Harnessing biomaterials to study and direct antigen-specific immunotherapy

Abstract

Immunotherapies are an evolving treatment paradigm for addressing cancer, autoimmunity, and infection. While exciting, most of the existing therapies are limited by their specificity—unable to differentiate between healthy and diseased cells at an antigen-specific level. Biomaterials are a powerful tool that enable the development of next-generation immunotherapies due to their tunable synthesis properties. Our lab harnesses biomaterials as tools to study antigen-specific immunity and as technologies to enable new therapeutic vaccines and immunotherapies to combat cancer, autoimmunity, and infections. Our efforts have spanned the study of intrinsic immune profiles of biomaterials, development of novel nanotechnologies assembled entirely from immune cues, manipulation of innate immune signaling, and advanced technologies to direct and control specialized immune niches such as skin and lymph nodes.

I. Introduction

Cancer, autoimmunity, and infectious disease impact people of all nationalities, genders, and socioeconomic statuses. In each of these disease areas, immunotherapies have emerged as an exciting area of study to provide better outcomes for patients. Immunotherapies harness the power of the immune system to drive potent effects. During healthy immunity, fragments of pathogen or cancerous cells – termed antigens, are recognized by antigen-specific B and T lymphocytes. These cells then differentiate and expand to seek out and destroy the corresponding pathogen or cancerous tissues. Conversely, during healthy immune function, tolerance is maintained against host tissue and self-antigens; this prevents inflammation and autoimmunity – attack of host tissue. Despite this well-regulated system, abnormal cell growth, autoimmunity, and infectious diseases become pathologic when normal mechanisms of immunity break down. For example, during cancer immunosuppressive tumor environments prevent effective T cell responses from infiltrating tumors and destroying cancerous cells. Cancer immunotherapies such as chimeric antigen receptor T (CAR-T) cells and checkpoint blockade inhibition seek to enable cancer cell killing and overcome such suppressive environments. However, these therapies are limited in the varieties of cancers and patient cohorts they treat, as well as significant time and monetary costs. In autoimmunity, tolerance to self-antigen breaks down, and self-tissue is attacked as though it were pathogenic. Novel monoclonal antibodies that target specific aspects of the immune system—such as CD3⁺ T cells in Type 1 Diabetes (T1D),¹ can improve the course of disease. Despite these advances, monoclonal therapies do not differentiate between disease-causing and healthy cells. Likewise, in infectious disease vaccines have been transformative and even eradicated some diseases, but infections such as HIV and malaria avoid immune cell killing by targeting immune cells directly (HIV) or by evasion through changing antigens. In each case, of these contexts – infection, cancer, autoimmunity – next generation immunotherapies will benefit from increased selectivity and potency to meet unmet need.

The expansion of T cells to kill cancer cells (cytotoxic CD8⁺ T cells), treat infection (inflammatory CD8⁺ and helper CD4⁺ T cells), or promote tolerance via regulatory T cells (T_REG) requires a precise coordination of signals both spatially and temporally. Antigen presenting cells (APCs) such as dendritic cells (DCs) present antigen - as well as soluble cues and costimulatory molecules - to modulate immune function towards either inflammation or tolerance. Processing of antigens via APCs is a critical because it helps direct the polarity of a T cell response. For example, pathogen-sensing receptors, such as toll-like receptors (TLRs), recognize conserved pathogen-associated motifs and drive activation and maturation of APCs.² This promotes the presentation of antigen in inflammatory contexts. In contrast, APCs exposed to regulatory cues – including certain metabolites or antagonistic innate cues (e.g., TLR antagonists, which regulate rather than activate TLR pathways) can prevent development of inflammatory responses. Strategies to control the juxtaposition of antigen and such cues that control APC activation—and ultimately T cell activation—are one important way to gain increased control over the nature of immune responses.

Biomaterials are tools that enable precision control over presentation and delivery of immune cargo through tunable properties, including cell and tissue targeting, co-loading of multiple cargos, and controlled release over time.^3–6 These qualities can be influenced by the inherent properties of materials, the cargo loaded on or within biomaterial constructs, and the local microenvironments in which these constructs are processed.⁷ Biomaterials are thus valuable technologies to study and create immune engineering applications areas spanning autoimmunity,⁸ cancer,⁹ and infectious disease¹⁰. Over the past decade our lab has focused on biomaterials as tools to study and direct immune processes, with an emphasis on T cell function and antigen-specific therapies (Fig 1). Our work has spanned the engineering of intrinsic material properties, leveraging entropically-favorable self-assembly processes to build biomaterials entirely from immune cues, targeting disease-associated signaling cascades, and using scaffolds and depots to locally condition immunological niches.

Fig 1: Overview.

A variety of technologies can be harnessed to study and manipulate immunotherapies across tolerance, immunity, and intrinsic immune profiles of materials.

II. Controlling intrinsic properties of biomaterials to dictate immune response.

A key reason biomaterials have emerged as exciting tools for immunomodulation is because of their tunable characteristics that provide precision control to direct immune signal processing and response.^11,12 One aspect of this tunability is the intrinsic immune profiles of many materials that engage innate immune pathways, often as a function of changing characteristics, such as polymer degradation, molecular weight, and electrostatic interactions. Understanding and manipulating these factors empowers biomaterials unique control over how immune responses develop. Our lab has used each of these variables to better control T cell responses both in the context of inflammatory and regulatory contexts.

Polymers, along with other materials, offer controllable properties such as molecular weight, functional groups, and blends that enable tunable biological outputs. Varying molecular weight or makeup can impact key biophysical properties including degradation rate and stiffness, which can each affect the way a material is interacted with by a cell. One polymer class of interest for vaccine and immunotherapy delivery are poly(β-amino esters) (PBAEs). PBAEs are positively charged polymers that complex readily with cargo such as nucleic acids to form particles for delivery. An advantage of PBAEs is they degrade quickly (i.e. hours to days) relative to more ubiquitous polymers such as PLGA; this time scale is favorable for many immunotherapy applications. Specifically, our lab has focused on understanding how the immunogenic properties of PBAEs vary with polymer characteristics (i.e. molecular weight, degradation rate). Understanding how intrinsic material properties impact immune pathways is critical for rational design of vaccines and immunotherapies. In one set of studies, we tested the impact of PBAEs with varying molecular weight on immune cell activation, as well as the difference between soluble PBAE chains and PBAEs particles (2A).^13,14 PBAE particles were found to upregulate dendritic cell activation markers more strongly while soluble PBAEs were less immunogenic (Fig 2B). Importantly, the immunogenicity of degraded PBAEs was highest at early points in degradation, while decreasing over time. This finding demonstrated that immunogenicity evolves during the course of material degradation, which has significant implications across vaccine and immunotherapy applications incorporating degradable polymers.

Fig 2: Tunable material properties dictate immune outcomes.

A) Schematic of experimental design of PBAE and DC activation study. PBAEs were fragmented, then administered to DCs as either particles or free PBAE polymers. B) APC Activation as measured by mean fluorescence intensity (MFI) of the DC maturation marker CD80. Darker bars indicate particulate formulation and lighter bars indicate soluble formulations. C-D) NF-κβ (C) and interferon response factor (IRF) signaling (D) as a function of PBAE treatment after 48 hours of culture. E) Schematic of PLGA microdisk synthesis with immunogenic cargo—SIINFEKL antigen (SIIN) and CpG adjuvant. H) Cartoon of QD functionalized with MOG antigen. G) Clinical Score curves for experimental autoimmune encephalomyelitis among mice treated with QDs synthesized with varying antigen density. A higher clinical score indicates increased disease severity. H) T_REG proportions in draining LNs as a function of antigen density in QD synthesis. I) Proportion of drug release measured before and after ice-cooling release step.

To connect natural molecular weight decreases during polymer degradation specifically to molecular weight, we synthesized a library of PBAEs with varying chain lengths via Michael addition. We found that there was an ideal range of 1500-3000 kDa for inducing immunogenic responses in DCs. While understanding the innate role of PBAE characteristics on immunogenicity is important, a strength of this polymer class is the ability to complex with negatively charged nucleic acids to deliver oligonucleotides, genes, and other cargo. As these complexes have characteristics distinct from polymer alone, we sought to test the impact of PBAEs complexed with RNA on innate immune cells.¹⁵ To focus on the effects of the biophysical changes of RNA complexation, an immune-irrelevant RNA sequence was used. We found that the polymer complexes activate macrophages independently of NF-κβ, while still activating the interferon-response factor (IRF) pathway (Fig 2C–D).¹⁵ NF-κβ is a major innate immune pathway, so the independence of this mechanism is both surprising and significant. The importance of material properties on immune activation also suggest that variability in synthesis could lead to variable immune activation, a challenge for clinical translation. To overcome variability in polymer particle synthesis, we have also developed a uniform microdisk vaccine platform where particles are prepared in a PDMS stamp mold (Fig 2E).¹⁶ The result of this is incredibly uniform particles that alleviate processing variability in other synthesis methods while allowing incorporation of a range of immune signal combinations and concentrations. These types of advanced manufacturing technologies could uniquely enable tuning of distinct polymer characteristics and immune cue delivery provides for modulating immunity.

Another key variable that can be manipulated using biomaterials is signal density and display conformation. The display density of antigen, for example, impacts interactions with immune cells by altering the strength and valency (i.e., number of contacts) with cell surface receptors. To test the impact of antigen density on immune responses we used a molecularly defined quantum dot (QD) platform.¹⁷ QDs are monodispersed, nanoscale particles with the capacities to present biomolecules at controllable densities (Fig 2F). We synthesized QDs functionalized with varying quantities of myelin autoantigen self-peptides to test the impact of this parameter on efficacy during a mouse model of multiple sclerosis—experimental autoimmune encephalomyelitis (EAE). The EAE model drives inflammation and destruction of myelin, resulting in progressive paralysis in mice over time; paralysis severity is indicated by a clinical score. Using the EAE model with QDs, treatment with a higher number of QDs with a lower MOG density reduced clinical scores compared to fewer QDs with a higher antigenic density, underscoring the role of antigen density in therapeutic efficacy (Fig 2G). Additionally, lower density QDs resulted in a delay in disease onset (Fig 2H). Thus the ability biomaterials provide to control ligand display density creates additional options to direct how immune cues are integrated for selecting or enhancing responses.

In addition to the described properties above, the ability of a material to respond to particular environmental triggers enables more precise immunotherapy delivery. For example, many carriers are designed to be responsive to either heat or pH to promote degradation in conditions such as a lysosome or tumor microenvironment. This selectivity is enabled by rational design of polymers and functional groups that are responsive to desired conditions.¹⁸ Our team has developed a novel polymer particle formulation that is specifically responsive to cold conditions, such as those present in cryoablation therapy for cancer. To achieve this, particles were synthesized with a polymer that has a lower critical solution temperature lower than room temperature—poly(N-isopropylacrylamide-co-butyl acrylate) (PNIPAM-B). When cooled, the polymer becomes soluble in solution, driving rapid release of an anti-cancer payload (Fig 2I).

III. Self-assembled materials to probe and regulate immune responses.

Beyond their functional role, immune signals themselves – such as proteins, nucleic acids, and peptides - can serve as structural components of biomaterial systems. Our lab has reported work in this area using carefully selected components based on both their immunological functions and chemical properties. We have expanded this concept for immunotherapy platforms that can be applied to a variety of disease settings¹⁹. These strategies allow different modalities to be integrated to specifically activate or polarize APCs and T cells²⁰. From a synthesis perspective, our self-assembly platforms utilize important classes of interactions, including electrostatics and hydrophobicity.^21,22 These approaches enable precise loadings of immune cargo, such as antigen and adjuvant, at ratios that are ideal to elicit an immune response. Specifically, our lab has pioneered nanoscale immune complexes (Fig 3A) and layer-by layer templated immune capsules (Fig 3B). We have utilized both traditional exogenous polymers as well as natural immune cues to form these materials (Fig 3C).

Fig 3: Self- assembly of immune signals polarizes APCs and causes expansion of antigen-specific T cells, leading to improved disease outcomes.

Our lab has synthesized (A) nanoparticle complexes and (B) used layer-by-layer (LbL) synthesis to study the impact of immune signals with or without traditional polymers. (C) Combinations of nanoparticle complexes or layer-by-layer assembled particles can be formed with or without traditional polymer carriers. (D) Representative SPR graph used to identify the binding affinity of antigen to TLR antagonist. (E) The type of amino acid appended to antigen to increase positive charge for self- assembly dictates the binding affinity of antigen to a TLR antagonists—specifically, arginine- modified antigens have a smaller K_D and theys higher binding affinity than lysine- modified antigens. (F) The lowest binding affinity complex improved T_REG induction. (G) Both disease-relevant antigen and tolergenic cue are critical for driving antigen-specific, tolergenic responses in DCs, as measured by the gene expression of DCs isolated from spleens of C57BL/6 mice were analyzed following stimulation with CpG and treatment with iPEMs. Unsupervised hierarchical clustering of samples and genes separates formulations containing GpG and those containing ODN-CTRL. Gene expression in the heatmap is standardized for comparison across multiple genes. Each culture replicate (a single column of the heatmap, n = 4 per condition) is an aggregation of n = 7 technical treatment replicates. (H) Human patient samples treated with iPEMs containing disease-relevant antigen and tolergenic cue (blue bars) attenuate inflammatory cytokine secretion compared to untreated (black) or PBMCs treated with iPEMs comprised of disease-relevant antigen and control cue (blue). Study was conducted with 3 patient samples; representative plots shown. (I) iPEMs can contain combinations of different immune adjuvants, which can enable rationally-selected adjuvants. (J) In a prophylactic vaccine study, mice treated with a combination of TLRas delivered together in iPEMs significantly improved survival and (K) slowed tumor growth compared to single-TLRa treated mice or untreated mice. All data were collected in triplicate and are representative of three comparable experiments. Error bars represent the mean standard error. SEM and p values were considered significant as defined by: *p<0.05; **p<0.01; ****p<0.0001, or as indicated in the figure.

First, our lab focused on incorporating immune signals with traditional polymers to enable controlled delivery of immunotherapies and vaccines. To test the effect of electrostatic forces on cancer immunotherapies, we created self-assembled complexes comprised of positively charged PBAEs and a negatively charged TLR agonist, CpG ²³. The rationale of this immunotherapy was that CpG, a potent immune adjuvant, must be internalized within cells to be recognized by a TLR. We hypothesized that CpG complexed with positively charged PBAEs could drive increased cellular uptake, increasing immunogenicity for cancer immunotherapy. To control electrostatic interaction strength, the ratio of positive to negative charge (PBAE to CpG) was varied in complex synthesis. We found that complexes exhibiting maximal positive surface charge were most effectively internalized into antigen presenting cells (APCs). These complexes were more tightly bound to one another compared to other charge ratios. Interestingly, despite a stronger affinity and uptake, these complexes were less effective at stimulating inflammatory pathways as measured by APC activation and IFN-γ secretion. We consistently found that more positively charged complexes drove stronger interactions; however, these strongly complexed particles were less effective at stimulating downstream immune responses for immune activation for cancer²⁴.

In addition to complexing PBAEs with TLR agonists and antagonists in simple mixtures, our lab used polyelectrolyte multilayer (PEM) technology to layer oppositely charged PBAEs with TLR antagonists onto microparticle templates. These self-assembled particles exhibit controllable geometry, loading, charge, and release to promote immune cell uptake and response. Specifically, this concept enabled sustained release of a TLR9 antagonist, GpG, ultimately reducing inflammatory cytokine secretion and polarizing T cells towards tolerogenic phenotypes in multiple sclerosis (MS)²⁵.

New advantages, such as 100% signal density and eliminating inherent immunogenicity from exogenous polymers, are achieved through two novel classes of self-assembled materials that our lab reported. These materials are comprised entirely of immune signals spontaneously self-assembled using electrostatic interactions, which juxtapose signals at an extremely high density. Both of these properties are important in efficiently polarizing the effector cells that drive immune responses. Furthermore, these biomaterials eliminate inherent immunogenicity from exogenous polymers, enable colocalization of positively charged antigen with negatively charged TLRa at 100% signal density since there is no diluting carrier polymer or matrix. This improves uptake of components by immune cells compared to soluble signals²⁶. One type of material are immune polyplexes, which are condensed nanoparticles of antigen and adjuvant. By replacing the positively-charged PBAE with antigen modified to contain additional positive amino acid residues, polyplexes effectively protected negatively-charged GpG nucleic acid from enzymatic degradation and blunted TLR signaling in vitro. By blunting signaling in the context of this antigen, these particles provide a way to selectively downregulate inflammatory responses against antigen specifically. This has important implications in the field of autoimmunity, where TLR signaling is frequently upregulated.

In EAE, these polyplexes reduced myelin-driven inflammation. Interestingly, polyplexes where smaller amounts of positive charge appended to the Trp2 human melanoma antigen resulted in significantly slowed disease onset and prolonged survival.²⁷ Both of these studies show how nanoparticles comprised entirely of immune signals can be used to improve outcomes in autoimmune and cancer contexts.

Recently, our lab has used the definable properties and antigen sequences in immune polyplexes to study how the binding affinities between the positively- and negatively- charged components influence the biophysical properties and resulting immune response of self-assembled particles. To examine this role in the context of autoimmunity, we tested the effect of varying antigen sequence – and thus electrostatic affinity - in complexes self-assembled from myelin autoantigen and the toll like receptor (TLR) antagonist GpG—an oligonucleotide.²⁸ TLR signaling drives activation of innate immunity in response to pathogen-associated molecular patterns, so the rationale of this design is to inhibit activation via this pathway during delivery of autoantigen to promote non-inflammatory or tolerizing antigen presentation to T cells. To modify the charge of the myelin antigen, the peptide was functionalized with varying numbers of positively charged amino acids: lysine and arginine. We tested the binding affinity of these modified antigens to GpG using surface plasma resonance (SPR) (Fig 3D). SPR analysis revealed that arginine-modified antigens had a smaller K_D and thus a higher binding affinity than lysine-modified antigens (Fig 3E). Within the same type of amino acid, more positively charged groups – and thus higher valency interactions - drove stronger binding. Excitingly, we found that this impacted the generation of antigen-specific T_REG cells. The lowest binding affinity complex promoted the greatest induction of T_REG, while the higher binding affinity groups had reduced T_REG (Fig 3F). This suggests that electrostatic forces can directly affect the ability to induce a potentially therapeutic cell population.

Our lab invented a new class of biomaterials, using a layer-by-layer approach for precision assembly of multiple classes of immune cues. We termed these structures immune polyelectrolyte multilayers (iPEMs) and have developed approaches for injectable capsules and for coating on to microscale devices such as microneedles. Like immune polyplexes, these particles are carrier-free and rely on electrostatic interactions between adjuvants and antigens. However, iPEMs have the added advantage of control over the total and relative loading of components into particles. iPEMs are similar to traditional PEMs containing PBAEs discussed above in that these particles leverage layer-by-layer technology. In contrast, iPEMs involve coating layers of each immune component onto a calcium carbonate core, which is then removed by chelation to create carrier-free capsules. For instance, the size and loading of these particles could be controlled by varying the number of layers and the pH of template dissolution.²⁹ This unique, modular platform enables probing of the immune system without confounding impacts of biomaterials themselves. Our lab elucidated the relative role and importance of two immune signals in EAE ³⁰. By creating a library of iPEMs containing either relevant antigen (myelin self-antigen) or an irrelevant antigen, as well as a tolerogenic cue (a TLR9 antagonist, GpG) or non-signaling cue, our lab elucidated the relative role and importance of each component in EAE³¹. We found that the disease relevant antigen (MOG) and tolergenic cue (GpG) were both critical for driving tolerance in EAE (Fig 3G). This finding also has relevance to human samples. In peripheral blood mononuclear cells (PBMCs) derived from patients with MS, iPEMs containing GpG restrained myelin-derived activation relative to an irrelevant (non-signaling) nucleic acid. Compared to untreated samples, iPEMs containing disease relevant antigen and tolerogenic cue were able to preserve metabolic function while reducing inflammatory cytokine IL-6 (Fig 3H), TNFα, and IFNγ³². In a cancer disease model, iPEMs containing a model antigen (SIINFEKL) with a single TLR agonist (TLR3a) caused both components to drain to the lymph node (LN), expanding T cells within the draining LN and spleen of treated animals³³. This expansion in the LN is critical because it shows the systemic effects of a locally-delivered therapy³⁴. More recently, our lab sought to include multiple TLRas to discern how immune responses could be tuned via synergistic outcomes (Fig 3I). Specifically, this advanced iPEMs as a platform technology by testing how different outcomes could be generated against the same antigen depending on which TLR agonists were included. These studies showed that the ratio of included TLRas dictated levels of corresponding TLR activation and levels of tumor protection in a mouse melanoma model. Interestingly, in a prophylactic vaccine regimen, a vaccine of iPEMs containing a 1:1 ratio of TLR3a:TLR9a with antigen protected mice better than soluble signals or iPEMs containing a single TLRa and antigen (Fig 3J,K) ³⁵.

IV. Direct targeting of immunological niches/metabolism

Previous sections discussed our lab’s use of biomaterial design parameters and cargo selection to study and direct T cell function by manipulating specific pathways. Another goal of our lab has been to use biomaterials to exploit immunological geography, locally conditioning key immunological niches at the cell and tissue level. In this area, we have delivered a variety of cues, metabolites, and drugs to niches including skin and LNs.

Immunological processes are governed and accompanied by broad changes in metabolic states; for example, memory T cell metabolism is driven by aerobic respiration, while actively dividing effector cells are highly glycolytic. As such, metabolites have been widely demonstrated to promote changes in immune cell phenotype, cytokine production, and function (Fig 4A). As a result, delivery of immunometabolic cargo has emerged as potential way to revolutionize immunotherapy. These types of therapies are used clinically to treat solid tumors, with 17 clinical studies completed as of 2020.³⁶ As suggested in previous sections, combining this approach with materials could enable more effective manipulation of metabolic pathways in immune cells to drive stronger therapeutic outcomes. However, one challenge associated with metabolic cargo is that these are often small compounds that require a high effective concentration to drive immunomodulatory effects. Materials offer the potential to overcome this by promoting higher concentrations of metabolites in cells of interest while decreasing off-target effects. One metabolite our lab has used in this context is Phenyl-7-(hydroxyimino)cyclopropa[b]chromen-1a-carboxamide (PHCCC), a small molecule glutamate receptor enhancer and DC metabolite. Both liposomes and nanoparticles encapsulating this hydrophobic metabolite improved DC viability compared to soluble delivery (Fig 4A), indicating that the particle form could potentially be safer than the soluble form. They also more effectively altered DC function and blunted secretion of inflammatory cytokines (Fig 4B), reducing inflammatory, antigen-specific T cell proliferation (Fig 4C).^37,38 Delivering a metabolite alone promoted changes in the function of a key cell population, highlighting the potential potency of these approaches. As alluded to, metabolic changes also underpin the differentiation of T_REG cell populations from naive populations. To take advantage of this, our lab has co-loaded T1D autoantigen with various metabolic modulators demonstrated to promote T_REG.³⁹ We demonstrated in vitro that these compounds could promote an increase in T_REG while decreasing DC activation levels among disease-relevant cell populations. In similar work, we showed that multiple analogs of the modulator Rapamycin were able to promote T_REG and decrease activation levels.⁴⁰ Taken together, this work shows how materials can be used to harness the modulatory power of immunometabolites.

Fig 4: Biomaterials enable reprogramming of the metabolic- and skin- immune niches.

(A) Metabolites can alter DC function, resulting in differential polarization of T cells. CD11c⁺ cells were isolated from spleens, stimulated with LPS (1 μg/mL), then PHCCC (at 400, 200, 100, and 50 μM concentrations) was encapsulated in liposomes and added to cells. (B) Liposome-encapsulated PHCCC significantly improved viability compared to soluble PHCCC. (C) Liposomal PHCCC significantly blunted inflammatory IL-10 secretion and (D) reduced MOG-specific cell proliferation, as measured by CFSE staining using flow cytometry. (E) Mice were treated with vaccines comprised of SIINFEKL antigen and TLR9a through a variety of delivery routes using a prime-boost regimen. Intradermal delivery resulted in significantly more antigen-specific cells compared to any other delivery route. (F) Immune cells are concentrated in the dermal layers. MNA delivery to these layers targets these cell populations. (G) Dissolvable microneedle arrays can be fabricated from gelatin and loaded with a TLR9a and P47 malaria antigen. Scale bar = 100μm. These needles effectively maintained (H) the immunogenicity of both the P47 antigen and (I) the ability of TLR9a to stimulate TLR9. (J) iPEM- coated PLLA- polymeric microneedle arrays were effectively able to penetrate through the outer layer of dermis, (K) delivering colocalized antigen and adjuvant to the dermal layers. Scale bar = 200μm. In all images, arrows indicate insertion point of a single needle. All data were collected in triplicate and are representative of three comparable experiments. Error bars represent the mean standard error. SEM and p values were considered significant as defined by: *p<0.05; **p<0.01; ****p<0.0001, or as indicated in the figure.

Our lab has also selected delivery routes and locations to target high concentrations of immune cells. The skin provides one such niche, as it is the first line of defense against many types of pathogens; for example, four-channel whole-mount immunofluorescence staining of human dermis revealed that the total area of dermis in a human contained about ten times as many myeloid DCs than the entire blood volume.^41,42 Specifically, the skin contains high levels of two specialized APCs: dendritic cells, which have been defined previously, and Langerhans cells (LCs). LCs are skin-resident APCs located in the epidermis, the uppermost layer of live skin cells. They specialize in quickly taking up antigen and trafficking to LNs.⁴³ In contrast, DCs specialize in presenting antigen to effector cells and are located within the dermis, slightly deeper in the skin.⁴⁴

Because of these benefits, delivery to the dermis is an attractive way to directly activate and study APCs because these cells are so much more highly concentrated in the skin compared to the blood stream (Ackun Frammer, 2022). In fact, the FDA approved a change in the route of administration of monkeypox vaccine in August 2022. Intradermal delivery, compared to subcutaneous (S.C.) delivery achieved the same immune response as subcutaneous delivery with only a fifth of the dose, enabling stretched supply⁴⁵. In a prime boost study conducted by our lab (Fig 4D), intradermal delivery of vaccines comprised of a model peptide, SIINFEKL, with a TLR9a, CpG, yielded the greatest proliferation of CD8+ effector T cells compared to S.C. or intramuscular (I.M.) delivery (Fig 4E). ⁴⁶.

Despite these benefits, intradermal delivery of soluble cues still presents certain challenges. First, antigens and agonists are not necessarily colocalized, which can lead to broad inflammation (if agonists only are taken up) or a tolerizing response (if antigens only are taken up). Next, a bolus injection into the skin can lead to relatively quick clearance of components, reducing the opportunity for APCs to encounter vaccine components and take them up. Finally, the depth of I.D. injection varies physician-to-physician which means that the targeted type of APC (i.e. DC vs LC) also varies. To overcome these goals, microneedle arrays (MNAs) can be used (Fig 4F). These polymeric patches contain hundreds of micron-scale projections long enough to reach immune cells, but short enough to avoid pain receptors. Their length can be carefully chosen to directly target the cell type of interest. The small-volume needles promote retention in the skin and enable co-encapsulation of components in or on the needles. Finally, MNAs can be self-applied and their fixed length lessens variability between injection depths.

Despite these benefits, the material, type, and cargo of MNAs must be carefully selected to effectively penetrate the skin and understand the inherent immunogenicity of materials and cargos.⁴⁷ Our lab has used MNA technology for both infectious diseases and cancer.⁴⁸ In the former case, dissolvable gelatin MNAs were loaded with P47 malaria antigen and TLR9a (Fig 4G). Incorporating antigen into these MNAs did not affect binding of P47 to anti-P47 antibodies (Fig 4H) or the ability of TLR9a to bind TLR9 (Fig 4I) ⁴⁹. To create a candidate cancer vaccine, poly(L-lactide) MNAs were coated with iPEMs comprised of conserved human melanoma antigens with a TLR9a. These MNAs effectively penetrated the epidermis of mouse skin (Fig 4J), delivering colocalized components (Fig 4K). These resulted in T cell expansion both in vitro and in vivo, providing motivation to see whether this response protects against tumors.⁵⁰ With these exciting results, future work will need to be conducted to elucidate the role of LCs compared to DCs in driving immune outcomes. Additional work will also focus on other types of diseases, such as autoimmune and allergic diseases.

Controlling Immunity via the LN Microenvironment

LNs are critical immunological tissues for the organization execution of adaptive immune responses—both in healthy immune responses, such as to a vaccine, or in pathologic responses in autoimmunity. In these tissues, antigen-presenting cells (APCs) such as dendritic cells (DCs) present antigen to antigen-specific B cells and T cells along with activating cues to drive expansion. Whether through passive drainage or cell-mediated transport, immunological signals must reach LNs to be “seen” by the immune system. To facilitate their organization role, LNs are highly organized structures and adapt their structure in response to various conditions. Of particular interest are laminin stromal proteins. It has been demonstrated that the ratio of laminin α4 to laminin α5 is indicative of whether the LN will promote an inflammatory or tolerizing response.^51–53 Due to the important role of LNs as domains of immunological organization, directly re-engineering their microenvironment is a potential way to increase control over immune responses.

In efforts to target this critical tissue, there have been clinical attempts to use ultrasound-guided injections of soluble antigens to increase potency of immune responses.^54–56 However, since these cues are soluble, they readily drain out of LNs following injection. Building on this concept, our lab developed a biomaterial platform to reprogram the LN microenvironment using direct injection of microparticle depots^57,58(Fig 5A). These depots are diffusion limited; instead, they remain in LNs to slowly release immune cargo over time—offering great control over this critical immune tissue (Fig 5B). In models of vaccination, microparticles loaded with antigen and adjuvant were able to greatly increase the proportion of antigen-specific CD8+ T cells, as measured by tetramer. This increase required both direct LN injection as well as microparticle delivery compared to soluble cues. A follow up study demonstrated that this increase was maintained at least 28 days compared to a sham and was accompanied by increased lymphocyte numbers in treated LNs (Fig 5C).⁵⁹ In a model of cancer, signals split between contralateral LNs remained separate as shown via in vivo imaging (Fig 5D–E). Delivery of both adjuvant and antigen in the same LN was shown to provide stronger protection against tumors than splitting the signals between distinct LNs.⁶⁰ The requirement for both signals to be localized to the same geographic location suggests that the two signals work together within the LN to promote expansion of tumorkilling T cells (Fig 5F). In additional studies, we used the modularity of the MP platform to incorporate additional drugs to further condition phenotype, such as low-dose mTOR inhibitors to drive central memory T cells.⁶¹

Fig 5: Direct LN immunotherapy promotes increased potency and modular protection against disease.

A) Schematic of intra-LN injection. Briefly, mice are injected with tracer dye at the tail base to visualize LNs. Particle suspensions are then directly injected into visibly LNs. B) Histology of particle-treated LN. Particles are depicted in green, B cells in blue, and T cells in red. C) Antigen specific CD8 T cell populations as a proportion of total CD8 cells measured using tetramer. Mice were treated with no treatment (naïve), sham, empty MP, or PolyIC MP with OVA. D) Schematic showing treatment regimens for one, two, and split LN treatments. E) IVIS imaging of contralateral LNs after treatment with either one LN (Cy5 PolyIC and FITC OVA) or split (Cy5 PolyIC Left, FITC OVA right). F) Cancer survival in mice treated intra-LN with formulations described in legend. Split formulations deliver antigen and adjuvant to contralateral LNs. One LN indicates full treatment formulation in one LN, while Two LN indicates full treatment formulation to both LNs. G) Histological analysis of particle treated LN depicting MPs (red) loaded with FITC-MOG (green) with B cells (blue) and T cells (white). H) Clinical score of mice treated at peak disease with MOG/Rapa MPs or a control intra-LN. I) Diabetes-free survival of mice treated iLN with Empty, Rapa, antigen (p31 or NRP-V7), or antigen-rapa MPs. J) Allograft survival of transplanted pancreatic islets of mice treated with empty, rapa, alloantigen (Ea), or alloantigen-rapa MPs.

Engineering the local LN microenvironment—where T cell fate is often determined, creates an opportunity to modify the course of other disease classes, such as autoimmunity. While effective at increasing vaccine potency and protective in models of cancer, we have in parallel adapted this platform to drive immune tolerance in the context of autoimmunity. In this approach, instead of adjuvant, autoantigen is co-loaded with an immunomodulator such as rapamycin (Rapa)—an mTOR inhibitor that potentially induces and expands T_REG. The rationale for this approach is that maintaining a concentration of both autoantigen and Rapa in the same context, autoantigen-specific T_REG will be generated with suppressive functions toward that antigen; thus, promoting tolerance without broad immunosuppression. We first tested this in a model of the autoimmune disease multiple sclerosis, wherein the immune system mistakenly attacks the myelin of the nerves. MPs delivered antigenic cargo to LNs without disrupting T and B cell zones as shown by histology (Fig 5G). By treating intra-LN with antigen-Rapa microparticles, just a single treatment was able to durably reverse paralysis—even when administered at the peak of disease (Fig 5H).⁶² This protection was associated with a significant increase in T_REG. Further, when an irrelevant antigen was substituted for the autoantigen of interest, this efficacy was lost, highlighting the antigen-specific nature of this approach. In a recent report, we applied this strategy to effectively promote tolerance in T1D and improve allograft transplantation survival, showing that this approach is potentially highly modular.⁶³ In an accelerated model of T1D, iLN MPs containing antigen (p31 or NRP-V7) and rapamycin protected mice from disease onset (Fig 5I). Additionally, in an mismatched transplant model, MPs loaded with alloantigen (Ea) and rapamycin significantly prolonged allograft functionality (Fig 5J). Further, we showed that intra-LN MPs particles potentially drive the differentiation of a memory-like cell population that could be critical for long term protection in the context of disease.

Conclusion

In summary, biomaterials delivered via a variety of routes provide an opportunity to overcome many of the challenges facing clinical translation of vaccines and immunotherapies. These hurdles include quick clearance of vaccine and immunotherapeutic components, a lack of antigen and adjuvant colocalized delivery, and drugs or metabolites that lack the proper biophysical features for cellular uptake. Encapsulation into a variety of biomaterials allows these challenges to be overcome, enhancing both the innate and effector immune response in vitro and in vivo to a variety of diseases, including cancer, a mouse model of MS, and a model antigen. Biomaterials themselves, such as QDs and liposomal nanoparticles, can be engineered to exhibit the desired immunological properties. Our lab also seeks to use immune signals themselves as biomaterials by harnessing their inherent biophysical properties, either with or without exogenous materials. We have developed the iPEM platform, which are coated, templated particles comprised entirely of immune signals held together through opposing charge. We have also designed polyplexes, which are nanoparticles that rely on those same electrostatic interaction to combine immune cues.

Finally, we seek to overcome the challenge of quick systemic clearance of immune signals by directly delivering cues encapsulated by biomaterials to engineer immune-cell rich niches; specifically, the metabolism of DCs by metabolite delivery in nanoparticles, the dermal layers via MNAs and LNs via intra-LN injection. Cues delivered via MNA enable expansion of antigen-specific immune cells, while iLN protects mice against T1D and MS at least in part by retention of immune signals in immune-cell rich niches. This exciting work provides numerous future directions. First, the agonists we have currently tested in mice can be optimized for use in humans by identifying specific features of innate immune cells unique to humans. This will provide an exciting opportunity to prove how many of these platform technologies can be used clinically. Next, combining different materials platforms with the cues and delivery routes used throughout the lab will provide unique, combinatorial therapies and potentially yield synergistic responses—for example, delivering iPEMs via the iLN route or delivering nanoparticles via MNAs.