Surface-Functionalized Microgels as Artificial Antigen-Presenting Cells to Regulate Expansion of T Cells

Abstract

Artificial antigen-presenting cells (aAPCs) are currently used to manufacture T cells for adoptive therapy in cancer treatment, but a readily tunable and modular system could enable both rapid T cell expansion and control over T cell phenotype. Here, we show that microgels with tailored surface biochemical properties can serve as aAPCs to mediate T cell activation and expansion. Surface functionalization of microgels was achieved via layer-by-layer coating using oppositely charged polymers, forming a thin but dense polymer layer on the surface. This facile and versatile approach is compatible with a variety of coating polymers and allows efficient and flexible surface-specific conjugation of defined peptides or proteins. We demonstrate that tethering appropriate stimulatory ligands on the microgel surface efficiently activates T cells for polyclonal and antigen-specific expansion. The expansion, phenotype and functional outcome of primary mouse and human T cells can be regulated by modulating the concentration, ratio and distribution of stimulatory ligands presented on microgel surfaces as well as the stiffness and viscoelasticity of the microgels.

Graphical Abstract

Microgels are fabricated as artificial antigen-presenting cells to present stimulatory ligands specifically on the surface. Surface-specific functionalization is achieved via layer-by-layer coatings, forming a thin but dense polymer layer on the surface. Microgels modified with appropriate ligands promote polyclonal and antigen-specific T cell expansion and regulate T cell phenotype and function.

1. Introduction

Adoptive cell therapy (ACT) of T cells, in which isolated T cells are manipulated and expanded ex vivo before infusing into patients, has proven to be an effective treatment for certain cancers.^[1–3] The activation and expansion of T cells involves signals for T-cell receptor (TCR) stimulation and co-stimulation together with growth factors such as interleukin 2 (IL-2) to stimulate isolated T cells ex vivo, which in the body are provided by antigen-presenting cells (APC).^[4–5] Technologies that allow rapid T cell expansion and tune T cell phenotype in a controlled manner provide powerful tools to generate functional therapeutic T cells.

Biomaterials have served as artificial antigen-presenting cells (aAPCs) by locally providing required stimulatory cues for T cell activation to mimic the endogenous T cell-APC interaction and improve the therapeutic efficacy of ACT.^[6–7] Leveraging the flexible design in various material properties allows biomaterials to modulate T cell proliferation, function, and phenotype. Inorganic,^[8–9] polymeric,^[10–11] liposomal^[12–13] and lipid-modified^[14] particles conjugated with stimulatory ligands for T-cell receptor (TCR) stimulation and co-stimulation have been explored for T cell activation, and provide various advantages owing to their preparation process and physical properties.^[6] The size,^[10,15] morphology,^[16] ligand composition^[9,17] and mobility^[18–19] of aAPCs have profound effects on T cell expansion and phenotype. In addition to particle-based materials, APC mimetic scaffolds assembled from carbon nanotube bundles^[20] or lipid-coated mesoporous silica rods^[21–22] provide a 3D niche with large surface area for clustering of ligands and cells, resulting in efficient expansion of T cells. Extracellular matrix-mimetic hydrogels incorporating bioactive ligands are also capable of activating T cells and regulating their functions,^[23–25] in a manner dependent on the mechanics of the hydrogel.^[25–27] Despite the development of aAPCs, those with flexibly tunable mechanical properties are still underexplored. And considering the importance of various biochemical and physical properties in T cell activation, a biomaterial system with multiple layers of tunability is of interest for research and T cell manufacturing.

Hydrogels can be fabricated as microscale particles, also known as microgels, with tailored size, morphology and mechanics, providing a highly tunable, modular and biocompatible system.^[28] When jammed together, microgels can assemble to form granular hydrogels,^[28–29] a type of injectable microporous scaffold that has been used in 3D bioprinting,^[30] wound healing^[31] and tissue regeneration.^[32–35] Microgels enable encapsulation and release of bioactive factors in a controlled manner^[36] and exhibit mechanical properties similar to cells,^[37–38] showing potential as aAPCs and APC mimetic scaffolds. However, bioactive ligands are generally conjugated throughout the entire microgels, while only those presenting on the surface can typically bind to T cell surface receptors to regulate T cell activation. Therefore, efficient and flexible conjugation to the microgel surface of target peptides or proteins that can bind to cell surface receptors is still challenging, thus limiting their application for T cell activation.

Herein, we developed a powerful approach to fabricate microgels as aAPCs via surface functionalization of microgels using layer-by-layer coating. Sequentially adsorbing oppositely charged polymers formed a thin but dense layer on the surface with a high stability. This strategy is applicable to a variety of microgel polymers, coating polymers and allows versatile chemistry for further modification, thus providing a convenient means to modulate microgel surface properties independent of the mechanical properties. Efficient conjugation of stimulatory ligands specifically to the microgel surface promoted polyclonal and antigen-specific T cell expansion. We further demonstrate that modulating the concentration, ratio and distribution of stimulatory ligands on microgel surfaces as well as the stiffness and viscoelasticity of microgels allows control over the expansion, function and phenotype of primary mouse and human T cells.

2. Results

2.1. Synthesis of Microgels and Granular Hydrogels

Alginate microgels were fabricated using microfluidic emulsion, which provides defined size and shape by controlled droplet formation (Figure 1a). Alginate was first modified with norbornene (Alg-Nb) or tetrazine (Alg-Tz) by carbodiimide coupling to achieve an average degree of substitution (DS) of 13 or 11.5 functional groups per alginate chain respectively, as quantified by proton nuclear magnetic resonance spectra (Figure S1, S2). Stock solutions of Alg-Nb and Alg-Tz were then mixed at a final concentration of 2 wt% in the microfluidic device and injected to form microdroplets by emulsion, which then crosslinked overnight to generate microgels with a diameter of 77 ± 2 μm (Figure 1b). The elastic moduli of the microgels could be tuned by varying the ratio between Alg-Nb and Alg-Tz (Figure 1c).

Figure 1.

Fabrication and characterization of microgels and granular hydrogels. a) Schematic representation of microgel preparation using microfluidic emulsion. Alginate microgels were crosslinked by norbornene-tetrazine click chemistry. b) Phase-contrast image of alginate microgels crosslinked by norbornene-tetrazine click chemistry. Scale bar: 100 μm. c) Elastic moduli of 2 wt% alginate microgels containing different Nb/Tz ratios as measured by AFM. All the data sets are significantly different (**** P < 0.0001) except the two compared in the figure. d) Schematic representation of microgel assembly and jamming. Concentrated microgels were loaded on a membrane filter to remove continuous phase by centrifugation, which resulted in jamming of microgels. e) Void space in granular hydrogel calculated from 2D confocal slices as a function of increasing centrifugation time. In [c] and [e], values represent mean ± s.d. an ordinary one-way ANOVA with post hoc Tukey’s multiple comparisons was used. *P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001 and NS, not significant.

Alginate microgels can be jammed by centrifugation^[30–32,39] over a membrane to remove a portion of the continuous aqueous phase between particles to fabricate granular hydrogels (Figure 1d). The resulting microporous structure was visualized by incorporating 2 MDa fluorescein (FITC)-labelled dextran (Figure S3, S4). Varying the time for centrifugation modulated the porosity of the granular hydrogels in a reproducible manner, independent of the stiffness of the microgel building blocks (Figure 1e, S5). The microporous structure allows cells to penetrate through granular hydrogels of 900 μm thick following cell seeding on the surface after 2 days (Figure S6). As T cells can migrate at 10–15 μm/min,^[40] they likely had completely infiltrated the gels earlier than the 48 h observation point.

2.2. Surface Functionalization of Microgels

We next sought an efficient chemical strategy to functionalize the surface of microgels by non-covalent polymer coatings. Alginate microgels were first immersed in a solution of poly(D-lysine) (PDL) to deposit a layer of positively charged polymers and then immersed in a solution of functionalized alginate to coat the second layer and introduce functional groups on the surface (Figure 2a,b, Table S1). Both layers of polymers were uniformly and efficiently coated on the surface of microgels, and the thickness of both PDL layer and alginate coating was 0.74 ± 0.11 μm (Figure 2c, S7). The diameter of the microgels slightly decreased to 72 ± 2 μm after coating (Figure S8). This non-covalent coating showed high stability on microgel surfaces, as more than 90% of polymers remained when microgels were soaked in beads buffer (HEPES) over 3 weeks and in T cell culture media over 7 days (Figure 2d,e). Surprisingly, after soaking microgels in beads buffer at 4°C over 10 months, we still observed a thin alginate layer of 0.91 ± 0.22 μm, despite some polymer aggregation on the surface (Figure S9).

Figure 2.

Surface functionalization of microgels. a) Schematic representation of PDL and subsequent alginate coating on the surface of microgels. b) Confocal images of both PDL and alginate coatings. Scale bar: 100 μm. c) Confocal image of alginate-Rhodamine B coating on microgel surface at 100x magnification; thickness = 0.74 ± 0.11 μm. Scale bar: 20 μm. d) Change in quantity of alginate coating on microgel surface over 3 weeks in beads buffer. e) Change in quantity of alginate coating on microgel surface over 1 week in T cell culture media. f) Density of alginate polymer coated on microgel surface as a function of alginate concentration in coating solutions. g) Confocal image of azide-coated microgels labelled with Rhodamine-DBCO. Scale bar: 200 μm. h) Confocal images of alginate-Rhodamine B coated on the surface of microgels formed from hyaluronic acid, gelatin and an alginate/collagen interpenetrating network. Scale bar: 100 μm. i) Confocal images of alginate microgel presenting tetrazine functional groups coated with alginate-sulfoCy5. Microgel core in red, free tetrazine in green and polymer coating in blue. Scale bar: 20 μm. j) Quantification of fluorescent intensity of Rhodamine B (red), FITC (green) and sulfoCy5 (blue) as a function of distance from microgel surface. In [d]-[f], values represent mean ± s.d. an ordinary one-way ANOVA with post hoc Tukey’s multiple comparisons was used. *P < 0.05, **P < 0.01, ***P < 0.001 and **** P < 0.0001.

The thin and stable coating layer of alginate coating allows incorporation of sufficient ligands exclusively on the surface to mediate biological functions without introducing functional groups throughout the entire microgel that are not available to cell surface receptors. The surface ligand density can be efficiently and precisely engineered through multiple approaches during the surface functionalization process. First, the surface ligand density can be tuned by varying the DS of functional groups coupled to the alginate polymers used for coating. When increasing the DS of FITC on alginate, for example, the fluorescent intensity of alginate-FITC on the microgel surface significantly increased while the coating density remained constant (Figure S10). Second, the surface ligand density can also be easily tuned by modulating the density of coated polymers. Varying the concentration of ligand-modified alginate solution used to create the second layer from 0.01 to 1 mg/mL resulted in a 25-fold increase of polymer density without significant changes in thickness (Figure 2f, S11). Third, surface ligands can be engineered via post-functionalization using orthogonal click chemistries to conjugate the target molecules to the microgel surface. For example, microgels crosslinked via the norbornene-tetrazine strategy can be subsequently coated with azide-modified alginate, allowing surface-specific conjugation of dibenzocyclooctyne (DBCO)-modified ligands through strain-promoted azide–alkyne cycloaddition (SPAAC) in a controlled manner (Figure 2g).

The surface functionalization strategy is also applicable to a range of coating and core polymers. FITC-labelled hyaluronic acid (HA) can be uniformly coated on the surface of alginate microgels (Figure S12). In addition, core microgels made of HA, gelatin and alginate-type Ⅰ collagen interpenetrating network were fabricated using microfluidic emulsion, and a uniform and thin layer of fluorescent dye-labelled alginate was also observed on the surface of these microgels after coating, demonstrating the versatility of our approach (Figure 2h). Overall, the surface-specific chemical modification achieved via surface coating allows efficient fabrication of microgels with different surface functionalities by leveraging different polymers and chemo-selective chemistries to modify pre-synthesized microgels.

We next examined the distribution of the coating polymer in the microgels. We synthesized Rhodamine B-labelled alginate microgels containing excess tetrazine (Nb/Tz = 1/2) and used sulfo-Cy5 labelled alginate as the coating polymer. FITC-TCO was allowed to react with residue tetrazines on the microgels after coating, to detect their availability in the microgel. As expected, the fluorescent signal from all three dyes overlapped at the outer shell of the microgels (Figure 2i,j), indicating the diffusion of polymers vertical to the surface into the microgels instead of solely depositing a layer on the surface. This diffusion mechanism is further supported by the finding that polymer coatings had no significant impact on interparticle covalent crosslinking between microgels with complementary functional groups. Interparticle crosslinking between microgels with excess Tz (Nb/Tz = 1/2) and excess Nb (Nb/Tz = 2/1) in the jammed state resulted in a significant improvement of granular hydrogel stability compared to those crosslinked only by physical interactions (Figure S13). Formation of interparticle covalent crosslinking remained efficient between microgels with polymer coatings, as these granular hydrogels were stable over 3 weeks when subsequently soaked in buffer and the porosity remained similar after 3 days (Figure S13, S14), suggesting that the Nb and Tz functional moieties are exposed on surface rather than remaining underneath the polymer coatings. This phenomenon is likely attributed to the out-of-plane diffusion of charged polymers, a process commonly observed in layer-by-layer assembly approaches.^[41–42] The PDL and alginate coating likely undergo some interdiffusion, although the high molecular weight and high charge density of PDL and alginate results in low chain mobility,^[43] likely restricting their diffusion within a thin layer (< 1 μm). To test this possibility, low molecular weight PDL (1–5 kDa) was used for coating, and was found to diffuse throughout the entire microgel (Figure S15).

2.3. Primary Mouse T Cell Activation

We next leveraged this surface-specific functionalization strategy to conjugate essential antibodies for T cell activation and expansion (Figure 3a). T cell activation generally requires both T cell receptor stimulation and co-stimulation. Activating antibodies to the appropriate T cell surface receptors CD3 (αCD3) and CD28 (αCD28) were modified with DBCO groups by first reducing the disulfide linkages and then conjugating with maleimide-PEG12-DBCO via thiol-ene reaction. The amount of DBCO labeled on each antibody was approximately 4.5 on average, as measured by UV-vis spectroscopy (Figure 3b, S16). These activating antibodies were surface presented from microgels using the post-functionalization approach, by first coating alginate-azide on the surface of microgels and then reacting with DBCO-modified αCD3 and αCD28 (Figure 3a).

Figure 3.

Polyclonal and antigen-specific activation of primary mouse T cells. a) Schematic representation of modification of αCD3 and αCD28 antibodies on microgel surface by coating microgels with azide-modified alginate and conjugating antibodies using azide-DBCO click chemistry. b) UV-vis absorption spectra of unmodified αCD3, DBCO-modified αCD3 and DBCO model compounds. c) Carboxifluorescein diacetate succinimidyl ester (CFSE) histogram evaluated using FACS flow cytometry indicating the proliferation profile of stimulated CD4+ T cells. d) Percentage of proliferating CD4+ T cells when cultured with Dynabeads or microgels of different formulations. e) Representative phase contrast images of primary mouse CD4+ T cells cultured with blank microgels, microgels conjugated with anti CD3/CD28 over the entire microgel and microgels functionalized with anti CD3/CD28 on the surface as shown by representative images using phase contrast. Scale bar: 100 μm. f) Schematic representation of modification of MHC-Ⅰ and αCD28 on microgel surface. Microgels were first coated with biotin-modified alginate, reacted with streptavidin and then conjugated with ligands using biotin-streptavidin interaction. g) CFSE histogram evaluated using FACS flow cytometry indicating the proliferation profile of stimulated antigen-specific CD8+ T cells. h) Representative plots and i) quantification showing enrichment of live CD8+ cells specific for SIINFEKL peptides when mixed CD8+ T cells were cultured with Dynabeads or MHC-Ⅰ/antigen functionalized microgels. j) Fold expansion of CD8+ T cells specific for SIIFEKL peptide cultured with Dynabeads or MHC-Ⅰ functionalized microgels. In [d], [i] and [j], values represent mean ± s.d. an ordinary one-way ANOVA with post hoc Tukey’s multiple comparisons was used. *P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001 and NS, not significant.

Polyclonal T cell activation was evaluated by culture of CD4+ primary mouse T cells isolated from C57BL/6J mice with suspended microgels conjugated with αCD3 and αCD28 on the surface (surface specific) at an overall antibody density of 0.4 μg/cm², which is half of that on Dynabeads. By day 3, the proliferation rate significantly increased from 5% without antibodies to 90% in the presence of αCD3 and αCD28. This number was slightly higher than commercial CD3/CD28 T-cell expansion beads (Dynabeads), which also showed robust proliferation (Figure 3c,d). The reduction of disulfide linkages in the site-specific antibody modification may reduce the stability and bioactivity of the antibodies. Our observation of T cell proliferation suggests that modified antibodies are still functional. Massive cell clusters were observed surrounding the microgels (Figure 3e, surface-specific). When the same amount of TCO-modified αCD3 and αCD28 antibodies were directly conjugated to microgels via free tetrazines on the polymer used to fabricate the microgels (Nb/Tz = 1:2), only 13% of the CD4+ T cells were activated to proliferate (Figure 3c,d,e, entire microgel), which is attributed to a significantly reduced surface density of antibodies when conjugated throughout the entire microgel (Figure S17).

We also investigated whether our approach is applicable to antigen-specific activation and expansion of primary mouse CD8+ T cells, which is clinically relevant to cancer treatment. Microgels were first coated with biotin-modified alginate, which allows streptavidin to specifically bind to the surface due to high affinity between biotin and streptavidin (Figure 3f, S18). A biotinylated H-2K(b) MHC class I monomer presenting the peptide epitope SIINFEKL and biotinylated αCD28 were attached to the surface for antigen-specific activation of OT-1 cells. When OT-1 cells (CD8+ T cells whose T-cell receptor recognize SIINFEKL) were co-cultured with the surface-coated microgels for 3 days, a dramatic enhancement of proliferation rate was observed, compared to conditions without antibodies or with Dynabeads (Figure 3g, S19, S20).

We next assessed the antigen-specific enrichment of a subpopulation of CD8+ T cells specific to SIINFEKL peptide. To introduce SIINFEKL-specific T cells, we mixed CD8+ T cells from wild type and OT-1 mice at a ratio of 200:1 and co-cultured with the MHC class I and αCD28-presenting microgels. By day 7, the frequency of SIINFEKL-specific CD8+ T cells increased from 0.7% to 87%, corresponding to a 190-fold expansion (Figure 3h–j). By contrast, only a slight increase in frequencies of the antigen-specific subpopulation was observed when cultured with Dynabeads, as expected. To address the possibility of antigen-specific expansion directly from primary T cell isolation, we vaccinated wild type mice with ovalbumin antigen to induce endogenous antigen-specific T cells. The frequency of SIINFEKL-specific CD8+ T cells isolated from the spleen or lymph nodes 7 days after vaccination was 1.5% and 0.6%, respectively (Figure S21, S22). When CD8+ T cells isolated from vaccinated mice were co-cultured with the MHC class I (SIINFEKL) and αCD28-presenting microgels over 7 days, we also observed significant increases in the frequency of antigen-specific CD8+ T cells (Figure S21, S22).

2.4. Tuning T cell proliferation and phenotype of primary mouse T cells by modulating surface biochemical properties

The surface-specific functionalization strategy enables precise and efficient engineering of the concentrations and types of antibodies presented on the surface, thus allowing us to explore the expansion and phenotypic change of T cells in response to different presentation of cues. Raising the concentration of αCD3 and αCD28 over 2 orders of magnitude, at a αCD3/αCD28 ratio of 1, led to an increase in T cell fold expansion when CD4+ and CD8+ T cells were co-cultured with microgels (CD4/CD8 = 1) (Figure 4a, S23). CD4+ and CD8+ T cells that were expanded with higher ligand density upregulated the expression of CD25 and OX-40 activation markers (Figure S24). Commercial Dynabeads and microgels presenting antibodies at a matched density of 0.79 μg/cm² exhibited similar expansion.^[21] The ultimate CD4/CD8 ratio was also dependent on the ligand density (Figure 4b). Increasing the ligand concentration above a threshold of 0.1 μg/cm² promoted rapid and substantial CD8-biased skewing. Increasing ligand density also resulted in a reduction of CD44-CD62L+ T cells in both the CD4+ and CD8+ populations, suggesting fewer T cells associated with a naive-like phenotype but more T cells associated with central memory-like (CD44+CD62L+) and effector-like phenotypes (CD44+CD62L−) (Figure 4c). The cytotoxicity function of expanded CD8+ T cells was also evaluated using an in vitro killing assay of B16-F10 target cells presenting ovalbumin by expanded OT-I T cells. Cytotoxic function initially increased with ligand density, and then saturated (Figure 4d). When T cells were co-cultured with B16-F10 cells that don’t express ovalbumin, minimum killing was observed, as expected (Figure S25).

Figure 4.

Polyclonal mouse T cell expansion (CD4+ and CD8+ co-culture) by varying biochemical properties of microgels. a) Expansion of primary mouse T cells (inlcuding CD4+ and CD8+ T cells), b) CD4/CD8 ratio of CD4+ and CD8+ single-positive cells, c) CD44 and CD62L expression by live CD4+ (left) or CD8+ (right) T cells that were co-cultured with microgels as a function of overall surface antibody density or Dynabeads on Day 3. αCD3/ αCD28 ratio = 1, CD4/CD8 ratio = 1 on Day 0. d) Quantification of in vitro killing of OVA-expressing B16-F10 target cells by CD8+ OT-I T cells that were co-cultured with microgels as a function of overall surface antibody density. Effector/target cell ratios = 10:1. e) Expansion of primary mouse T cells (inlcuding CD4+ and CD8+ T cells), f) CD4/CD8 ratio of CD4+ and CD8+ single-positive cells, g) CD44 and CD62L expression by live CD4+ (left) or CD8+ (right) T cells that were co-cultured with microgels as a function of αCD3/ αCD28 ratio on Day 3. Overall antibody density = 0.4 μg/cm², CD4/CD8 ratio = 1 on Day 0. h) Schematic representation of a single type microgel coated with antibodies (medium purple, left) and a mixture of microgels coated with antibodies (dark purple, right) and without antibodies (light purple, right). i) Expansion of primary mouse T cells, j) CD4/CD8 ratio of CD4+ and CD8+ single-positive cells that were co-cultured with a single type microgel or mixed microgels at the same overall surface antibody density on Day 3. αCD3/ αCD28 ratio = 1, CD4/CD8 ratio = 1 on Day 0. k) Schematic representation of modification of αCD3 and αCD28 antibodies and IL-2 on microgel surface. l) Expansion of primary mouse T cells, m) CD4/CD8 ratio of CD4+ and CD8+ single-positive cells that were co-cultured with microgels as a function of IL-2 density on Day 3. Overall antibody density = 0.8 μg/cm², αCD3/ αCD28 ratio = 1, CD4/CD8 ratio = 1 on Day 0. In [d]-[f], [i], [j], [l] and [m], values represent mean ± s.d. an ordinary one-way ANOVA with post hoc Tukey’s multiple comparisons was used. *P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001 and NS, not significant.

The αCD3/αCD28 ratio was next altered. When increasing αCD3/αCD28 from 1:1 to 7:1 at an overall ligand density of 0.4 μg/cm², negligible difference was observed in fold expansion, CD4/CD8 ratio and differentiation status when CD4+ and CD8+ T cells were co-cultured with microgels (CD4/CD8 = 1) (Figure 4e–g, S26). However, decreasing the αCD3/αCD28 ratio affected the activation of T cells, leading to a reduction in fold expansion and an increasing population of T cells associated with a naive-like phenotype (CD44-CD62L+) in accordance with downregulation of CD25 and OX-40 (Figure S27). This finding aligns with the finding that TCR stimulation is required before co-stimulation for optimal T cell activation.^[44]

To explore if spatial heterogeneity of signaling would impact T cell expansion and phenotype, we combined microgels conjugated with αCD3 and αCD28 (αCD3/ αCD28 ratio = 1) and blank microgels (without antibody conjugation) (Figure 4h). The dose of antibodies in the mixed microgels conditions were tuned by varying the mixing ratios of microgels to match the dose in conditions of a single type of antibody-presenting microgel, giving the same average ligand density over the entire population of microgels. CD4+ and CD8+ T cells (CD4/CD8 = 1) cultured in mixed microgels demonstrated an increase in fold expansion and decrease in CD4/CD8 ratio, compared to cells cultured with a single type of microgel at the same dose of antibodies (Figure 4i,j). Heterogeneous ligand distribution also resulted in an upregulation of the expression of CD25 and OX-40 and a reduction in the population of naive-like T cells (Figure S28, S29). Heterogeneous distribution of the ligands provides some local areas with high density of ligands for promoting stimulation, which likely leads to enhanced T cell activation and expansion. Altogether, these results highlight the modular versatility of our approach and indicate one can leverage the flexibility of the surface-specific functionalization strategy to modulate the concentration, ratio and distribution of activating antibodies to regulate T cell expansion and the resulting phenotypes.

IL-2, the third signal for productive T cell activation, is generally supplemented as soluble factors in the culture medium, but we hypothesized that DBCO-modified cytokines could also be immobilized on the microgel surface to regulate T cell activation. IL-2 was first functionalized with DBCO by reacting with DBCO-sulfo-NHS via aminolysis, and conjugated to coated microgels together with αCD3/αCD28 (Figure 4k). After multiple washing steps to remove unreacted IL-2, minimum IL-2 release was detected. When cultured with CD4+ and CD8+ T cells (CD4/CD8 = 1), IL-2 immobilized on the surface of microgels supported T cell activation, with increasing IL-2 density resulting in greater fold expansion (Figure 4l, S30, S31). Interestingly, the low density of immobilized IL-2 resulted in a CD4-biased expansion, while a high density promoted a substantial CD8-biased skewing (Figure 4m). At low level of IL-2, an increase of FOXP3+CD25+ T cells was observed (Figure S32). As regulatory T cells express a high affinity IL-2 receptor,^[45] the low level of IL-2 likely enriched these cells, skewing the CD4/CD8 ratio.^[46] Immobilizing IL-2 at a surface density of 2000 U/cm² led to similar activation of the T cells compared to supplementing IL-2 in the media, with minimum differences in T cell expansion and CD4/CD8 ratio (Figure 4l,m). Immobilizing cytokines on the surface of scaffolds can potentially provide a convenient means to locally regulate T cell activation in vivo and minimize non-targeted cytokine release to avoid unexpected side effects.^[47–49] Our method allows precise engineering of IL-2 density on the surface to mediate the expansion and phenotypes of T cells, suggesting a strategy to specifically expand regulatory T cells when desired, and can also potentially capture subsequently administered DBCO-modified cytokines for presentation in a time-dependent manner.

To examine the impact of CD4+ cells on CD8+ cell expansion, we performed co-culture studies or with CD8+ only when activated by microgels. The presence of CD4+ cells resulted in an enhanced expansion of CD8+ cells, an upregulation of activation markers, and a reduction in naïve-like population (Figure S33–35). However, no significant difference was observed in the expression of IFNγ, TNFα and IL-2, suggesting that the T cell products were functionally similar (Figure S36).

2.5. Regulating T cell proliferation and phenotype of primary mouse T cells by modulating microgel mechanical properties

The surface-specific functionalization strategy also enables modulation of the mechanical properties of the microgels independent of their surface biochemical properties. Microgels with different stiffness were synthesized by varying the ratio between Alg-Nb and Alg-Tz at an overall alginate concentration of 2 wt% and functionalized with stimulatory antibodies on the surface. When microgels of different stiffness were co-cultured with CD4+ and CD8+ T cells (CD4/CD8 = 1), increasing the elastic moduli of microgels from 1kPa to 3 kPa led to an increase in T cell fold expansion and upregulation of the expression of CD25 and OX-40 activation markers (Figure 5a, S37). Increasing the stiffness also resulted in a substantial CD8-biased skewing (Figure S38). A reduction of CD44-CD62L+ T cells was observed in both the CD4+ and CD8+ populations, suggesting fewer T cells associated with naïve-like phenotype (Figure 5b).

Figure 5.

Polyclonal mouse T cell expansion (CD4+ and CD8+ co-culture) while varying the physical properties of microgels. a) Expansion of primary mouse T cells and b) CD44/CD62L expression by live CD4+ (left) or CD8+ (right) T cells that were co-cultured with microgels as a function of stiffness for 3 days. c) Expansion of primary mouse T cells and d) CD44/CD62L expression by live CD4+ (left) or CD8+ (right) T cells that were co-cultured with elastic or viscoelastic microgels for 3 days. Overall antibody density = 0.4 μg/cm², αCD3/ αCD28 ratio = 1 in all studies.

The influence of the viscoelasticity of the microgels on T cell activation was next studied. Ionically crosslinked alginate gels exhibit viscoelastic features^[50–52] and are used here as viscoelastic scaffolds for 3D cell culture. We synthesized microgels with calcium crosslinking, in place of covalent crosslinking. Alginate stock solution was first combined with calcium-ethylenediaminetetraacetic acid (EDTA) at neutral pH and then injected into the microfluidic device to mix with acid-containing oil. The binding affinity between EDTA and calcium decreased significantly from pH 7 to pH 4, thus releasing free calcium ions to crosslink the hydrogels when forming the microdroplets. The diameter of the microgels was 79 ± 2 μm and the elastic modulus was 3 kPa, similar to elastic microgels used for T cell activation (Figure S39, S40).

Compared to elastic microgels with the same surface ligand density, viscoelastic microgels modified with αCD3/αCD28 showed a reduction in T cell expansion (Figure 5c), similar to previously reported results in which T cells were activated in 3D collagen matrices of different viscoelasticity.^[26] Viscoelastic microgels also led to the downregulation of the expression of CD25 and OX-40 activation markers and an increase of CD44-CD62L+ T cells, suggesting a less differentiated phenotype (Figure 5d, S41). In addition, no significant difference in CD4/CD8 ratio was observed (Figure S42).

To investigate if the size of the microgels affects T cell activation, we synthesized covalently crosslinked microgels with diameter of 18 ± 2 μm (Figure S43), and functionalized with αCD3/αCD28 antibodies. While maintaining consistent surface density and dose of antibodies, decreasing microgel size resulted in a small increase in fold expansion and a reduction in CD4/CD8 ratio (Figure S44, S45). Reduced microgel size also led to an upregulation of the expression of CD25 and OX-40 and a reduction in the population of T cells associated with a naive-like phenotype (Figure S46, S47).

We also examined the capability of our strategy for T cell activation in 2D culture. 96-well tissue culture plates were coated with PDL and alginate, and then functionalized with stimulatory antibodies at different densities. Similar to microgels, increasing ligand density on the plate resulted in an increase in T cell expansion (Figure S48). However, a threshold of 0.2 ug/cm² was required for activation, which is significantly higher than found with 3D microgels, and the expansion plateaued at 0.8 ug/cm². Overall, the regime of ligand density that has tunable impact on T cell phenotypes was much narrower for coated plates compared to coated microgels (Figure S48–S51). More generally, compared to 2D plate culture, 3D culture provides higher achievable cell densities and more tunable properties, which is greatly preferable for T cell manufacturing.

2.6. Polyclonal expansion of primary human T cells

Finally, we investigated whether microgels could be used for polyclonal expansion of primary human T cells. Primary human T cells (mixture of CD4+ and CD8+) were isolated from blood samples from two healthy donors and cultured with microgels presenting human αCD3 and αCD28. Higher ligand densities initially led to an increase in fold expansion in cells from Donor #1, similar to the trend observed in mouse T cells, but expansion then diminished at ligand densities greater than 0.4 μg/cm² (Figure 6a). The trend in fold expansion agreed with the expression of CD25 (Figure 6b). The reduction of fold expansion at high ligand density is likely attributed to activation induced cell death (AICD), in which over activation results in apoptosis.^[53] Donor #1 cells initially had a CD4/CD8 ratio of 1.9 (Figure 6c), and when activated with hydrogels presenting a ligand density below 0.4 μg/cm², the resulting T cell population was biased towards the CD8+ population. Increasing the ligand density above 0.4 μg/cm² resulted in a more balanced CD4+ and CD8+ ratio. This phenomenon may be caused by the different sensitivities of CD4+ and CD8+ T cells to activating antibodies. The expanded T cells were also phenotypically different as the density of ligands presenting on hydrogels was varied. A high ligand density resulted in a reduction of CD45RA+CCR7+ T cells, suggesting a more differentiated phenotype in both CD4+ and CD8+ T cells (Figure 6d). A similar trend was also observed in the expression of CD45RA and CD62L (Figure S52). T cells exposed to high density of ligands also expressed more CD39+, suggesting a more effector-like phenotype (Figure 6e). In addition to differentiation status, the expression of inhibitory markers was also affected by the surface ligand density (Figure S53). PD-1 and Lag-3 expression among both CD4+ and CD8+ T cells first progressively increased with ligand density, suggesting a more exhausted phenotype. However, increasing the ligand density above 0.4 μg/cm² corresponded with a slightly lower level of PD-1 and Lag-3 expression. Expression of these inhibitory markers suggests that persistent stimulation of T cells at high does will lead to exhaustion over time, and ultimately T cell dysfunction. A potential strategy to balance proliferation and exhaustion is to limit the timeframe of stimulation using degradable materials.

Figure 6.

Polyclonal human T cell expansion (mixture of CD4+ and CD8+) by varying biochemical properties of microgels. a) Expansion of primary human T cells that were co-cultured with microgels as a function of overall surface antibody density on Day 6. b) CD25 expression by live CD4+ (left) or CD8+ (right) T cells that were co-cultured with microgels as a function of overall surface antibody density on Day 6. c) CD4/CD8 ratio of cells cultured with microgels as a function of overall surface antibody density on Day 6. d) CD45RA and CCR7 expression, e) CD39 expression by live CD4+ (left) or CD8+ (right) T cells that were cultured with microgels as a function of overall surface antibody density on Day 6.

To examine whether the T cell response would be consistent between different donors, we analyzed T cells from Donor #2 after activation with antibody-presenting hydrogels. Similar trends of expansion and phenotypic changes from cells of the two donors were observed (Figure S54–S60). However, a difference was found in activation potential and sensitivity between donors. For example, the proliferation for Donor #1 peaked at 0.4 μg/cm², while the largest expansion for Donor #2 occured at 1.6 μg/cm² (Figure 6a, S54). Donor-to-donor variability due to different health conditions, genders and ages is an important factor affecting how T cells respond to stimulation.^[22,54–56] The facile functionalization strategy developed here for microgels could readily be used to provide patient-specific stimulation to accommodate individual differences in the donor cells.

3. Conclusion

Here we demonstrate a microgel platform that presents bioactive ligands specifically on the surface to regulate T cell expansion and phenotypic change. Surface functionalization was achieved by coating the microgel surface with oppositely charged polymers, resulting in a thin yet stable layer of functional polymers decorating the surface of microgels. Conjugation of activating antibodies and mitogenic cytokines via chemo-selective chemistry allows one to modulate the surface biochemical cues to T cells precisely and efficiently. Microgels modified with appropriate ligands promoted efficient polyclonal and antigen-specific T cell expansion. Our findings demonstrate that the concentration, ratio and distribution of antibodies during T-cell activation have profound effects on the resulting phenotype of primary mouse and human T cells. In addition, stiffer and more elastic microgels promote the expansion and activation of the T cells.

This surface-specific functionalization strategy provides a convenient and versatile means to modulate the surface biochemical properties of microgels, which could be exploited to manipulate the stimulation dose for personalized T cell therapies.^[22] The ready injectability of microgels and granular hydrogels^[30–32,39] and stability of polymer coatings during injection (Figure S61) potentially could also allow these materials to be delivered with minimally invasive procedures in the future for in situ expansion of immune cells for cancer treatment, minimizing the risks of off-target toxicities. Quantitative analysis of the stability of the coating will need to be further analyzed if future studies use needle injection of these microgels. Considering the flexibility of microgels in other aspects, including their physical properties and ligand mobility, these can also be leveraged to investigate the impact of other important material properties on T cell expansion and phenotypic regulation, and used for the expansion of a variety of cell types.

4. Experimental Section

Elastic Microgel Preparation.

The synthesis of covalently crosslinked microgels was adapted from an existing protocol.^[57] The dispersed phase containing 2 wt% alginate was prepared as a mixture of Tz and Nb modified alginate dissolved separately at 1–3 wt% in DI water. A mixture of fluorosurfactant (1%) in fluorocarbon oil was used as the continuous phase. Alginate-Tz and alginate-Nb solutions were injected at 150 μL/h and the continuous phase was injected at 1000 μL/h. The emulsion was then collected in a tube and left at room temperature for 24 h to allow covalent crosslinking between alginate polymers. After the reaction was complete, the continuous phase was removed, and 33% 1H,1H,2H,2H-perfluoro-1-octanol in HFE was added in excess at 1:3 volume ratio to the collected microgels to break the emulsion. Finally, microgels were washed three times with beads buffer (130 mM NaCl, 25 mM HEPES, 2 mM CaCl₂, pH 7.5), redispersed in beads buffer and stored at 4°C until further use.

Viscoelastic Microgel Preparation.

The synthesis of Ca²⁺ crosslinked alginate microgels was adapted from an existing protocol.^[58] The dispersed phase containing 1–2 wt% unmodified alginate and 50 mM CaEDTA was identically prepared as described above for batch emulsion technique. A mixture of fluorosurfactant (1%) and acidic acid (0.05–0.2 v%) in fluorocarbon oil was used as the continuous phase. Alginate solution and the continuous phase were injected at flow rates of 300 and 1000 μL/h, respectively. The emulsion was then collected and mixed with 50% 1H,1H,2H,2H-perfluoro-1-octanol in HFE at 1:1 volume ratio to break the emulsion. Microgels were washed three times with beads buffer, redispersed in beads buffer and stored at 4°C until further use.

Polymer Coating.

Microgels were first concentrated by centrifugation at 300 rcf for 3 min and redispersed in a solution of poly(D-lysine) (PDL) (50–150 kDa, 0.1 mg/mL in beads buffer) at a concentration of 4*10⁵ microgels/mL. Microgels were then immediately concentrated by centrifugation at 300 rcf for 3 min, washed three times with beads buffer, redispersed in beads buffer at initial stock solution concentration, and stored at 4°C until further use. Microgels were redispersed in a solution of functionalized alginate (0.01 – 1 mg/mL) in beads buffer at a concentration of 4*10⁵ microgels/mL and collected by centrifugation at 300 rcf for 3 min. Microgels were washed three times with beads buffer, redispersed in beads buffer at initial stock solution concentration, and stored at 4°C until further use.

Coating Density.

The amount of alginate coated on the microgels was determined by the difference between the amount of alginate used for coating and the remaining amount in solution after coating. Alginate-Rhodamine B was used as a model polymer for coating to quantify the concentration of alginate in solutions. Microgels were washed three times after coating and all the supernatants were collected after each centrifugation. Alginate concentration in the original solution used for coating and all the supernatants were quantified by fluorescent intensity at 586 nm (excitation wavelength 561 nm) based on a calibration curve. The density of coating was calculated by the amount of alginate-Rhodamine B coated on the surface and overall surface area of microgels. Three independent experimental replicates were used for all experiments.

Coating Stability.

The stability of polymer coating was determined by polymer dissolution in the surrounding buffer solution. Alginate-Rhodamine B coated microgels were soaked in beads buffer (4*10⁵ microgels/mL) at room temperature. All buffer solutions were collected and replaced by fresh beads buffer on Day 1, 4, 7, 10, 14, 21. The concentration of released alginate-Rhodamine B was determined as described before. Three independent experimental replicates were used for all experiments.

Microgel Jamming.

Microgels were first concentrated by centrifugation at 300 rcf for 3 min, if applicable, mixed with the desired complementary microgels collected separately. A pre-rinsed membrane (0.22 μm) was folded into a cone shape and placed in a 1.5 mL eppendorf tube. The pellet was then loaded onto the membrane and centrifuged at 50 rcf for 20, 5 or 1 s. The jammed microgels were retrieved from the membrane and placed between two glass slides with spacers until the assembly was completed.

Porosity.

Characterization of porosity was adapted from a previously reported method based on the fluorescence of void space between particles.^[59] Briefly, a labeling solution for the interstitial space was prepared by dissolving FITC-dextran (2 MDa) in beads buffer at 40 μg/mL. Microgels were dispersed in the FITC-dextran solution before jamming. Granular hydrogels were imaged using an inverted confocal microscope (LSM 700 Confocal Microscope) and post-processed with ImageJ to analyze the pores. Thresholding was based on the Triangle or Huang algorithm to binarize the stacks and the size range in the Analyze Particles function was set to 5 μm² to infinity. The resulting %area was averaged over all stacks to obtain the porosity of the sample.

Atomic force microscopy (AFM).

The elastic modulus of microgels were measured using AFM as previously described.^[60] The nanoindentation tests were conducted on a NanoWizard II AFM (JPK Instruments AG). Silicone cantilevers with a polystyrene tip, a force constant of 0.2 N/m, and a resonance frequency of 13 kHz were used (NanoAndMore GmbH, Watsonville, CA, USA) for the measurements. The contact force was set to 0.1 V, and the pulling range was set from 1500 to 3000 nm. 4,096 force-distance curves in 20 × 20 μm area were recorded and calculated to give the elastic modulus.

Antibody modification.

αCD3 or αCD28 antibodies were modified with DBCO by reducing the disulfide linkage using TCEP-HCl (1:30 molar ratio) and then reacting with DBCO-PEG12-maleimide (Conju Probe, 1:60 molar ratio) at 4°C overnight. The mixture was purified using desalting column (3kDa), washed 7 times with 1X PBS and stored at 4°C before further use. The degree of modification on antibodies was quantified via UV-vis spectroscopy using nanodrop. The absorbance of unmodified antibody and DBCO-PEG12-maleimide at 280 nm and 310 nm was measured and plotted versus the concentration to obtain the standard calibration curve and extinction coefficient. The concentration of DBCO conjugated to antibodies was quantified based on the calibration curve of DBCO-PEG12-maleimide absorption at 310 nm. The concentration of antibodies after modification was quantified based on the calibration curve of antibody absorption at 280 nm after subtracting the absorbance from DBCO. The number of DBCO per antibody = concentration of DBCO/concentration of antibody, indicating an average of 4.5 DBCO groups on each antibody.

Antibody conjugation.

For polyclonal expansion, alginate microgels coated with alginate-azide were mixed with DBCO-modified αCD3 and αCD28 at 4°C overnight. Microgels were washed three times with beads buffer, soaked in T cell media at 4°C overnight and washed three times with T cell media to remove physically absorbed antibodies. The conjugation efficiency was quantified by measuring the amount of unreacted antibodies in the buffer and culture media using nanodrop during reaction and washing steps. For antigen-specific expansion, microgels coated with biotin-modified alginate were mixed with streptavidin at room temperature for 1h at a biotin/streptavidin ratio of 1. Then biotinylated H-2K(b) MHC class I monomer presenting SIINFEKL peptide and biotinylated αCD28 were added to react for 1 h at room temperature. Microgels were washed three times with beads buffer, soaked in T cell media at 4°C overnight and washed three times with T cell media to remove physically absorbed antibodies.

Primary mouse T cell isolation.

C57BL/6J mice were used for polyclonal T cell expansion studies and C57BL/6-Tg(TcraTcrb)1100Mjb/J (OT-I) mice were used for antigen-specific T cell expansion studies and cytotoxic function analysis. Primary mouse T cells were obtained from the spleen and isolated using CD4+ or CD8a+ T cell isolation MACS kits (Miltenyi Biotec) to obtain CD4+ T cells or CD8+ T cells. Mouse T cells were cultured in T cell media (RPMI 1640 supplemented with 10% HI-FBS, 2 mM L-glutamine, 1 mM sodium pyruvate, 50 μM beta-mercaptoethanol, 0.1 mM non-essential amino acids, 1 mM sodium pyruvate, 10 mM HEPES, and 1% penicillin-streptomycin) supplemented with 100 U/ml recombinant mouse IL-2. All procedures involving animals were done in compliance with National Institutes of Health and Institutional guidelines with approval of Harvard University’s Institutional Animal Care and Use Committee. Animals were purchased from The Jackson Laboratory, female and between 6 and 9 weeks old. Animals were maintained on 12 h light cycles and fed chow and water ad libitum.

Vaccination against ovalbumin.

Vaccines were prepared following protocols published previously.^[61–62] In brief, mesoporous silica rods (MSRs) were suspended in sterile DPBS at 50 mg/mL. 2mg of MSRs were incubated with 200μg ovalbumin (InvivoGen) and additional 2mg were incubated with 100ug CpG-ODN 1826 (5′-TCCATGACGTTCCTGACGTT-3′) (Integrated DNA Technologies, Chicago, IL). The suspensions were gently shaken at room temperature for 7h, flash-frozen, and lyophilized overnight. The following day, a separate 1mg aliquot of MSR suspension was mixed with 1μg granulocyte-macrophage colony-stimulation factor (GM-CSF, Peprotech) and shaken for 1h at 37°C. The three MSR suspensions were combined into one single 150μL shot by adding sterile DI water, and injected through an 18G needle into the left flank of two C57BL6/J mice (The Jackson Laboratory).

Primary mouse T cell isolation from lymph nodes.

Vaccinated mice were euthanized 7 days post-vaccination. Their ipsilateral draining inguinal, axillary, and brachial lymph nodes and their contralateral lymph nodes were harvested into 4mL RPMI containing 10% FBS, 150 U/mL collagenase IV (Thermo Fisher Scientific Inc.), and 0.1ug/mL DNAse (F. Hoffmann-La Roche AG). The lymph nodes were dissociated using a GentleMACS Tissue Dissociator (Miltenyi Biotec) and incubated at 37°C for 30 min under mild agitation. Subsequently, the tissues were strained through a 40μm strainer, washed with PBS twice. T cells were isolated using CD4+ or CD8a+ T cell isolation MACS kits (Miltenyi Biotec).

Primary human T cell isolation.

Human healthy de-identified blood collars were obtained from Brigham’s and Woman’s Hospital, processed in a Ficoll gradient to obtain enriched peripheral blood mononuclear cells (PBMCs) and frozen prior to use. Primary human T cells were isolated from PBMCs using the human pan-T cell isolation kit (Miltenyi Biotec) to obtain a mixture of CD4+ and CD8+ T cells. Human T cells were cultured in T cell media (RPMI 1640 supplemented with 10% HI-FBS, 2 mM L-glutamine, 1 mM sodium pyruvate, 50 μM beta-mercaptoethanol, 0.1 mM non-essential amino acids, 1 mM sodium pyruvate, 10 mM HEPES, and 1% penicillin-streptomycin) supplemented with 30 U/ml recombinant human IL-2.

T Cell Proliferation Assay.

The isolated CD4+ or CD8+ cells were pre-labelled with 5 μM CellTrace yellow (ThermoFisher Scientific) at 37°C for 15 minutes. After PBS washing, the CellTrace-labelled CD4+ cells were mixed with activation stimuli (i.e., Dynabeads or microgels), and seeded at a density of 10⁵ cells/well. Microgels were seeded at a density of 10⁴/well in suspension. Commercial Dynabeads (ThermoFisher Scientific) were used according to the manufacturer-optimized protocol included with the kit at a cell/Dynabeads ratio of 1. After 3 d, the CellTrace yellow fluorescence was measured using an Aurora Spectral Analyzer (Cytek). Microscopic Images were taken using EVOS FL microscope.

Polyclonal T cell Expansion Studies.

A mixture of isolated primary mouse or human CD4+ and CD8+ T cells (CD4/CD8 ratio = 1) were mixed with activation stimuli and cultured for 3 d as described above. Media was added to maintain the cells below a density of 2.5×10⁶ cells/mL throughout the culture period. Fold expansion was calculated by dividing the number of cells at the respective time point by the number of cells seeded at the start of culture.

T Cell Phenotypic Analysis.

T cell phenotype was evaluated by using flow cytometry (Cytek Aurora). Gates were set using fluorescence minus one (FMO) controls. Data was analyzed using FCS express flow cytometry software. Anti-mouse antibodies for flow cytometry were obtained from BioLegend and ThermoFisher: CD4-BV785 (RM4–5), CD8a-eFluor450 (53.6.7), CD62L-BV510 (MEL-14), CD44-FITC (IM7), PD-1-PE/Dazzle (RMP1–30), Lag-3-PE (C9B7W), CD25-APC (PC61), OX-40-PE/Cy7 (OX-86), Live/Dead (Fixable blue dead stain). Anti-human antibodies for flow cytometry were obtained from BioLegend: CD4− PerCP (SK3), CD8− APC/Cyanine7 (SK1), CD45RA− PE/Cyanine7 (HI100), CD62L-BV510 (DREG-56), CCR7-FITC (G043H7), CD25− PE/Dazzle (M-A251), CD39-BV711 (A1), PD-1-BV421 (EH12.2H7), Lag-3-APC (7H2C65), CD127-PE (A019D5).

Cytotoxic Function.

The B16-F10 murine melanoma cells and B16-F10 expressing ovalbumin were obtained from American Type Culture Collection (ATCC) and expanded subconfluently in growth medium consisting of 10% fetal bovine serum, 1% penicillin/streptomycin in high-glucose Dulbecco’s Modified Eagle media (DMEM). Cells were passaged at 80% confluency and used at passage 10 or lower for all experiments. In killing assays, B16-F10 expressing ovalbumin were incubated with 1 μg/mL Calcein AM (Invitrogen) for 30 min at 37 °C, and then pulsed with 2 μg/ml SIINFEKL peptide for 60 min at 37 °C. T cells and B16-F10 were mixed at a ratio of 10:1 and co-cultured for 4 h. Cells were pelleted and fluorescence intensity of supernatant samples were quantified using a plate reader. For B16-F10 without expressing ovalbumin, cells were incubated with Calcein AM and directly mixed with T cells for co-culture.

Statistical Analysis.

All statistical analysis was performed in R studio (version 2022.7.0.548). All levels of significance for differences observed between groups were evaluated using ANOVA one-way, followed by Tukey Honest Significant Differences test. Standard deviation is illustrated by error bars, and significance levels are stated as follows: * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.