcutaneous injections of hydrogels in the dorsal flank of healthy mice allowed for a basic characterization of the host immune response to our materials, with Masson’s trichome stained histology images shown in Figure 2D-I. As expected, the K2-MDP, which has been previously characterized,²⁹ showed homogenous cellular infiltration of immune cells (Figure 2E-F) known to be primarily monocytes and macrophages (as first responders participating in an acute immune response to the biomaterial).⁴ However, it was interesting to observe that the L-NIL-MDP showed a much less homogenous cellular infiltration at early time points (day 3 post injection). Cell nuclei were instead observed to be more concentrated in pools and channels around large islets of undegraded hydrogel throughout the implant (Figure 2H-I), suggesting a distinct immune response to this material compared to K2-MDP. Our prior studies have shown L-NIL-MDP hydrogels degrade much slower in vivo, and are easily observable 21 days post injection, compared to similar size K2-MDP hydrogels not being easily observable past approximately 14 days.²⁷ We hypothesize that this slower degradation profile may be directly related to the way interacting immune cells (such as macrophages, which are known play a large role in material phagocytosis) infiltrate and remodel the material. Consequently, the material’s iNOS inhibitory activity may influence the normal function of macrophages (which are high expressors of iNOS) in ways we do not yet understand, and we hypothesize that the sustained presence of the bioactive hydrogel in its local environment in vivo allows for longer therapeutic action. However, these results also confirmed the biocompatibility of L-NIL-MDP in vivo, showing that the hydrogel can interact with and be infiltrated by immune cells, which is a critical element of any immunotherapy-based approach.

Efficacy of CDN-loaded L-NIL-MDP (SynerGel) in a murine model of HNSCC

We previously reported the efficacy of STINGel (K2-MDP hydrogel loaded with CDN) in a non-palpable murine model of human papilloma virus (HPV)-associated HNSCC, termed MOC2-E6E7. In this current study, we evaluated the efficacy of L-NIL-MDP hydrogel loaded with CDN (SynerGel) compared to STINGel in MOC1 tumors, a murine model of non-HPV-associated HNSCC which does not express ectopically generated viral antigens. Figures 3 and 4 show the results from tumor treatment efficacy studies in C57BL/6 mice bearing established (4-5 mm diameter) MOC1 tumors in the oral cavity. Growth of MOC1 tumors in mice treated with HBSS (Hank’s Balanced Salt Solution), CDN alone, K2-MDP, K2-MDP+CDN (STINGel), L-NIL-MDP or L-NIL-MDP+CDN (SynerGel) was investigated to compare L-NIL-MDP+CDN’s anti-tumor efficacy with these treatment groups. Following treatments with L-NIL-MDP+CDN, we observed decreased tumor growth in MOC1 tumor-bearing mice compared to K2-MDP+CDN (Fig. 3A). Individual tumor growth curves over the course of study and including a secondary MOC1 rechallenge after day 100 post-tumor inoculation indicate L-NIL-MDP+CDN (SynerGel) to be a more effective biomaterial-based immunotherapy as compared to K2-MDP+CDN (STINGel) (Fig. 3B-G).

Figure 3.

MOC1 tumor growth curves. (A) Mean tumor growth until time of first mouse euthanization, (B-G) Individual tumor growth curves for controls and L-NIL-MDP+CDN treated mice bearing established MOC1 oral tumors over course of study. Tumor rechallenge was at day 100 post-inoculation (indicated by dotted line). The number of tumor-bearing mice that were euthanized is listed above each plot. L-NIL-MDP+CDN (SynerGel) treated mice had decreased tumor growth. a, p<0.01 vs K2-MDP; b, p<0.0001 vs. HBSS (Hank’s balanced salt solution), K2-MDP, L-NIL-MDP; #, p<0.01 vs. K2-MDP+CDN; c, p<0.0001 vs. HBSS, K2-MDP, K2-MDP+CDN and L-NIL-MDP.

Figure 4.

Survival curves for mice bearing established MOC1 oral tumors. Kaplan-Meier curves of the different experimental groups is based on euthanasia timepoints due to tumor burden. Intratumoral injections (IJ) were given when tumors were at 4-5 mm, represented on the x-axis as 5mm(IJ). Survivor rechallenge (RC) was done at post day 100 time point. L-NIL-MDP+CDN (SynerGel) treated mice had improved survival. SynerGel increased survivorship from 20% to 33% compared to CDN alone. Log-rank test; **, p< 0.01 vs. HBSS; ^^, p< 0.01 vs. K2-MDP; ##, p< 0.01 vs. L-NIL-MDP.

Survival was significantly increased when the L-NIL-MDP+CDN (median survival, 67.5 days) treatment group is compared to HBSS (median survival, 44 days); L-NIL-MDP (median survival, 44 days) and K2-MDP (median survival, 46 days) (Fig. 4). Additionally, L-NIL-MDP+CDN was the only group to display statistically significant improvement in median survival over HBSS control. Treatment with L-NIL-MDP+CDN resulted in 33% of mice surviving to the 150-day endpoint compared to 20% for CDN alone and 8% for K2-MDP+CDN (although not statistically significant). These data suggest L-NIL-MDP+CDN (SynerGel) had better anti-tumor efficacy than the control groups and was more effective than K2-MDP+CDN (STINGel) in the treatment of MOC1-bearing mice, illustrating the potential of L-NIL-MDP as a platform for biomaterial-based immunotherapies.

Successful immunotherapeutic targeting of established HNSCC tumors remains a challenge. Elimination of established MOC1 tumors has typically required multiple injections of cyclic dinucleotides (CDNs).¹⁶ We previously reported the successful treatment of early, HPV-associated MOC2-E6E7 tumors with STINGel.¹³ However, in this study more established, non-HPV associated MOC1 tumors were relatively unaffected by single intratumoral STINGel injections. Several reasons may contribute to the differences in outcome. Established oral tumors are known to create an immunosuppressive tumor immune microenvironment (TIME), driving resistance to standard immunotherapy by suppressing the immune system’s ability to recognize and eliminate cancerous cells, also known as immune escape.^30-31 Therefore, established MOC1 oral tumors are likely more challenging to treat than the non-palpable MOC2-E6E7 oral tumors. Furthermore, MOC1 tumors lack any ectopically expressed antigens such as the E6/E7 HPV viral proteins which may help drive a robust, specific, anti-tumor immune response.¹³

Additional mechanisms of resistance may be present in the MOC1 tumor microenvironment. The immunosuppressive MOC1 TIME is likely reducing the efficacy of CDN and STINGel treatments, which are mainly designed to enhance effector immune mechanisms. However, MOC1 tumors are highly infiltrated by immunosuppressive cell types such as myeloid derived suppressor cells (MDSCs).³² The strong presence of inhibitory immunocytes in the MOC1 tumor microenvironment³³ informed the rational design of SynerGel, replacing the K2-MDP hydrogel component in STINGel with a second generation, drug-mimicking L-NIL-MDP hydrogel. L-NIL has been shown to reverse MDSC-mediated immunosuppression through modulation of tumor-induced inflammation, as iNOS is overexpressed in nearly every solid tumor type where it supports the development of a profoundly immunosuppressive TIME.^34-36 Our results support our hypothesis that biomaterial-based iNOS inhibition combined with STING activation (SynerGel) can effectively treat established tumors, successfully overcoming a tumor immune microenvironment non-responsive to traditional immunotherapy.

CONCLUSIONS

We have shown that the L-NIL-MDP is an effective, bioactive carrier material for immunotherapies in an established tumor model, requiring only a single intratumoral injection of L-NIL-MDP+CDN (SynerGel) for full elimination of 4-5 mm tumors. We confirmed that the cationic properties of the L-NIL-MDP hydrogel allow it to extend the release of anionic cyclic dinucleotides, with 4- to 20-fold greater controlled release compared to various other commonly used hydrogel systems. The combination of the L-NIL-MDP hydrogel with its inherent iNOS inhibition and the controlled release of STING agonist immunotherapy successfully increased the survival of MOC1-tumor burdened mice, with a median survival of 67.5 days compared to 44 days in a no-treatment control. Total survivorship also increased from 20% in CDN drug alone groups to 33% in SynerGel treated groups. We have demonstrated that we can use the L-NIL-MDP biomaterial as a successful alternative to treatment strategies that utilize the small molecule drugs L-NIL or CDNs. Future studies will involve characterization of the mechanisms of action of L-NIL-MDP formulations (SynerGel as well as other combinatorial immunotherapies). These formulations will continue to be designed for long-lasting, immune cell-infiltrated biomaterial immunotherapy at the tumor site, allowing for transformation of the local immunosuppressive microenvironment to a more treatment-responsive tumor.