Biomaterials to Enhance Antigen-Specific T cell Expansion for Cancer Immunotherapy

Abstract

T cells are often referred to as the ‘guided missiles’ of our immune system because of their capacity to traffic to and accumulate at sites of infection or disease, destroy infected or mutated cells with high specificity and sensitivity, initiate systemic immune responses, sterilize infections, and produce long-lasting memory. As a result, they are a common target for a range of cancer immunotherapies. However, the myriad of challenges of expanding large numbers of T cells specific to each patient’s unique tumor antigens has led researchers to develop alternative, more scalable approaches. Biomaterial platforms for expansion of antigen-specific T cells offer a path forward towards broadscale translation of personalized immunotherapies by providing “off-the-shelf”, yet modular approaches to customize the phenotype, function, and specificity of T cell responses. In this review, we discuss design considerations and progress made in the development of ex vivo and in vivo technologies for activating antigen-specific T cells, including artificial antigen presenting cells, T cell stimulating scaffolds, biomaterials-based vaccines, and artificial lymphoid organs. Ultimate translation of these platforms as a part of cancer immunotherapy regimens hinges on an in-depth understanding of T cell biology and cell-material interactions.

1. Introduction

Immunotherapy relies on the manipulation of the immune system to induce a potent and durable antigen-specific attack on diseased cells. Most immunotherapies to date have specifically relied on the work of effector T cells, as they accomplish many of the goals set by both personalized medicine[1] and targeted drug delivery[2]. T cells accumulate in the diseased site[3], kill diseased cells with high specificity and efficiency, and potentiate not only responses from other T cells, but from other branches of the immune system as well. In addition, while coordinated immune responses involving both the humoral and cellular arms of the adaptive immune system are important in the response to most infections, diseases such as cancer[4] and the novel SARS-CoV-2[5] specifically require strong T cell responses for lasting immunity. As a result, much work has gone into the development of therapies to replace, induce, or potentiate T cell responses in the treatment of cancer, infectious diseases, and autoimmunity.

Our native T cell repertoire provides safe and efficient protection from a range of infections and malignancies. Even before individuals have gone through puberty, they have approximately 4x10¹¹ T cells in circulation[6]. Each of these cells has survived positive selection, ensuring the creation of billions of distinct and functional T cell receptors (TCRs), and negative selection, which deletes any self-reactive clones. This process results in a repertoire with enough breadth to protect against future unknown pathogens while avoiding overactivity or autoimmunity. However, this system can become dysfunctional for a variety of reasons – from genetics to disease – rendering otherwise potent T cells ineffective. For instance, natural T cell responses can be suppressed in cancer through a wide range of immunoevasive tactics[7].

Strategies to replace or supplement T cell responses have been in development for decades. In this review, we will discuss the importance of biomaterials for bringing existing cellular therapies to the forefront of cancer immunotherapy, with a focus on recent advances in technologies for ex vivo and in vivo antigen-specific T cell activation and expansion. These technologies are particularly relevant for endogenous T cell therapy (ETC), an approach in which rare, naturally present, tumor-specific T cells are expanded to therapeutic levels from the peripheral blood mononuclear cells (PBMC) of cancer patients[8]. ETC is poised to provide a path toward personalized immunotherapies but comes with many challenges; biomaterials have the potential to alleviate many of these challenges, and in turn, facilitate widescale adoption of ETC. Alternative cellular therapies are also at various stages of clinical use and development. Analogously, these approaches can be further augmented through use of biomaterial platforms for control over the phenotype, function, dosing, and timing of therapeutic T cell administration. The development of biomaterial platforms for these therapies requires an in-depth understanding of T cell biology and careful consideration of design parameters that can allow for rapid expansion and fine control of T cell phenotype and function.

2. Endogenous T Cell Therapy: Opportunities, Challenges, and Alternatives

In the natural immune response, disease-specific antigen is taken up by antigen presenting cells (APCs), such as dendritic cells (DCs), at the site of infection or disease. When these antigens are internalized along with pathogen-associated or danger-associated molecular patterns (PAMPs and DAMPs respectively), the DCs are activated and travel to secondary lymphoid organs (SLOs) such as lymph nodes (LNs), in which naïve lymphocytes are concentrated. In the SLO, DCs travel to T cell rich zones and present three essential signals to naïve T cells. The first signal, signal 1, confers specificity in the form of the peptide-MHC (pMHC) complex which binds to the T cell receptor (TCR). The second signal, signal 2, is costimulatory and works to amplify downstream events from the TCR. The third signal, signal 3, is soluble cytokine support, secreted from DCs or other nearby cells to support cell growth and differentiation. Once activated, T cells leave the SLO and travel through systemic circulation until they encounter signals presented at the diseased site. T cells are then cued to extravasate into the tissue where they begin searching for cells expressing the cognate antigen.

The types of T cells that are targeted and their functions, are dependent on the signals they encounter during activation. For signal 1, peptide-loaded major histocompatibility complex I (pMHC I) molecules activate cytotoxic (CD8+) T cells, whose function is primarily to lyse cells infected with intracellular pathogens, whereas pMHC II molecules activate helper (CD4+) T cells, whose function is further specialized based on the cytokines they encounter upon activation. Alternatively, CD3 engagement is an approach used commonly in the clinic to bypass antigen-specific TCR signaling and activate polyclonal populations of CD4+ and CD8+ T cells[9, 10]. T cell memory is also determined early on after activation through several potential mechanisms[11]. The resulting memory state of the T cells determine whether they die off several weeks after initial activation or if they persist in the body for months to years.

Endogenous T cell therapy (ETC) seeks to mimic natural antigen-specific T cell responses ex vivo through the enrichment and expansion of rare, circulating tumor-reactive T cells from patients’ peripheral blood, followed by reinfusion of large numbers of autologous tumor-specific T cells into cancer patients (Fig. 1A). Enrichment of tumor-specific T cells can be performed in a number of ways, most commonly through fluorescent-activated cell-sorting (FACS) based on T cell binding to cognate, fluorescently-labelled, and multimerized pMHC molecules known as tetramers. Other common approaches include sorting on T cell activation or inhibitory markers, such as PD-1[12, 13] or CD137[14], or immunomagnetic bead-based enrichment[15]. Expansion of tumor-specific T cells is most commonly performed using autologous APCs that are either pulsed with tumor-specific peptides or transfected with RNA encoding tumor antigens[15].

Figure 1.

Protocols of T cell production for (A) ETC, (B) TIL, and (C) TCR/CAR T therapies. (A) For ETC, T cells and APCs are isolated from patient-sourced PBMCs. APCs are engineered to express the antigen of interest, then incubated with T cells to expand tumor-specific T cells. (B) TILs may be isolated from an excised tumor and rapidly expanded with IL-2. Tumor-reactive T cells from this pool are then selected for longer term expansion. (C) T cells for TCR and CAR T therapies are derived from patient-sourced PBMCs. T cells are transduced with genes for a tumor specific TCR or CAR, then expanded to large numbers. Created with Biorender.com.

2.1. Opportunities

Endogenous T cell therapy has much potential as a cancer immunotherapy, as evidence by its clinical use for almost two decades (see ref[8] for a summary of ETC clinical studies). ETC has minimal requirements of clinical grade peptide or RNA and patient PBMCs to produce a cell product, and it can be easily tailored to specific antigens simply by altering which epitope(s) are pulsed onto autologous APCs. Since only autologous cells are used, this approach presents few regulatory hurdles or complex pipelines, allowing for rapid, ad hoc targeting of patient-specific tumor antigens[8] or even neoantigens[16, 17], novel epitopes that result from tumor-specific mutations[18]. The benefit of modularity in this approach is further emphasized by studies which have shown optimal antitumor responses may require simultaneous targeting of multiple tumor antigens[19,20]. Additionally, by targeting naïve T cells, ETC inherently provides flexibility over the memory phenotype of the final T cell product. This is particularly important, as there is significant evidence from mouse models and human clinical trials that less differentiated naïve, stem cell memory, or central memory T cells show significantly greater in vivo persistence and antitumor efficacy compared to more differentiated effector memory or terminally differentiated effector cells[21,22]. The resulting T cells tend to be relatively safe, as these endogenous cells have gone through negative selection and therefore are unlikely to cross-react with healthy cells. Patient preconditioning steps for ETC also tend to be relatively safe and can range from no[23,24] to mild lymphodepletion[25] and no[24] to low doses of IL-2[23,25], both common sources of toxicities for other cellular therapies (see section 2.3.1)[26]. Lastly, recent studies have shown that receptors inserted into the native TCR locus generate more potent immune responses[27], suggesting that endogenous T cells which inherently signal through their native TCR may generate more effective immune responses compared to most engineered cellular therapies.

2.2. Challenges

Replicating or inducing natural T cell immune responses for cancer immunotherapy has been no small feat. Despite significant advancements in T cell culture systems, several challenges remain that prevent ETC from becoming a first-line therapy.

The identification of appropriate tumor antigens is a critical yet challenging process that must be completed iteratively as each patient has a unique human leukocyte antigen (HLA) phenotype as well as a unique tumor mutanome. Tumor associated antigens (TAAs) or neoantigens may be identified from a biopsy or resected portion of the tumor which is then digested and purified for sequencing of mutations, overexpressed antigens, and HLA expression [28]. For neoantigens, once potential epitopes are uncovered, prediction software is used to determine if the mutated epitopes are capable of being processed and presented on expressed HLA molecules[29]. Tumor specificity and immunogenicity of the resulting pool of targets can then be determined by monitoring T cell responses to peptide or cDNA libraries through a variety of methods including cytokine secretion, activation or inhibitory marker upregulation, or peptide-HLA multimer staining [29] to choose the final epitope(s) to target.

Even when an immunogenic target epitope is found, T cells against that antigen can be very rare in the naïve repertoire, especially for cancer antigens, as the numbers of T cells that recognize overexpressed self-antigen or mutated-antigen are naturally very low[30]. These extremely rare cells must be expanded to large numbers; infused cell products for ACT can be up to 10¹¹ cells[16,31] to confer clinical benefit. The 4-5 weeks long culture that is required to reach these cell numbers[8,32] can often lead to T cell exhaustion due to overstimulation[33].

T cell therapy is traditionally an autologous cell therapy, meaning that patient to patient variability not only affects the response to the treatment but also the production of the treatment. Variability in the precursor frequencies and phenotype of patient T cells, as well as the availability and function of patient antigen presenting cells, can have a significant impact on T cell expansion and resulting function. Patient age, disease status, co-infections, and drug usage all affect the outcome of the T cell product.

The autologous nature of ETC also has impacts on manufacturing, as personalized cellular products face challenges such as lack of quality control markers and high cost. The length of time in manufacturing facilities, resources required for expansion, transport of cells between facilities and hospitals, as well as any necessary sequencing of patient samples to determine the proper T cell treatment all contribute to the cost of ETC. Prices may fall as technology improves, but currently the cost excludes many patients from benefiting ETC, and even for patients who can afford it, it is still not offered as a first- or second-line treatment.

2.3. Alternatives

Several additional approaches to ACT have been developed including polyclonal expansion of Tumor Infiltrating Lymphocytes (TIL) (Fig. 1B), and genetic engineering host cells to express antitumor TCRs or CARs (Fig. 1C).

2.3.1. TIL Therapy

TIL therapy, one of the earliest forms of ACT, involves isolating lymphocytes that have infiltrated the stroma of patient tumors and expanding them on irradiated allogeneic feeder layers for 5-6 weeks in the presence of an αCD3 antibody and IL-2[31]. TIL provide a source of T cells that are naturally enriched for tumor-reactivity, in comparison to PBMC[31]. The approach has shown tremendous promise as a therapy for metastatic melanoma[34] and has since been refined to include chemotherapy-based lymphodepletion immediately prior to TIL transfer[35] to improve the persistence of adoptively transferred cells, resulting in objective response rates and complete tumor regression in up to 72% and 40% of patients, respectively[36]. Despite these impressive clinical results, TIL therapy suffers from several drawbacks. Tumor-reactive TIL have only been successfully expanded from melanoma tumors[31] limiting the applicability of this therapy to other forms of cancer. Even for melanoma, there can be wide variations in the success of expanding patient TIL[37]. The necessary preconditioning steps, including lymphodepletion and high doses of IL-2 post adoptive transfer, are associated with harsh toxicities and thus require close monitoring of patients[26]. Furthermore, the process of activating TIL, which traditionally requires 5-6 week ex vivo cultures with irradiated feeder cells and complex rounds of selection for cultures that show antitumor activity[31], is not easily amenable to large-scale translation. TIL also often exhibit a terminally differentiated, exhausted phenotype with impaired effector function, limiting their therapeutic efficacy[38–41].

2.3.2. TCR and CAR-T Immunotherapies

TCR and CAR T cell immunotherapies have been developed to widen the range of targetable cancers and produce more “off-the-shelf” alternatives to ACT. In these therapies, a patient’s endogenous T cells are genetically reprogrammed to recognize tumor antigens either through an engineered alpha-beta TCR or through a CAR, which contains an extracellular antigen-binding domain that resembles the variable region of an antibody and an intracellular signaling domain for T cell activation and co-stimulation[42]. The transduced cells are then isolated and expanded for several weeks before being reinfused into the patient. A single TCR or CAR construct can be used to treat many patients with the cognate tumor antigen. In fact, CAR T cell therapies bypass HLA restrictions, so they can treat any tumors with the target antigen and maintain therapeutic efficacy even for tumors that develop defects in antigen processing and presentation[43]. However, CAR T cell therapies are limited to surface expressed antigens[18], while TCR therapies can target intracellular antigens presented on HLA molecules.

While the scalability of these technologies has provided tremendous translational potential, these therapies are associated with several life-threatening complications. First, CAR T cell therapies, in particular, can lead to excessive release of cytokines that require careful medical management in specialized facilities[44]. Secondly, numerous clinical trials have resulted in severe on-target off-tumor toxicities because these high affinity, exogenous receptors have not gone through negative selection[45,46] and are thus more prone than endogenous T cells to attack healthy cells that express low levels of antigen or cross-reactive epitopes[47–50]. Clinical success at minimizing toxicities by targeting tumor antigens that are only otherwise present on nonessential healthy cells have been seen for a variety of hematological malignancies[51–53]; however, an analogous approach has proven elusive for solid tumors, as many tumor-specific antigens are either expressed at levels that are too low on tumor cells or are also present on healthy cells[31]. Neoantigens are promising targets for TCR immunotherapy as they are only present on tumor cells; however, personalized neoantigen-specific TCR therapies are currently too expensive and cumbersome for widespread clinical use[54]. To further complicate matters, patients may require TCRs or CARs specific to multiple tumor antigens for optimal tumor control, as single-antigen targeting can result in escape of antigen-loss variants, especially for solid tumors[55–58]. Several recent publications have identified pan-cancer TCRs[59,60] and CAR T targets[61] which may prove to be universal cancer immunotherapies for solid tumors.

Additionally, the high affinity nature of current TCR or CAR T immunotherapies raise concerns of chronic antigen stimulation that can lead to T cell exhaustion and hyporesponsiveness[4,62]. There is ongoing work to reduce exhaustion and improve persistence of CAR T cell therapies either by including additional costimulatory domains[63,64], knocking out inhibitory molecules[65] and transcription factors[62,66], or even using lower affinity CARs[67].

2.3.3. Biomaterials for Antigen-specific T cell Expansion

As engineered cellular therapies continue to address current limitations of T cell-based therapies, biomaterials that can expand antigen-specific T cells also have an opportunity to play significant roles in the advancement of cancer immunotherapy (Table 1). First, they could present a path forward for ETC therapies, by providing a scalable means of identifying and targeting a patient’s unique tumor-specific T cells with “off-the-shelf” platforms. One of the main barriers to personalized immunotherapies is that current clinical approaches to identification and expansion of endogenous tumor-specific T cells rely on autologous antigen presenting cells, such as monocyte derived dendritic cells (moDCs)[16,24,71]. While T cells expanded by moDCs have shown clinical efficacy[24], DC-based T cell expansion is limited by availability[72], potential dysfunction[73,74], and complex manufacturing of autologous DCs. In contrast, “off-the-shelf” and relatively inexpensive biomaterials can stably and reproducibly present all the necessary cues for T cell activation and co-stimulation. These platforms can also be modularly designed to accommodate a range of tumor antigens across multiple HLA types, providing a facile manner of identifying and targeting patient-specific tumor antigens or neoantigens[75].

Table 1.

Summary of advantages and limitations of existing immunotherapies and how biomaterials can help overcome those limitations.

Cellular Therapy	Advantages	Limitations	Examples of How Biomaterials Can Help
ETC	Ease of personalization Minimal regulatory hurdles Can target neoantigens Control over memory phenotype Safe Mild Preconditioning Native TCR signaling	HLA restricted Complexity of identifying antigens Rarity of cells Difficult to expand to therapeutic levels Long culture period Patient Variability Manufacturing Challenges Cost of Treatment	Adaptive aAPCs allow for simple screening of tumor antigen immunogenicity[75] aAPCs[79,80] and APC-ms[81] can rapidly expand highly functional T cells from very low precursor frequencies aAPCs can eliminate variability of moDC function through stable antigen-presentation on particles or scaffolds[79] Easy and inexpensive GMP production of synthetic platforms[82]
TIL	Personalized to a patient’s tumor antigens or neoantigens Enrichment of tumor-reactive T cells Streamlined screening for tumor immunogenicity Native TCR signaling	HLA restricted Limited, mostly to melanoma Patient variability Harsh preconditioning steps Long culture period Exhausted phenotype and reduced effector function Cost of treatment	aAPCs[79,80] and APC-ms[81] can allow for a more rapid culture period Local delivery and sustained release of T cells from scaffolds can reduce length of culture period by requiring fewer cells for therapeutic efficacy[83,84] aAPCs can be designed to reverse T cell exhaustion[85] Easy and inexpensive GMP production of synthetic platforms[82]
TCR	Scalable Short culture period Can recognize intracellular antigens, such as neoantigens High affinity response Control over memory phenotype	HLA restricted Difficult to personalize Overstimulation can lead to exhaustion On-target, off-tumor toxicities Cost of Treatment	Targeted in vivo vaccination can maintain TCR-transduced cells within the therapeutic window to reduce toxicities[54,61] Conjugation of particles with soluble factors to cells can allow controlled delivery of immunomodulatory factors to the tumor[86]
CAR T	Scalable Short culture period Not HLA restricted High affinity response Control over memory phenotype	Restricted to surface-expressed antigens Difficult to personalize Overstimulation can lead to exhaustion Cytokine-related toxicities On-target, off-tumor toxicities Cost of Treatment	Targeted in vivo vaccination can maintain CAR-transduced cells within the therapeutic window to reduce toxicities[54,61] Conjugation of particles with soluble factors to cells can allow controlled delivery of immunomodulatory factors to the tumor[86]

aAPCs, artificial antigen presenting cells; APC-ms, antigen presenting cell mimetic scaffold; CAR, chimeric antigen receptor; ETC, endogenous T cell; GMP, good manufacturing practices; moDC, monocyte derived dendritic cell; TCR, T cell receptor; TIL, tumor infiltrating lymphocyte.

Second, biomaterial-based platforms can augment existing cancer immunotherapies by enhancing the persistence, phenotype, and function of transferred cells. In contrast to moDCs, which provide limited control over the signals T cells encounter during activation, biomaterial shape, size, stiffness, porosity, and biodegradability as well as the organization, dose, and composition of ligands and soluble factors can each have profound effects on T cell expansion, function, and phenotype[76], and be precisely tuned for optimal T cell stimulation. This in turn can allow for greater consistency in final ETC, TCR, or CAR T cell products. The flexibility and control offered by biomaterials is especially important, as the field of cancer immunotherapy continues to discover more about the most ideal targets and T cell phenotypes for achieving therapeutic responses. By example, while it has been known that cytokine secretion alone does not predict cytotoxic T cell killing ability and by extension potential clinical efficacy[77], the precise mechanisms of cytotoxic T cell killing through thrombospondin-1 dependent supramolecular attack particles were only discovered recently, motivating future biomaterial approaches that target this pathway to specifically enhance cytotoxicity[78]. Furthermore, biomaterial platforms for T cell expansion can allow for control of the dosing, timing, and localization of ETC, CAR, or TCR therapies in vivo to maintain antitumor immune responses and maximize their efficacy, while minimizing the risks of off-target toxicities, T cell exhaustion, or immune escape. The dynamics and localization of T cell therapies can be controlled in vivo by targeting specific cells such as T cells or dendritic cells, specific sites such as lymph nodes or the tumor, or specific external stimuli to control the timing of immune responses. Finally, biomaterials that can sufficiently activate and direct T cell immune responses in vivo could drastically streamline production processes for antigen-specific T cell therapies. In the next two sections, we will discuss the progress that has been made as well as the specific design parameter considerations for biomaterial platforms used for ex vivo and in vivo antigen-specific T cell activation and modulation. Tables of representative biomaterials platforms for antigen-specific T cell expansion can be found in references [68–70].

3. Ex Vivo Biomaterial Platforms for Antigen-Specific T cell Activation

Much of the progress in understanding critical design parameters for T cell activation in the last 40 years has come from biomaterial-based technologies for ex vivo T cell expansion. In contrast to cell-based approaches for T cell expansion, these technologies provide reductionist systems to directly and independently observe how material properties, dosing, and choice of ligands and soluble factors each modulate T cell proliferation, function, and phenotype. Ex vivo technologies for T cell stimulation are in some ways simpler and, in others, more complex to design than in vivo technologies. On the one hand, ex vivo platforms do not require tuning of physical or biochemical properties to meet biocompatibility requirements, to enhance in vivo biodistribution, or to allow for organ or cell-specific targeting; on the other hand, these technologies generally aim to replace and not simply augment or modify the function of endogenous APCs and thus, may require optimization of a large number of design parameters to show similar if not improved efficacy compared to endogenous APCs.

The most fundamental requirements for antigen-specific T cell stimulation by professional APCs include recognition through interaction between a TCR and its cognate pMHC (signal 1), co-stimulation most fundamentally through CD28 on T cells and B7.1/2 (signal 2) on antigen-presenting cells, and soluble factors known as cytokines (signal 3), which direct T cell fate and lineage[82]. In addition to the composition of these signals, the mechanical forces they transmit, their density and dose, and their spatial organization all affect T cell activation. Initial T cell activation does not only require binding of TCR to its cognate pMHC, but also the specific mechanotransduction that occurs due to this interaction[87–89]; this means that even high affinity TCR-pMHC binding events can occur without leading to T cell activation[89]. Upon initial activation, TCRs which are pre-clustered on naïve T cells into 35-70 nm nanoislands[90] begin to coalesce into microscale clusters and eventually form the central portion of an immunological synapse with an APC[91]. The cytoskeletal dynamics associated with formation of the immunological synapse exert mechanical forces at the T cell-APC interface[91]. Moreover, immobilization of intracellular cell adhesion molecule 1 (ICAM-1) on APCs during the formation of the immunological synapse is necessary for mechanically activating the integrin lymphocyte function associated antigen-1 (LFA-1) on T cells[92]. The cytoskeletal reorganization and mechanical forces that occur at the T cell-APC interface have implications on the arrangement and density of signals as well as the mechanical properties of biomaterials that can lead to robust T cell stimulation. Thus, a wide range of APC-mimetic biomaterials have been developed to recapitulate these fundamental properties for efficient ex vivo T cell stimulation.

Biomaterials for antigen-specific T cell stimulation have been studied using particle and scaffold-based platforms (Fig. 2A,B). Both modalities offer flexibility of ligand choice and density, with particles more closely mimicking endogenous APCs and allowing for control over the curvature of interaction with T cells, and scaffolds enabling easier and more precise ways to pattern signaling molecules and tune biophysical properties such as stiffness and porosity. As dendritic cell (DC) based T cell stimulation requires at the bare minimum signal 1 and 2 within a specific cytokine milieu, APC-mimetic particles and scaffolds minimally present pMHC and either B7.1/2 or an agonistic αCD28 monoclonal antibody[82]. Beyond those minimal requirements, each modality has its unique design considerations for antigen-specific T cell stimulation.

Figure 2.

Overview of design considerations for T cell stimulating platforms. Stimulating signals may be delivered either via particles (A) or scaffolds (B). Each modality has unique interactions with T cells and with the body that influence design parameters. With both modalities, material choice (C), size and shape (D), and choice of ligands (E) all have significant effects on T cell activation, both ex vivo and in vivo. PK, pharmacokinetics. Created with BioRender.com.

3.1. Particles

Particles used for antigen-specific T cell stimulation are often referred to as artificial antigen presenting cells (aAPCs). Beyond the minimal requirements of presenting signals 1 and 2, the major design considerations for aAPCs include material, size and shape, and ligand choice (Fig. 2C–E).

3.1.1. Materials

The materials which have been used for aAPCs have ranged from biomimetic or biological such as liposomes or cell membranes, to inorganic such as iron oxide or carbon nanotubes, to polymeric such as polystyrene or poly(lactic-co-glycolic acid) (PLGA), or some combination thereof (Fig. 2C). Material choice can play an integral role in determining the function of aAPCs, as it can impact particle properties such as membrane fluidity, nanoscale organization of ligands, stability, stiffness, degradability, surface area, ease of encapsulation of soluble factors, and responsiveness to magnetic fields.

Some of the first aAPCs were developed over 40 years ago, using phospholipid membranes reconstituted with cell-derived MHC molecules to induce cytotoxic T cell responses[93,94]. Liposomal aAPCs, like cell membranes, form dynamic lipid bilayers that can provide fluidity in interacting with T cells, in contrast to inorganic or polymeric particles that are rigid and therefore have constrained orientations of T cell signaling molecules bound to their surface[95]. In addition, the lateral mobility of liposomal aAPCs, especially at higher temperatures such as 37°C, can allow for migration of pMHC molecules and costimulatory ligands towards the site of formation of the immunological synapse[96], in a manner akin to endogenous T cell-APC interactions. Lipid formulations can also allow for formation of microdomains containing pre-clustered T cell ligands, mimicking the microdomains of pMHC molecules found on APCs[97]. Several studies have shown that liposomes which contain these regions of pre-clustered T cell ligands more potently stimulate T cells than liposomes with randomly distributed signaling ligands[98,99]. Capitalizing upon these findings, one group directly isolated the lipid rafts of pre-clustered MHC molecules from DCs and reconstituted them onto liposomes to form RAFTsomes[100]. Not only did these RAFTsomes lead to cytokine secretion and antigen-specific T cell proliferation in vitro but they also led to tumor protection in an in vivo immunization model[100].

Despite these benefits, liposomes tend to have lower stability than polymeric or inorganic materials and may have stiffnesses that are too low to provide the necessary mechanical cues for T cell stimulation. This has led some groups to pursue composite materials, known as supported lipid bilayers, that contain a polymeric or inorganic core surrounded by a lipid outer membrane. Some of the earliest work examining the ability of supported lipid bilayers to stimulate cytotoxic T cell responses against tumor antigens involved production of large multivalent immunogens formed through the incorporation of tumor cell plasma membrane vesicles onto silica or latex microspheres[101]. These initial studies found that the composite particles led to improved cytotoxicity against tumor antigens and significant reduction in tumor growth for several syngeneic tumor models, whereas tumor-derived liposomes had no effect on cytotoxicity or tumor growth[101]. Since these reports, several other groups have pursued analogous approaches for generating tumor-specific T cell responses using 100-200 nm supported lipid bilayers composed of a PLGA core and a cell membrane coating derived from unmodified tumor cells[102], tumor cells genetically modified to express the costimulatory molecule B7.1[103], or mature dendritic cells pulsed with tumor lysate[104]. These approaches have shown promising results both in vitro and in vivo and may provide an “off-the-shelf” antigen-agnostic approach of generating personalized tumor-specific immunotherapies.

Other commonly used materials for aAPCs are inorganic, such as iron oxide. Iron oxide microparticles have been used extensively for expansion of antigen-specific CD4+ and CD8+ T cells in both humans[95,105] and mice[106]. Clinically, they are an attractive platform for stimulating T cells for ACT, as the particles can be easily removed using a magnet prior to infusion[82,107]. Nanoscale iron oxide aAPCs can additionally be used to enhance T cell activation by driving aggregation of aAPCs on the surface of T cells with a magnetic field[108,109]. This approach leads to enhanced TCR clustering and much greater antigen-specific CD8+ T cell expansion ex vivo, as well as improved antitumor efficacy in vivo[108]. More recently, nanoscale iron oxide aAPCs have been used to simultaneously enrich and expand rare, endogenous antigen-specific CD8+ T cells[79,80,110]. aAPCs are first incubated with a diverse population of endogenous T cells to allow rare, cognate T cells to bind to aAPCs. Next, bound T cells are enriched using a magnet, depleting non-cognate T cells and simultaneously clustering TCRs on rare cognate cells to enhance their activation[80]. Not only does this process of enrichment and expansion result in a higher frequency of antigen-specific T cells, but it also improves the expansion of cognate cells compared to aAPC-based stimulation without enrichment[79,80]. In turn, this approach allows for identification of putative neoantigens from murine tumor models, as well as greater than 1000-fold expansion of antigen-specific murine and human T cells in one week[80]. Despite this rapid expansion, which can often lead to exhaustion, endogenous tumor-specific T cells expanded with this approach showed impressive antitumor responses in vivo[80]. Enrichment and expansion has also been applied to melanoma patients from various immunotherapy trials and has resulted in up to 1000-fold expansion of Melanoma Antigen Recognized by T cells 1 (MART-1) tumor-specific CD8+ T cells over two weeks[79]. Even after this rapid expansion, these MART-1 CD8+ T cells displayed a predominantly “stem-like” phenotype and were highly functional based on cytokine release and cytotoxicity assays[79].

The magnetic properties of iron oxide nanoparticles have also been used to produce more modular approaches to antigen-specific T cell expansion, allowing for greater control over the phenotype and specificity of the expanded cells. For instance, signal 1 and 2 can be displayed by separate nanoparticles and then co-localized on the surface of T cells using magnetic clustering[85]. This approach can allow for the dosing and composition of signal 1 and a range of signal 2s to be varied independently of each other, allowing for rapid prototyping of an array of aAPC designs to skew toward a specific memory phenotype or to optimize T cell expansion[85] (Fig. 3A). Furthermore, as naïve T cells constitutively express CD28[111], enrichment with signal 1 only particles can reduce nonspecific binding of non-cognate cells, leading to a purer enriched fraction that can subsequently be expanded through co-clustering of signal 1 and 2[85]. The magnetic properties of iron oxide aAPCs have also been utilized to produce high throughput platforms for enrichment and expansion of a range of T cell specificities[75]. In this setting, an adaptive aAPC conjugated with unloaded MHC molecules and αCD28 can be surface loaded with a range of peptides and then washed using a 96-well plate magnet, to parallelize production of aAPCs targeting a variety of T cell specificities (Fig. 3B). T cells can be incubated with batches of these particles in a 96-well plate, enriched, and expanded, to allow for much wider screens of endogenous antigen-specific T cell responses in cancer immunotherapy, autoimmunity, and infectious diseases[75].

Figure 3.

Examples of modular nanoparticle platforms for custom T cell expansion. (A) Magnetic field clustering of nanoparticles allows for separation of signal 1 and 2 to allow a wide range of ratios and combinations of signal 2s to be studied for T cell activation. (B) Plate magnets and adaptive aAPCs allow for high throughput expansion and screening of antigen-specific T cells.

A) Reprinted with permission from Kosmides, A. K., Necochea, K., Hickey, J. W. & Schneck, J. P. Separating T Cell Targeting Components onto Magnetically Clustered Nanoparticles Boosts Activation. Nano Lett. 18, (2018), Copyright (2020) American Chemical Society. (B) Reprinted with permission from Hickey, J. W. et al. Adaptive Nanoparticle Platforms for High Throughput Expansion and Detection of Antigen-Specific T cells. Nano Lett. (2020). doi:10.1021/acs.nanolett.0c01511, Copyright (2020) American Chemical Society.

Carbon nanotubes are another inorganic material that has been studied for both polyclonal and antigen-specific expansion of CD8+ T cells[112,113]. Some of the advantages offered by carbon nanotubes include their clustering of signaling ligands due to their unique topography, as well as their high surface area and aspect ratio, resulting in increased multi-avidity interactions and contact with T cells[112]. In addition to conjugating signal 1 and 2 to carbon nanotube aAPCs, one study also conjugated PLGA nanoparticles loaded with IL-2 and magnetite to the nanotube, allowing for local delivery of IL-2 for enhanced ex vivo T cell stimulation, as well as a means of magnetically removing particles prior to adoptive transfer of T cells[112]. These composite aAPCs enhanced long-term ex vivo expansion of murine and human antigen-specific CD8+ T cells and delayed tumor growth in a mouse melanoma adoptive transfer model.

A variety of polymeric materials have also been used to produce aAPCs. Due to their widespread use, ease of coating, and biocompatibility, polystyrene microspheres were used commonly in the initial production of fully synthetic aAPCs[114–118]. The flexibility of size and control over ligand composition and density using polystyrene microparticles provided some of the first insights into optimal particle-based stimulation of T cells (see sections 3.1.2 and 3.1.3). PLGA aAPCs have also been explored, due to their in vivo biocompatibility as well as some unique properties that enable efficient ex vivo T cell stimulation. Since they are formed through emulsion techniques and are biodegradable, they can be used to encapsulate and slowly release soluble signals such as cytokines. A number of studies have used this property to design aAPCs that mimic endogenous APCs in providing paracrine delivery of IL-2 when aAPCs bind to and activate cognate T cells[112,119,120]. These studies found that paracrine release of IL-2 was significantly more potent at activating antigen-specific murine and human CD8+ T cells than similar concentrations of bulk IL-2[119]. PLGA particles are also plastic, allowing them to be reshaped to alter their surface contact with CD8+ T cells[121,122].

The range of materials explored for ex vivo antigen-specific T cell expansion reflect the various aspects of endogenous T cell-APC interactions that researchers have sought to recapitulate synthetically. For instance, liposomes, iron oxide, and PLGA particles can each provide control over the T cell-aAPC interface in distinct manners. Liposomes mimic endogenous T cell-APC interactions through their membrane fluidity and incorporation of microdomains with pre-clustered stimulatory signals, iron oxide particles can use an external magnetic field to allow for nanoscale clustering of TCRs bound to aAPCs, and PLGA particles can be stretched to tune the number of available ligands at the T cell-aAPC interface.

The various formulations also need to satisfy important criteria for translation of aAPC technologies to the clinic, such as shelf-life, safety, and flexibility in tuning T cell specificity and optimizing T cell function for personalized cancer immunotherapies. In terms of scalability, liposomal formulations tend to be less stable and more difficult to manufacture than iron oxide or PLGA particles, somewhat hindering their “off-the-shelf” potential. In terms of biocompatibility, liposomes and PLGA aAPCs are biodegradable and safe for in vivo administration, whereas less is known about the in vivo biocompatibility of iron oxide particles[123]. On the other hand, several pre-clinical[112] and clinical[9] studies have successfully removed iron oxide particles prior to adoptive cell transfer using an external magnet. With regards to control over T cell specificity, liposomal aAPCs can be targeted towards tumor antigens by sourcing lipid bilayers from the cell membranes of a patient’s tumor cells or dendritic cells pulsed with tumor lysate. These approaches are appealing in their efforts to provide an antigen-agnostic but patient-specific cancer immunotherapy; however, they also introduce risks such as inadvertent expansion of T cell specific to antigens that are also present on healthy cells. In contrast, iron oxide aAPCs can provide control over T cell specificity through surface-loading of tumor-specific peptides and magnetic enrichment and expansion of T cells specific to these tumor antigens. This process is more labor-intensive than the antigen-agnostic approach provided by liposomes, but it is nonetheless amenable to high-throughput screens[75]; it is also safer and easier to monitor, as the T cell specificities in this case are known. Lastly, in terms of control over function and phenotype, liposomal and polymeric aAPCs present only the costimulatory molecules that were initially conjugated to their surface, whereas a range of T cell costimulatory molecules, combinations, and ratios can be co-clustered with iron oxide aAPCs within an external magnetic field[85]. On the other hand, the biodegradability of liposomes and PLGA aAPCs allows for encapsulation and paracrine release of a variety of soluble factors such as IL-2 that are vital for T cell survival, function, and phenotype.

3.1.2. Size and Shape

The size and shape of aAPCs can have a significant impact on T cell activation by altering the particle-cell surface contact area and, in turn, the avidity of TCR-pMHC interactions[110] (Fig 2D). Initial studies examining the effect of aAPC size on T cell activation found that large, 4-5 μm, cell-sized polystyrene microspheres led to greater T cell stimulation than aAPCs below 4 μm[114]. These results, along with the observation that higher doses of smaller or less dense aAPCs could not rescue T cell activation, suggested that T cell activation required large contiguous regions of TCR ligation, and did not depend solely on the total number of TCR-pMHC interactions[114]. These results were corroborated in later studies with biodegradable aAPCs, that showed 8 μm microparticles led to significantly greater activation of antigen-specific CD8+ T cells than 130 nm nanoparticles[119]. That said, later studies with 30-100 nm magnetic and quantum-dot based aAPCs showed that saturating doses of low or high density nanoparticles could lead to robust antigen-specific murine and human CD8+ T cell proliferation in vitro[124]. Mechanistic studies varying both size and ligand density of magnetic aAPCs found that 50 nm aAPCs led to lower antigen-specific T cell stimulation at similar densities of signal and doses of total protein, compared to larger 300 nm, 600 nm, or 4.5 μm aAPCs, which each performed comparably[109]. However, the suboptimal stimulation with 50 nm particles could be overcome at saturating doses. Interestingly, as in previous studies[124], the density of signal on 50 nm aAPCs did not affect T cell stimulation at similar total doses of protein, suggesting that these smaller aAPCs had monovalent or divalent interactions with T cells. Indeed, a simple calculation showed that larger 5 μm microparticles could have up to 200 bioavailable ligands at the T cell-aAPC interface, whereas 50 nm nanoparticles may have as few as one or two, due to their significantly higher degree of local curvature[109]. In contrast to the results with 50 nm particles, T cell stimulation with particles 300 nm or larger, showed a dependency on ligand density, requiring a ligand spacing below 100 nm for robust T cell activation[109]. Taken together, these results show a requirement for multivalent interactions between T cells and aAPCs when the particle footprint becomes large enough to otherwise preclude sufficient receptor occupancy within TCR nano-islands. Interestingly, a later study showed that even below 50 nm, pMHC-coated nanoparticles for activating antigen-specific CD4+ and CD8+ T cells depended on ligand density. The study found an agonistic threshold for T cell activation at an inter-pMHC distance of 17 nm[125], which corresponds roughly to the width of the TCR complex[126]. However, the nanoparticles used in this study did not have a costimulatory molecule conjugated to their surface, which may affect the minimum valency requirements for T cell triggering.

Another approach that has been used to modulate the contact area at the T cell-aAPC interface is changing the shape of the aAPC (Fig. 2D). In one study, ellipsoidal PLGA microparticles were formed by stretching spherical PLGA microparticles to aspect ratios of up to 6.6[121]. The group found that stretched particles led to increased T cell expansion across a range of total antigen doses and ligand densities, compared to spherical particles. These differences were heightened at low doses of total protein and correlated with increased binding and contact area between stretched aAPCs and T cells[121]. A follow-up study with 200 nm nanoscale PLGA aAPCs similarly found that ellipsoidal aAPCs led to increased T cell proliferation in vitro, indicating the importance of the geometry of the T cell-aAPC interface even at the nanoscale[122].

3.1.3. Ligand Choice

Another parameter that can affect aAPC function is the choice of ligands to include. As mentioned previously, the most basic signals attached to an aAPC include signal 1, which activates T cells, and signal 2, a costimulatory molecule (Fig. 2E). Signal 1 can be a pMHC molecule or an αCD3 antibody for antigen-specific or polyclonal T cell activation, respectively. Classically, agonistic αCD28 antibodies or B7.1 molecules have been used as signal 2[82]. However, a wide range of costimulatory molecules with immunomodulatory effects on T cell effector function, survival, and memory formation, have been characterized[127]. Many of these molecules, such as CD27, OX40, CD40L, 4-1BB, ICOS, and GITR[128,129], have been targeted directly with agonistic antibodies to stimulate T cell responses for cancer immunotherapy. The differential effects of co-stimulation through these molecules provide ample opportunity to design aAPCs that can elicit customized T cell responses. As an example, CD27 co-stimulation has been shown to produce short-lived but highly functional effector CD8+ T cells, whereas 4-1BB agonism can generate persistent memory cells[130]. In order to improve T cell survival and effector function, co-stimulation through 4-1BB has also been included as part of the TIL rapid expansion protocol[131].

Moreover, certain combinations of co-stimulatory molecules have been shown to have synergistic effects on T cell proliferation. For instance, co-stimulation of 4-1BB and OX40[132] has been shown to profoundly increase CD8+ T cell expansion. Several studies have applied these findings to produce more functional cellular aAPCs by using multiple co-stimulatory molecules[133,134]. Likewise, the ratios of these costimulatory molecules can affect T cell proliferation and function as well. For instance, one study with polystyrene microsphere aAPCs shows that a 3:1 ratio of αCD28 to α4-1BB could lead to up to five-fold higher frequencies of antigen-specific CD8+ T cells after several rounds of stimulation, compared to other ratios[135]. As discussed previously, another study in which signal 1 and 2 were attached to separate magnetic nanoparticles and then co-clustered with a magnetic field allowed for rapid comparisons of different ratios of co-stimulation with αCD28, α4-1 BB, and αCD27[85]. This study confirmed that a 3:1 ratio of αCD28 to α4-1BB particles led to increased proliferation and memory formation. Finally, a recent publication produced 11-molecule PLGA microparticle aAPCs, consisting of two MHC molecules, three costimulatory molecules, a CD47-Fc molecule for improved in vivo retention, and five encapsulated cytokines, chemokines, and antibodies[136]. The aAPCs showed significant expansion of endogenous tumor-specific CD8+ T cells both ex vivo and in vivo and inhibited tumor growth in a mouse melanoma model[136].

3.2. Scaffolds

While there have been fewer scaffolds reported for ex vivo expansion of antigen-specific T cells than particles, many parameters such as patterning and spacing of signaling ligands as well as material stiffness have been gleaned from studies with pMHC or αCD3 coated scaffolds.

3.2.1. Patterning of Signaling Ligands

One of the first studies showing how patterned surfaces could be used to study T cell activation used electron-beam lithography to produce micron sized grids on silica substrates and then patterned the support with lipid bilayers containing pMHC and ICAM-1 molecules[137]. The constraints imposed by these grids restricted ligand mobility and, in turn, prevented centralized TCR clustering and formation of immunological synapses. Analogous results were found using photolithography techniques to produce arrays of immobilized αCD3 and ICAM-1 molecules in various spatial arrangements. The study found that αCD3 needed to be arranged into focal, as opposed to annular, patterns to enable T cells to proliferate and secrete cytokines[138]. Later studies that incorporated αCD28 onto these APC-like arrays through multiple rounds of microcontact printing demonstrated that T cells could sense both the microscale distance between and orientation of signal 1 and 2[139,140].

Several studies have also used patterned surfaces to investigate the role of ligand density in T cell activation. A common technique used in these studies is known as block copolymer micellar nanolithography[141], which can produce arrays of gold nanoparticles with nanoscale control over interparticle distance. The arrays of gold nanoparticles can then be conjugated with signal 1 to produce surfaces with controlled ligand densities[142,143]. These studies demonstrated that αCD3-based T cell activation required interparticle spacing below 60 nm[142,143], in line with the previously mentioned study that varied ligand densities on nanoparticle-based aAPCs[109]. A more recent study has used electron beam lithography on glass coverslips to allow for patterning of αCD3 and ICAM-1 onto gold-palladium nanoparticle arrays with defined intermolecular distances[144]. Some of the arrays underwent an additional etching step to produce raised 10 nm glass pedestals. Thus, these arrays allowed for precise axial and lateral positioning of αCD3 molecules. By increasing the intermembrane distance between T cells and the activating surface, the glass pedestals allowed free diffusion of CD45 phosphatases during TCR engagement, thus inhibiting a critical component of T cell activation, CD45 exclusion[145]. In turn, arrays with raised glass pedestals had different requirements for lateral spacing of signals, compared to planar arrays[144], motivating further study of the three-dimensional patterning and spacing requirements for T cell activation.

3.2.2. Stiffness

Several studies have used materials with tunable bulk stiffnesses to understand the role of mechanotransduction during T cell activation. Poly(acrylamide)[146,147] or poly(dimethylsiloxane)[148] gels were produced with stiffnesses ranging from as low as 0.5 kPa to as high as 2 MPa by altering the amount of crosslinker present during formation of the gels. Gels were conjugated with αCD3 and αCD28 to activate T cells, and consistently across each of these studies, it was found that 100 kPa gels optimally stimulated T cells. Interestingly, one of the studies found through RNA microarray analysis that T cell transcriptional profiles show a graded response to substrate stiffness in the presence of CD3-based stimulation[147] (Fig. 4A). Furthermore, the study found that T cells secrete cytokines across a range of stiffnesses; however, only T cells on the stiffest substrates used for this study (100 kPa) became metabolically active and began cell cycle progression[147]. In turn, T cells showed the greatest proliferation after 72 hours on the stiffest substrates[147]. The importance of stiffness in modulating T cell activation has also been shown with antigen-pulsed APCs seeded onto two and three-dimensional alginate scaffolds[149]. Analogous to these previous studies, for both two-dimensional and three-dimensional gels, stiffer 40 kPa scaffolds led to better T cell activation, proliferation, and function than softer 4 kPa scaffolds[149]. Moreover, the stiffer gels led to a significant increase in the size of immune synapses (Fig. 4B).

Figure 4.

Substrate stiffness of T cell activating surfaces modulates T cell transcriptional profiled and immune synapse formation. (A) In the presence of αCD3, T cells show a graded transcriptional response to substrate stiffness. (B) Stiffness of three-dimensional substrates modulates size of immune synapses between T cells and APCs, with stiffer substrates leading to larger synapses.

A) Reprinted with permission from Saitakis, M. et al. Different TCR-induced T lymphocyte responses are potentiated by stiffness with variable sensitivity. Elife 6, e23190 (2017) under the Creative Commons Attribution License (B) Reprinted from Majedi, F. S. et al. T-cell activation is modulated by the 3D mechanical microenvironment. Biomaterials 252, 120058 (2020), Copyright (2020), with permission from Elsevier.

3.2.3. Applications for Cancer Immunotherapy

The findings above have been invaluable to the understanding of T cell biology and how to better create substrates for the activation of tumor-specific T cells for adoptive immunotherapy. One study synthesized a hydrogel from hyaluronic acid, a common component of the extracellular matrix (ECM), and conjugated it with pMHC and αCD28 to form an artificial T cell stimulating matrix (aTM) for antigen-specific T cell stimulation[150]. The study investigated the effects of stiffness, ligand density, and a variety of ECM proteins on T cell proliferation, function, and phenotype. Interestingly, in contrast to some of the studies previously mentioned, this study found that softer 0.5 kPa gels led to significantly greater proliferation, function, and CD3 cluster formation than stiffer 3 kPa gels. Moreover, culturing T cells on the aTM led to an increase in expansion and polyfunctionality of endogenous antigen-specific CD8+ T cells, compared to culturing T cells on a tissue culture plate or a blank hydrogel in the presence of magnetic nano-aAPCs. Similarly, endogenous T cells expanded on the aTM and adoptively transferred into tumor-bearing mice significantly slowed tumor growth and increased mouse survival, compared to T cells stimulated with nano-aAPCs on a tissue culture plate. Another study produced a composite APC mimetic scaffold (APC-ms) by forming a supported lipid bilayer with T cell signaling cues on high-aspect ratio mesoporous silica micro-rods (MSRs)[81]. The MSRs were pre-loaded with IL-2, coated with liposomes, and then conjugated with signal 1 and 2 for T cell stimulation. In vitro, the MSRs self-assembled into a three-dimensional scaffold with high enough porosity to allow for cell infiltration. T cells cultured with the APC-ms formed denser clusters with the MSRs than with traditional aAPC microbeads, due in part to both the larger size (70 μm vs 4.5 μm) and higher aspect ratio of the MSRs. Moreover, paracrine release of IL-2 from the MSRs was shown to be more potent at T cell stimulation than adding the same amount of bulk IL-2, as in previous studies[112,119]. This platform was shown to be more effective than aAPC microbeads for polyclonal expansion of primary T cells or tumor-specific CAR T cells. Similarly, the platform outperformed moDCs for expansion of rare antigen-specific CD8+ T cells. CAR T cells expanded with APC-ms showed similar antitumor efficacy compared to CAR T cells expanded with aAPC microbeads. Together, the results from these two studies show promise for future clinical studies using artificial scaffolds to expand patient-specific CD8+ T cells for ACT.

4. In Vivo Biomaterial Platforms for Antigen-Specific T cell Activation

As researchers have developed increasingly better particles and substrates for T cell activation that are efficient and avoid off-target activation, there has been a growing interest in using these materials to directly activate T cells in vivo. Biomaterial-based technologies for in vivo T cell activation are an attractive option to alleviate many of the challenges associated with adoptive cell transfer. Avoiding the ex vivo T cell culture step drastically lowers cost, time-to-treatment, and exhaustion of cells due to long culture periods. In vivo antigen-specific CD8+ T cell activation is a relatively new venture in T cell therapy, yet the field has generated a lot of interest and is propelling forward with several active clinical trials[151,152].

In vivo T cell activation strategies can broadly be divided into two categories: vaccines and direct T cell activation. Vaccines work by activating host APCs, which in turn activate host T cells (Fig. 5A,B). Direct T cell activation bypasses host APCs by recruiting, directly binding to, and activating host T cells (Fig. 5C,D). Each of these categories is largely approached with the use of either particle- or scaffold-based platforms. Particles typically travel via circulation or lymphatics and activate either DCs or T cells in the periphery or in the lymph node (Fig. 5A,C). Scaffolds are typically injected or implanted subcutaneously and act either through the release of soluble activating molecules to activate nearby cells or through recruitment of cells into the scaffold itself for activation through direct contact (Figure 5B,D).

Figure 5.

Methods for in vivo activation of antigen-specific T cells. (A) Vaccine particles may be injected subcutaneously or intravenously and either taken up by DCs in the periphery or travel directly to the LN where they interact with DCs. From the periphery, activated DCs will travel to T cell zones of the LN to activate cognate naïve T cells. (B) Vaccine scaffolds are typically injected or implanted subcutaneously with antigen and adjuvant encapsulated for slow release. The scaffold may either activate local tissue resident DCs or may actively recruit DCs via chemoattractants. Activated DCs leave the scaffold and travel to the LN for T cell activation. (C) aAPCs for direct T cell activation may be injected intravenously or subcutaneously. Intravenously injected particles may drain to the spleen, while subcutaneously injected particles may drain via lymphatics into the lymph node. Once in secondary lymphoid organs, particles can directly interact with and activate T cells. (D) Scaffolds for direct T cell activation are injected or implanted subcutaneously and have encapsulated or conjugated T cell stimulating molecules. Scaffolds, infused with chemokines, attract na’ive T cells from systemic circulation. T cells are activated within or on the surface of the scaffold, then travel back into systemic circulation. After T cells have entered systemic circulation, they can travel to the site of infection or cancer via endogenous mechanisms to kill cells expressing the target antigen. Created with BioRender.com.

General design considerations for injectable biomaterial-based particles and scaffolds have been reviewed extensively[153–155]. To summarize, particles and scaffolds must be biocompatible, biodegradable, and display favorable pharmacokinetics. Biomaterial scaffolds must also have a porous structure, be easily injectable, and avoid foreign body responses. These properties may be inherent to the chosen material or may be engineered. Biomaterials have been utilized in vivo to treat cancer for decades, mostly for drug delivery. Platforms specifically for T cell activation share some considerations with drug delivery vehicles, such as targeting distinct organs and cell types, but also have their own unique considerations.

4.1. Vaccines

Vaccines are arguably the oldest and most successful form of immune engineering[156]. Since the first vaccine – cowpox, by Edward Jenner – numerous pathogens have been successfully controlled by vaccines. The first vaccine for cancer can be traced to the end of the 19^th century. Dr. William Coley came across a case of a patient whose malignant tumor receded after developing cellulitis, a bacterial skin infection[157]. He began treating his cancer patients with streptococcal organisms, “Coley’s Toxin”, in his first attempts at a therapeutic cancer vaccine[157]. Coley’s results were eventually deemed irreproducible, but they sparked interest in the development of an effective, reproducible cancer vaccine.

At a minimum, vaccines must include an antigen to direct the specificity of the immune response and an adjuvant to activate the immune response. Initial vaccines typically injected these components as separate entities in one mixture. For most diseases, these early vaccines induced strong antibody development, which was enough to neutralize pathogens. However, a strong CD8+ T cell response is harder to elicit, yet necessary for cancer treatment[158]. Since vaccine components are delivered extracellularly, the antigen is by default loaded onto MHC class II molecules, stimulating CD4+ helper cells which in turn support B cell growth and maturation. CD4+ T cells are also known to support CD8+ T cell growth and differentiation, but CD8+ cells must first be activated by DCs presenting MHC class I. The process of uninfected DCs loading antigen onto MHC class I is referred to as cross-presentation, and it is often a central component of cancer vaccine design. CD4+ T cells are also thought to be necessary for anti-cancer efficacy of vaccines[159], but as MHC class II loading is the default loading pathway for exogenous antigen, it is often not necessary to add extra design considerations for eliciting CD4+ T cell responses.

Biomaterials have been invaluable to the field of vaccine development as they can not only improve trafficking of vaccine components in vivo, but they have been shown to induce cross-presentation of antigen by DCs. Biomaterial-based vaccine composition, delivery route, modality, timing, and dose have all been shown to have a strong impact on T cell activation, as discussed below[160,161].

4.1.1. Particle-based vaccines

Particle-based vaccines are designed to deliver antigen and adjuvant directly to antigen presenting cells, most commonly DCs. As mentioned, the particle must promote cross-presentation of the antigen onto MHC class I molecules in order to activate CD8+ T cells. Vaccine nanoparticles for T cell activation must be efficiently internalized by DCs, rather than other phagocytic cells which are inefficient at cross-presentation. Once taken up, the particles must release antigen and trigger DC maturation and activation. Intervention in each of these steps can be accomplished with the design of biomaterial-based vaccine carriers.

To target cross-presentation pathways, many vaccines are designed to enhance direct delivery of vaccine particles to DCs[162]. In the periphery, DCs constitutively uptake molecules through non-specific macropinocytosis and receptor-mediated endocytosis in order to survey the environment[163]. Vaccine particles less than 50 nm are internalized via macropinocytosis whereas particles larger than 500 nm are internalized via receptor-mediated pathways[162]. Passive targeting of DC uptake can be achieved through modulation of particle material, size, shape, and surface charge[164–167]. Active targeting can be achieved by directing the particles towards receptors involved in receptor-mediated endocytosis. For example, mannose and fructose moieties can interact with membrane-bound lectins on DCs, which are involved with phagocytosis[168]. Several vaccine particles take advantage of this pathway by coating the particle surface with mannose and mannose-mimicking molecules[169–171]. Other endocytic receptors on DCs, such as DEC-205, DC-SIGN, Clec9A, DNGR-1, and Fc receptors[166,172–177], have also been targeted. However, Clec9A, DNGR-1, and DEC205 seem to be particular effective at promoting cross-presentation through endosomal pathways [176,178–180]. More complex platforms can also be designed to deliver and release vaccine components at critical moments during uptake by DCs, such as during endosomal escape, to enhance cross-presentation efficiency[162,181,182].

The two DC subtypes that are most adept at cross-presentation of antigens are CD103+ and CD8+ DCs[183]. CD103+ DCs largely reside in nonlymphoid tissue, and the above DC targeting methods can be used to increase vaccine accumulation in these DCs. Targeting these populations is made easier by the fact that they exist throughout the body, in mucosal surfaces, the skin, and the lungs primarily. However, an alternative approach is to target CD8+ DCs, which reside in lymphoid tissue. In the spleen, these cells are located in the marginal zone or the red pulp, making them easily accessible to circulating particles. In the lymph node, CD8+ DCs have been found to line the subcapsular sinus along with macrophages[184] and are therefore the first cells that vaccine particles may interact with upon reaching the lymph node. In addition to high concentrations of CD8+ DCs being accessible to circulating particles, DCs in lymphoid tissue have the added benefit of already being in close proximity to T cells. Properties of biomaterials to enhance lymphatic drainage to the LN have been reviewed extensively[185]. Particle size, shape, charge, hydrophobicity, rigidity, and targeting ligands all have significant effects on not only LN drainage but also retention, which directly affects the T cell response[161]. Beyond modifying these particle surface properties, unique biomaterials designs can enhance vaccine particle delivery even further. One approach involves so-called “albumin hitchhiking”, linking vaccine cargo to albumin, which has been shown to efficiently drain to lymph nodes[54,186]. One barrier to efficient delivery to LNs is the differing size requirements for each stage of transport. Particles between 25 and 100 nm demonstrate enhanced direct lymphatic drainage; however, once in the LN, only particles less than 70kDa (~5nm) can penetrate deeply into the LN into T cell zones[185]. To overcome these opposing design considerations, one group has developed a multistage delivery platform[187]. The vaccine particles, which are 27 nm, preferentially drain through the lymphatics as they are too large for free diffusion through the vascular endothelial layer[188]. The individual vaccine components within the particles are initially sequestered within the nanoparticle by linkage to larger molecules. The linker degrades in the approximate time that it takes for the particles to reach the LN, allowing for the vaccine components to be released from the nanoparticle and passively diffuse into the interior of the LN[187]. Another recent study has shown that clodronate-mediated depletion of subcapsular sinus macrophages, the main barrier to the LN interior, can improve vaccine accumulation[189]. However, this method has only been tested for its effects on inducing a humoral response rather than a T cell response.

To efficiently facilitate activation and cross-presentation, it is important that DCs receive both components of the vaccine – antigen and adjuvant – simultaneously in order to avoid a tolerizing effect[190]. One group has avoided premature release of vaccine components with a pH switch[191]. The pH switch helps circumvent burst release and minimizes off-target effects. However, again, the real barrier that has plagued cancer vaccines is the lack of systems that can effectively promote cross-presentation. In the most common pathway of cross-presentation, exogenous antigen is phagocytosed or otherwise internalized, then can escape from the phagosome into the cytosol where it follows traditional loading onto MHC I in the ER [181,192]. Cytosolic exportation from the phagosome or endosome is a key mechanism through which vaccines may induce cross-presentation. Indeed, methods to induce endosomal rupture upon vaccine particle uptake have shown robust results[181]. Another popular approach is the use of liposomal carriers, which may fuse with the endosome membrane to release antigen into the cytosol[181]. Most notably, the RNA-Lipoplex system has shown robust T cell responses for cancer[61,152]. In this system, liposomes carrying RNA encoding tumor antigens were targeted to DCs throughout the body by altering the lipid composition to obtain a net positive charge on the liposome surface, leading to efficient uptake and expression of the target antigen on DCs and subsequent activation of effector T cells. This platform is currently in clinical trials for malignant melanoma[152]. More recent studies are working to elucidate how details of the particle synthesis affect antigen-specific responses. For example, one study has shown that both the peptide modifications and crosslinking chemistry used to encapsulate antigen affect DC maturation and T cell activation[193]; this study along with future mechanistic studies will be vital to the optimization of cancer vaccines.

4.1.2. Vaccine Scaffolds

Despite advancements, vaccine particles still struggle to overcome barriers such as efficient delivery to DCs and retention of vaccine components in cell-rich areas like lymph nodes[194]. An alternative approach is to inject or implant a biomaterial scaffold that will attract peripheral DCs to the implant site for activation. This approach may also allow for single-shot vaccines, a highly coveted standard of vaccination, as it can provide controlled and sustained release of its components. In addition to providing the two critical components of a vaccine, antigen and adjuvant, vaccine scaffolds must also release a DC attractant to enhance cell infiltration from the periphery (Fig. 2B). Activated DCs must then leave the scaffold to travel to the LN as they would after being activated endogenously.

The first step in the vaccine scaffold approach is recruitment of naïve DCs to the site. This has most commonly been accomplished using cytokine granulocyte–macrophage colony-stimulating factor (GM-CSF)[194–197]. Inclusion of a cell adhesion motif such as Arg-Gly-Asp (RGD) has also been shown to help recruit and retain DCs[198]. Vaccine scaffolds may be used solely as a depot for the release of adjuvant and antigen, but more recent designs are macroporous or otherwise assembled in such a way that allows the entry of immune cells into the scaffold, where they may access cytokines and other vaccine components. The pore size required for cell infiltration makes encapsulating small vaccine components challenging. One approach to address this issue is to create dual-porous scaffolds. For example, cryogelation may be used to create the large pores required for cell influx, while the base gel component maintains a smaller pore size that allows for sustained release of encapsulated components[196]. Another dual-porosity approach is to take advantage of self-assembling structures like mesoporous silica rods[195,198]. The rods themselves have pore sizes for small-molecule loading, while in vitro and in vivo they spontaneously assemble into structures that maintain large enough pore sizes for immune cell infiltration. Finally, researchers have combined scaffolds and particles to ensure sustained delivery of vaccines (Fig. 6A)[197,199]. In this study, MSRs carrying GM-CSF are loaded with mesoporous silica nanoparticles (MSNs) that contain with antigen (OVA) and adjuvant (Toll-like receptor 9 agonist). This approach takes advantage of the benefits of nanoparticle vaccines – controlled, simultaneous delivery of vaccine components into phagocytic cells – while improving on the limitations of particles by allowing them to activate cells efficiently in the periphery. While this study uses OVA as a model antigen, like many in vivo platforms to date, the technology is highly modular and can theoretically be adapted for loading of TAAs or neoantigens. Using defined antigens such as these has the benefit of being both personalizable and predictable. However, TAAs are only relevant for patients whose tumors express them at high enough levels and neoantigens can be challenging and time consuming to discover. Multiple antigens must also be targeted simultaneously to more effectively treat highly heterogeneous tumors. As an alternative, many studies have attempted to use tumor cell lysate in cancer vaccines[200], with several recent reports of clinical efficacy [201,202]. Vaccine scaffolds using tumor lysate have also shown efficacy with limited toxicities for a range of mouse tumor models [196,203,204], and one such scaffold is currently in Phase I clinical trials[205]. However, the approach of targeting undefined antigens in vivo carries a risk of inadvertently expanding autoreactive T cells, without the ability to pre-screen for self-reactivity of the expanded T cells, as is done with ex vivo expansions. As a result, extensive characterization of the T cell response in humans will be necessary for translation.

Figure 6.

Biomaterials-based nanoparticles for vaccines and aAPCs can elicit strong antigen-specific CD8+ T cell responses in vivo. (A) A combination vaccine approach containing mesoporous silica microrods (MSRs) coupled with mesoporous silica nanoparticles (MSNs) increases the population of antigen-specific CD8+ T cells and improves the anti-tumor response compared to particles or scaffolds alone. (B) Altering the shape of biodegradable aAPCs improves antigen specific T cell activation in vitro as well as decreases non-specific uptake by macrophages, resulting in enhanced T cell activation in vivo as well.

(A) Reproduced with permission from Nguyen, T. L., et al. Injectable dual-scale mesoporous silica cancer vaccine enabling efficient delivery of antigen/adjuvant-loaded nanoparticles to dendritic cells recruited in local macroporous scaffold. Biomaterials 239, 119859 (2020). (B) Reproduced with permission from Meyer, R. a et al. Biodegradable Nanoellipsoidal Artificial Antigen Presenting Cells for Antigen Specific T-Cell Activation. Small 11, 1519–1525 (2014).

While vaccine adjuvants may be loaded into the structures described above, the biomaterial itself may also act as an adjuvant. For example, it has been shown that mesoporous silica has adjuvant properties by stimulating pattern recognition receptors (PRRs)[195,198]. Other biomaterials may increase in immunogenicity as degradation occurs, such as poly(beta-amino ester) (PBAE), as the material experiences changes in surface properties, size, and shape[206]. In addition to the material having inherent adjuvant activity, it is also possible to modify surface properties of the material, such as charge, to increase immunogenicity[207].

Vaccines are an attractive approach for stimulating an antigen-specific T cell response as they avoid the difficulty of expanding rare T cell clones ex vivo. Several biomaterial-based cancer vaccines are currently in clinical trials, summaries of which can be found in references [208] and [209]. However, despite advances, it is still challenging to elicit a cellular response from vaccines. Furthermore, cancer patients often have dysfunctional DCs[210], rendering vaccine approaches ineffective.

4.2. Direct T cell Activation

While still a relatively new area of exploration, in vivo direct T cell activation may offer solutions to the problems that plague both adoptive T cell transfer and vaccination. Compared to ex vivo activation of T cells, in vivo activation saves time and money and moves the field closer to “off-the-shelf” treatments[82]. Compared to vaccination, direct T cell activation offers more direct control over the specificity and phenotype of the immune response as it avoids relying upon intermediary players, such as endogenous antigen presenting cells. It may also avoid adverse events that are experienced from current cancer vaccines[211,212] while reducing the number of injections necessary for efficacy[213].

This approach to expand endogenous antigen specific cells in vivo shares many design considerations with aAPCs used for ex vivo T cell expansion described above. However, with in vivo stimulation, there is increased concern of activating off-target cells, without the advantage of being able to characterize the cell product prior to infusion, as in ex vivo expansion systems. Platforms must include signals 1, 2, and 3 for T cell stimulation, activate cells without overstimulating them, and avoid activating non-specific T cells.

As mentioned with vaccine design, considerations like biocompatibility, biodegradability, and pharmacokinetics exist here as with any other pharmaceutical agent. However, there are several important distinctions between the vaccine approach and the direct T cell activation approach. First, unlike with vaccine platforms, T cell activators must avoid being internalized by phagocytic cells as they need to directly interact with T cells. Second, it is disadvantageous for the material itself to induce an immune response as this will likely attract innate immune cells rather than naïve T cells. Finally, the platform must be designed with a specific patient in mind, as endogenous T cell activation is HLA restricted. Here, we will discuss current technologies that enable in vivo T cell activation and how they are designed to enable antigen-specific activation without activating off-target effector cells.

4.2.1. Particle-based T cell activation

aAPCs were originally evaluated for their ability to expand T cells ex vivo for ACT but are now being adapted for in vivo applications for both CD4+ cells and CD8+ T cells[106,214]. Similar to aAPCs for ex vivo use, the minimum requirements for aAPC use in vivo are the incorporation of signals 1 and 2 for T cell activation. Particles with pMHC complexes and αCD28 for co-stimulation have been shown to be at least partially effective at treating tumors upon either intravenous or subcutaneous co-injection of aAPCs and naïve T cells[106,124,213,214] and an intraperitoneal injection of IL-2 (signal 3) to support T cell growth. These initial studies provided the proof of concept that aAPC particles can in fact support T cell growth in vivo. Improvements to these base aAPC designs have focused on particle circulation, LN drainage, and interaction with and activation of T cells.

One challenge with injectable aAPCs is directing accumulation to secondary lymphoid organs, where naïve T cells are concentrated. Since T cells continuously cycle through the LN, it is an ideal target for aAPC accumulation. Unlike vaccine particles, which can target DCs in the subcapsular sinus, aAPCs have to penetrate deeper into the T cell zone within the paracortex of the LN. If aAPCs are injected subcutaneously, small particles (< 5 nm) will enter the blood stream while large particles (~50 - 100 nm) will enter lymphatics[188]. While targeting lymphatic drainage is an attractive solution for vaccine accumulation in the LN, larger aAPCs would likely not be able to pass the subcapsular sinus barrier and thus would not interact with T cells. Instead, delivery via vasculature may be better as it more closely mimics native T cell trafficking[215]. Particles can even be functionalized to target draining LN High Endothelial Venules (HEVs), which are the same channel of entry used by T cells[216].

Several aAPC properties have been shown to impact in vivo T cell activation (Fig. 2). One study investigated PLGA-based aAPCs of different shapes with pMHC and αCD28 conjugated to the surface[122] . The study found that ellipsoidal aAPCs not only enhanced T cell activation but also reduced uptake by macrophages thus improving in vivo half-life over spherical particles (Fig. 6B)[122]. Research into the effect of ligand positioning on aAPCs in vitro has been extended to in vivo results, demonstrating that both lateral and axial control of ligand positioning has an impact on T cell expansion in vivo[217]. Using red blood cells as the base material for the aAPC with DNA linkers to attach pMHC and aCD28 proteins, researchers found that pre-forming clusters of pMHC and using shorter linkers for pMHC conjugation improves the performance of injectable aAPCs[217].

As described previously, many aAPC properties have been studied in vitro and these findings will likely be relevant in vivo as well. However, tradeoffs exist for injectable aAPC design as biocompatibility and pharmacokinetics must be balanced with aAPC properties that optimize T cell activation and expansion. As an example, the most efficient size for effective LN drainage may be at odds with the size necessary for enhanced T cell contact; it has been shown that effective LN drainage occurs with particles around 50 nm[185] whereas effective T cell activation occurs with particles above 300 nm[109].

Finally, as discussed for ex vivo technologies, ligand choice is an important aspect of injectable aAPC design (Fig. 2). The pivotal work that has been done to characterize contributions of various stimulating receptors on T cell activation has been extended in vivo; for example one group that incorporated pMHC, αCD28, and α4-1BB molecules onto a latex-based aAPC[218] showed enhanced in vivo T cell activation. One concern of in vivo T cell activation is the activation of off-target T cells through the costimulatory molecules on aAPCs[219]. In the study mentioned previously, the group did observe some non-specific activation of T cells due to the co-stimulatory molecules; however, for cancer applications it may be advantageous, as delivery of these molecules alone has been shown to have anti-tumor effects[220]. Beyond improving the specificity and efficiency of T cell activation itself, researchers may consider adding ligands to aAPCs to improve the pharmacokinetic profile of the particle, for example by reducing uptake of particles by the reticuloendothelial system (RES). In an 11-molecule aAPC, previously discussed in section 3.1.3, the group added CD47-Fc, the “don’t eat me signal,” to the surface of the particle. Addition of this signal improved accumulation in the spleen and LN, presumably through reduced phagocytosis, and enhanced the anti-melanoma immune response as compared to particles without CD47-Fc. There has been a large body of work describing methods to create so called “stealth” particles that show improved circulation in the body, which can be extended to the field of aAPCs[136,221].

4.2.2. Scaffolds

As with vaccine development, the difficulty of efficient accumulation of aAPCs in secondary lymphoid organs gave rise to a second approach for in vivo T cell activation – the T cell stimulating scaffold. This goal of this approach is to create an injectable or implantable biomaterial scaffold that incorporates chemokines and T cell stimulating signals that can recruit and expand T cells at the peripheral site; in other words, the creation of an “artificial lymph node” that recreates some or all of the functions of a LN. Thus far, this goal has not been fully realized, but the field is growing with many promising pre-clinical studies.

There have been several scaffolds developed to support pan-T cell expansion and subsequent in vivo delivery. One group demonstrated that T cells activated overnight with αCD3 and αCD28 antibodies can be embedded in a polyisocyanopeptide (PIC) hydrogel with IL-2 for subcutaneous delivery into mice[222]. The group was able to see continued T cell activation and slow release from the gel, indicating suitability of the platform for sustained local delivery of pre-activated T cells. The sustained release itself is a significant contribution to the field of ACT, as persistence of transferred cells has been a challenge in humans[223]. Another group has developed chitosan gels to co-encapsulate TIL with tumor fragments, which promote the expansion of TILs[83]. The group achieved a linear release of encapsulated TILs over time, which could then kill target cancer cells[83]. However, the approach is still in development and more in vivo testing is needed. Similarly, a group has developed an alginate scaffold to encapsulate tumor reactive T cells, which can be implanted at the tumor resection bed or at the site of an inoperable tumor[84]. The scaffold included a synthetic collagen-mimetic peptide for T cell adhesion. T cell activation was mediated by encapsulated silica microparticles coated with lipid bilayers including αCD3, αCD28 and α4-1BB and encapsulated IL-15 superagonist. The T cells were steadily released as the biomaterial degraded and treated mice showed enhanced survival after tumor challenge[84]. To further reduce the manufacturing burden of T cell therapy, eventually the goal of this approach would be to attract host T cells rather than co-delivering isolated T cells. One step that has been taken towards this end is the encapsulation of chemokine CCL21 into PEG hydrogels containing αCD3/αCD28 microparticles[224]. The platform was only tested for ex vivo T cell activation, but future experiments may use CCL21 to attract host cells into the scaffold for subsequent activation and release.

Another avenue of research has been to fully construct artificial lymph nodes by co-implanting biomaterial scaffolds with LN stromal cells. Efforts here were started in the early 2000s with formation of LN-like organoids that reconstitute some functions of the LN like B and T cell retention[225,226]. More recent research has produced much more sophisticated systems that are capable of enhanced functionality without the need to co-implant stromal cells. One group created artificial lymph node-like tertiary lymphoid organs (artTLOs) by implanting collagen sponge scaffolds with beads that facilitated the slow release of lymphorganogenic chemokines (lymphotoxin-α1β2, CCL19, CCL21, CXCL12, CXCL13, and soluble RANK ligand)[227]. The platform showed the capacity to produce an antigen-specific immune response to immunizations in SCID mice[227]. Another group created a bone-marrow mimicking alginate cryogel (BMC) to enhance T cell regeneration after hematopoietic stem cell transplantation (HSCT)[228]. These technologies that restore lymphoid function could be used in patients whose cancer has rendered them immunosuppressed, to allow them access to immunotherapy treatments.

4.2.3. Biomaterials that enhance ACT or the natural T cell response

In working towards developing technologies for in vivo endogenous antigen-specific T cell expansion, some platforms are currently acting as support mechanisms to traditional forms of ACT, such as ETC, TCR, or CAR T therapies. One challenge with ACT, either with endogenous cells or with engineered cells, is the persistence of transferred cells[229,230]. In order to potentiate the efficacy of the immunotherapy by extending the persistence, groups have developed various systems to support injected cells. One group has created a nanogel “backpack” that contains IL-15 superagonist (IL 15sa)[86]. This approach not only supports the cells in vivo, but it potentiates the immune response in an antigen-specific manner, as the release of IL-15sa is local and is only triggered upon T cell recognition of its target[86]. Another method has been the co-injection of T cells and a vaccine. For example, in one study DCs were pretreated with the adjuvant monophosphoryl lipid A[231]. Vesicles from these cells were isolated and co-injected into recipients with pre-activated CD8+ T cells. The vesicles acted as a vaccine to activate DCs in vivo which in turn activated antigen-specific T cells in vivo, potentially both transferred and endogenous cells[231]. Vaccines have also been co-delivered with CAR T cells to re-stimulate cells after transfer[54,61]. Another group used a non-specific vaccine to augment ACT; biopolymer scaffolds delivering stimulator of IFN genes (STING) agonists could expand tumor-specific T cells that recognize other tumor epitopes that the transferred cells did not, broadening the anti-cancer immune response[232]. Besides enhancing the persistence of adoptively transferred cells, one group has enhanced ACT by improving the accumulation of transferred cells in the tumor site. Prior to injection, T cells were decorated with magnetic APCs so that upon transfer, cells could be driven to the tumor by MRI and magnetic guidance[233].

There have also been developments in the enhancement of the endogenous T cell response without the need for ACT. These technologies use biomaterials to engineer tumor cells in situ to induce them to secrete or express molecules that support T cell growth and function within the tumor microenvironment. One group has developed a PBAE nanoparticle that delivers both IL-12 and 4-1BBL DNA into the tumor, turning tumor cells into APC-like cells, which restored function to tumor-specific T cells[234]. Another group has created tumor-targeting lipid-dendrimer-calcium-phosphate (TT-LDCP) nanoparticles that deliver IL-2 DNA and PD-L1 siRNA to induce tumor cells to support T cell growth while removing T cell inhibition [235]. Because these approaches work to enhance the endogenous T cell response, they can be antigen agnostic which circumvents many of the challenges faced by ETC. However, their efficacy is limited in settings where tumors inherently have poor T cell infiltration.

Finally, T cell therapies have been greatly enhanced by the advent of immune checkpoint blockade (ICB) where antibodies against “immune brakes” potentiate anti-cancer T cell responses in vivo. ICB has already seen great success clinically, alone and in conjunction with ACT[236,237]. In a murine model, ICB has also been shown to synergize with aApCs; upon co-delivery of aAPCs with αPD-1, the group saw enhanced IFN-γ secretion by CD8+ T cells as well as tumor regression in a melanoma model[213]. Biomaterials may also have a role in improving ICB delivery. For example, ICB may be coated onto nanoparticles to enhance their accumulation in the tumor microenvironment[238–241]. Another group has taken this idea further with the development of the “immunoswitch”[242], a nanoparticle conjugated with αPD-L1 and α4-1BB to not only block the checkpoint blockade signal, but also replace it with a co-stimulatory signal[242]. As technologies for in vivo T cell activation continue to advance, they will likely be used in conjunction with technologies like those mentioned here. Combination immunotherapies are already used in the clinic today and are rapidly becoming the standard of care[243,244].

In vivo generation of a cytotoxic, antigen-specific T cell response is an active field of research with growing excitement. Biomaterial-based therapeutic cancer vaccines are currently the closest to market, with many active clinical trials. However, there has been a long history of cancer vaccines failing to show anti-tumor efficacy in humans[209]. Significant progress has been made in improving the immunogenicity of these formulations through targeting neoantigens; nevertheless, the efficacy of such technologies may be inherently capped for patients with dysfunctional APCs. In these cases, biomaterials that directly activate T cells in vivo may be more effective. The approach of using aAPCs and scaffolds in vivo benefits from decades of ex vivo characterization and optimization of these technologies; however, the safety of direct T cell activation in humans has yet to be studied and may prove to be problematic. In patients who already have underlying antitumor T cell responses but have yet to “break” tolerance within the immunosuppressive tumor microenvironment, technologies that enhance the natural immune response may be the most effective and have the added benefit of being the closest to “off-the-shelf” technologies. As with direct T cell activation, these biomaterial technologies are still largely pre-clinical.

5. Summary and Conclusions

T cells play an integral role in directing immune responses against infected or cancerous cells. Biomaterial platforms that can mimic these natural immune responses can not only enhance our understanding of T cell biology, but can be harnessed towards design of immunotherapies tailored towards a variety of cancers and infectious diseases. Clinically, ACT-based therapies show great promise but are still not first-line therapies. The use of biomaterials for ex vivo manipulation of T cells, in vivo delivery of immunomodulatory agents, or in vivo activation of T cells has the potential to drastically change the landscape of cancer immunotherapy and bring the benefits of T cell therapy to more patients.

Cost, manufacturing difficulty, time-to-treatment, and patient variability all prevent the widespread use of ETC in the clinic. Mass-producible biomaterials that can provide the modularity necessary to induce robust T cell responses against a wide range of cancer antigens, with the flexibility to skew towards unique functional and phenotypic profiles, could translate into the next-generation of immunotherapies. “Off the shelf” aAPCs can help reduce the cost and length of ex vivo culture for personalized cell therapies while smart material choices can potentially simplify regulatory approval when the materials can be separated from T cell products prior to infusion. Cost reduction, customizability, streamlined regulatory proceedings, and lack of dependence on the availability and quality of patient-derived APCs can all help widen the pool of patients who can benefit from ACT.

Use of biomaterials in patients has been explored in depth in fields such as regenerative medicine and tissue engineering. However, biomaterials for in vivo activation of antigen-specific T cells is a relatively new, alternative approach to cancer immunotherapy. Vaccines have been around for a century, but rarely stimulate cytotoxic T lymphocytes. Biomaterials that enhance the vaccine’s ability to elicit a cellular response by increasing host APC cross-presentation of cancer antigens have renewed interest and hope in a cancer vaccine[152]. Alternatively, direct activation of host T cells in vivo may be particularly helpful for patients who have deficient APCs, a common occurrence in many cancers. The path towards translation of technologies for in vivo T cell activation has been partially paved by decades of biomaterials characterization. Many materials have defined toxicities, immune interactions, and degradation profiles that will alleviate some of the regulatory hurdles that exist for new therapies. However, materials for in vivo T cell activation have their own translational concerns, such as potential for on-target off-tumor effects. These risks can be mitigated through thoughtful materials design as discussed throughout this review, as well as careful monitoring of patient responses.

Several endogenous T cell therapies using iron oxide aAPCs for ex vivo expansion are currently in clinical trials[245,246]. As for in vivo expansion technologies, cancer vaccines are the most well studied platform in the clinic, with more than a dozen active clinical trials[208,209]. Biomaterials for direct T cell activation is a newer approach, with most research in pre-clinical stages. However, a novel antibody fusion protein designed for direct antigen-specific T cell activation in vivo is currently being investigated in clinical trials, providing evidence that the concept of direct T cell activation is translatable[219]. Continued development of these biomaterials-based technologies for endogenous antigen-specific T cell activation may offer a truly universal, “off-the-shelf” treatment for many cancer patients. As many studies have shown, there is great synergy between T cell-based therapies and other immunotherapies such as ICB. The future of cancer immunotherapy will likely involve intensive screening of the patient’s genome and mutanome to create a comprehensive treatment plan that includes targeted stimulation of a range of innate and adaptive immune cells in tandem with administration of immunomodulatory agents, such as ICB, to fight cancer cells in the tumor’s immunosuppressive microenvironment.

6. Appendix

Abbreviation

aAPC

Artificial antigen presenting cell

ACT

Adoptive cell transfer

AML

Acute myeloid leukemia

APC

Antigen presenting cell

APC-ms

Antigen presenting cell mimetic scaffold

artTLO

Artificial lymph node-like tertiary lymphoid

aTM

Artificial T cell stimulating matrix

BMC

Bone-marrow mimicking alginate cryogel

CAR

Chimeric antigen receptor

CRS

Cytokine release syndrome

DAMP

Danger associated molecular pattern

DC

Dendritic cell

ECM

Extracellular matrix

ETC

Endogenous T cell therapy

FACS

Fluorescently-activated cell-sorting

GM-CSF

Granulocyte–macrophage colony-stimulating factor

GMP

Good manufacturing practices

HLA

Human leukocyte antigen

HSCT

Hematopoietic stem cell transplantation

ICAM-1

Intracellular Cell Adhesion Molecule 1

ICB

Immune checkpoint blockade

LFA-1

Lymphocyte Function-Associated Antigen 1

LN

Lymph node

MART-1

Melanoma Antigen Recognized by T Cells 1

MHC

Major histocompatibility complex

MOA

Mechanism of action

moDC

Monocyte-derived DC

MSN

Mesoporous silica nanoparticles

MSR

Mesoporous silica micro-rods

PAMP

Pathogen associated molecular pattern

PBAE

Poly(beta-amino ester)

PBMC

Peripheral blood mononuclear cells

PIC

Polyisocyanopeptide

PK

Pharmacokinetics

PLGA

Poly(lactic-co-glycolic acid)

pMHC

Peptide Major histocompatibility complex

PRR

Pattern recognition receptors

RES

Reticuloendothelial system

RGD

Arg-Gly-Asp

SLO

Secondary lymphoid organ

STING

Stimulator of IFN genes

TAA

Tumor associated antigens

TCR

T cell receptor

TIL

Tumor infiltrating lymphocyte