Antigen-conjugated scaffolds enable sustained delivery of antigen and enrichment of antigen-specific T-cells

Abstract

A thorough understanding of T-cell dynamics and interactions could improve patient care in autoimmunity, cancer immunotherapy, and myriad other conditions, yet monitoring antigen-specific T-cell clones is challenging. T-cells recognize antigens presented by antigen-presenting cells (APCs) in the context of major histocompatibility complexes (MHCs). Specific T-cell clones are rare in the blood (<1 in 100,000), and thus cell expansion which consequently alters cell phenotype and function is typically necessary before analysis. This motivates the development of new methods for enriching T-cell populations of interest without phenotypically altering them. Recent work has demonstrated that implantable biomaterial systems can recruit disease-relevant cells in autoimmune conditions, and that if antigens are present, antigen-specific T-cells become enriched in these materials. To date, antigen-loaded materials have exhibited uncontrolled loading, burst release, and subsequent T-cell exhaustion. In this report, we engineer a novel biomaterial antigen delivery system by conjugating antigens to the polymer backbone prior to porous scaffold fabrication. We demonstrate that this technique enables precise antigen loading via ratiometric mixing of modified and unmodified polymer. We show controlled release of antigen into the microenvironment and demonstrate that released antigen is processed and presented by APCs. Using this fabrication method, we achieve sustained release of peptide antigens over a period of 3 weeks in vitro. When implanted in healthy mice, these antigen-conjugated scaffolds are invaded by host myeloid and lymphoid cells and exhibit a dose-dependent enrichment of systemically circulating antigen-specific T-cell populations, while avoiding significant T-cell exhaustion. Finally, we apply this system to an autoantigen from multiple sclerosis (MS) and show release and interaction with autoantigen-specific T-cells. Using this technique, disease-relevant T-cells can be recruited for diagnostic assessment or for immunological research. Future work will investigate the potential of these systems to monitor disease onset and progression in vivo, co-deliver multiple antigens for assessment of epitope spreading, therapeutically target disease-relevant cells within a local niche in situ, and expand the platform for controlled delivery of therapeutic peptides in models beyond autoimmunity.

1. Introduction

T-cells are critical to immune functions, from protecting against pathogens to inducing tissue repair and regenerative responses. Most mature T-cells exist as clonal populations expressing a single T-cell receptor that is specific to a unique antigen presented in major histocompatibility complexes (MHCs), but our ability to study these specific T-cell clones is limited by a lack of adequate tools. Blood draws are the gold-standard for collecting immune cells, and while easy to perform, cannot be easily used to study T-cell clones of interest due to the rarity of disease-specific T-cell populations in circulation [1-3]. Improved tools to isolate and study specific T-cells would be a boon to our understanding of how these cells orchestrate tissue repair and contribute to the development of therapeutics for cancers, vaccines, and autoimmunity [4]. Most tools designed to address the challenges associated with limited availability of T-cells in blood samples focus on ex vivo expansion of heterogeneous T-cell populations and subsequent selection of the clones of interest for downstream analysis [5]. However, ex vivo expansion of functional T-cells is complicated and requires the recreation of signals that would normally be provided by antigen-presenting cells in vivo, such as T-cell receptor (TCR) stimulation, costimulatory signals, and pro-survival cytokines [6]. While ex vivo culture systems can provide necessary signals for T-cell activation, alterations in dynamics of stimuli from the in vivo setting result in slow expansion rates, non-functioning T-cells, and costly manufacturing procedures [7-10].

An alternative approach to capturing immune cells locally involves harnessing the host response to implanted biomaterials to drive immune cell recruitment to the site of implant [11-14]. Recent work has shown that the foreign body response (FBR) to implanted biomaterials causes immune cell invasion, stromal cell infiltration, and vascularization to form an immunological niche (IN). Following subcutaneous implantation, the IN is readily accessible, providing a minimally-invasive location for enhanced disease detection and monitoring via imaging, cell collection, and biomarker identification [15,16]. Prior studies have applied this system to an experimental autoimmune encephalomyelitis (EAE) mouse model of MS, wherein the IN demonstrated utility as a platform for studying immune cells involved in disease pathogenesis [17].

INs however, are composed primarily of innate immune cells, and thus do not capture every aspect of disease. By incorporating antigens into these biomaterial scaffolds, there is a potential to increase recruitment of antigen-specific cells at the site of implantation [1,2]. Peptide antigens are an attractive option due to their low cost, stability in harsh conditions, and ability to be modified to complement conjugation chemistries of interest [18-21]. However, to date, antigen-loaded materials have exhibited uncontrolled loading, burst release, and sometimes massive T-cell proliferation, exhaustion, and activation-induced cell death (AICD) [22]. To ameliorate these drawbacks, we hypothesize that we can chemically conjugate peptide antigens to the poly(lactide-co-glycolide) (PLG) backbone prior to scaffold fabrication, allowing for precise loading of antigen in each individual scaffold and avoiding burst release and systemic activation.

In this report, we investigate the development of a novel antigen-conjugated scaffold platform for localized antigen-specific CD4 T-cell enrichment. We utilized carbodiimide crosslinking chemistry to chemically conjugate a variety of ovalbumin (OVA) antigen epitopes to the PLG backbone, from which we successfully fabricated porous scaffolds. Using this methodology, we fabricated scaffolds loaded with precise concentrations of antigen that allow for sustained antigen delivery for three weeks in vitro. Released peptide antigens were processed and presented by APCs, activating and inducing proliferation in antigen-specific T-cell populations. When implanted in mice in vivo, antigen-conjugated scaffolds recruited myeloid and lymphoid cell populations and exhibited a dose-dependent enrichment of antigen-specific CD4 T-cells. Additionally, CD4 T-cells in antigen-conjugated and blank scaffolds exhibited similar expression of exhaustion markers, showcasing our ability to fine tune antigen delivery and avoid massive T-cell exhaustion. To further demonstrate the modularity of this platform, we fabricated scaffolds conjugated with PLP_139–151, an MS autoantigen, and demonstrated the ability of these implants to activate disease-specific T-cells in vitro. Overall, we have demonstrated the development of a tunable antigen-conjugated scaffold, a modular tool for investigating rare T-cell populations, and adaptive-innate crosstalk that could be harnessed in a variety of inflammatory diseases, autoimmunity, cancers, and other pathologies.

2. Materials and methods

2.1. Reagents

Peptides (OVA_257–264 (SIINFEKL), OVA_321–339 (LKISQAVHAAHAEI-NEAGR), OVA_323–339 (ISQAVHAAHAEINEAGR), and PLP_139–151 (HCLGKWLGHPDKF)) were custom synthesized by Genscript with C-terminal amidation. For simplicity, all peptides will be referred to by the numerical amino acid positions within each protein (e.g. OVA_257–254) rather than the specific amino acid sequence (e.g. SIINFEKL). High molecular weight (MW) ester terminated 75:25 PLG with a MW of 75–85 kDa (Cat # AP125) and low MW acid terminated 75:25 PLG with a MW of 5–10 kDa (Cat # AP073) were purchased from PolySciTech.

2.2. Fabrication of antigen-conjugated scaffolds

2.2.1. Peptide-antigen conjugation

Carbodiimide crosslinking chemistry was used to modify 5–10 kDa 75:25 PLG and generate antigen-conjugated polymers in a similar manner to previously published reports [23]. Briefly, PLG was dissolved in dimethyl sulfoxide (DMSO) at a total polymer concentration of 2 % weight per volume (w/v) and set to stir. Reactions to PLG were calculated as molar ratios to PLG, hereafter denoted as “x to PLG”. 1-Ethyl-3-(dimethylaminopropyl)carbodiimide (EDC, 10× to PLG) (Sigma) was dissolved in DMSO at 2 % w/v and added dropwise to the stirring PLG solution. N-hydroxysuccinimide (NHS, 10× to PLG) (Sigma) was dissolved in DMSO at 1 % w/v and added dropwise to the stirring PLG solution. The reaction was allowed to occur for 15 min before proceeding. Peptide antigen (2× to PLG) was dissolved in DMSO at 2 % w/v. Triethylamine (TEA, 5× to PLG) (Sigma) was added dropwise to the peptide antigen solution. The peptide antigen and TEA solution was added dropwise to the stirring PLG solution. The reaction was allowed to proceed overnight at room temperature. The resulting antigen-conjugated polymer was purified via dialysis using 3500 MW cutoff snakeskin tubing (Fisher) against 3.5 L of distilled water over 8 h. Distilled water was replaced a total of 7 times. The resulting purified polymer was collected, frozen, and lyophilized overnight. Antigen coupling efficiency to PLG was determined via ¹H NMR and Diffusion Ordered Spectroscopy (DOSY) in DMSO-d₆ at the University of Michigan BioNMR core. The coupling efficiency of OVA_323–339 to PLG was calculated by comparing the integration values of the overlapping valine and isoleucine proton peaks present at 0.8 ppm in OVA_323–339 to the methylene proton peak present at 5.2 ppm in PLG [23]. The coupling efficiency of OVA_257–264 to PLG was calculated by comparing the integration values of the overlapping leucine and isoleucine proton peaks present at 0.8 ppm in OVA_257–264 to the methylene proton peak present at 5.2 ppm in PLG. The coupling efficiency of OVA_321–339 to PLG was calculated by comparing the integration values of the overlapping valine, leucine, and isoleucine proton peaks present at 0.8 ppm in OVA_321–339 to the methylene proton peak present at 5.2 ppm in PLG. The coupling efficiency of PLP_139–151 to PLG was calculated by comparing the integration values of the leucine peak present at 0.8 ppm in PLP_139–151 to the methylene peak present at 5.2 ppm in PLG.

2.2.2. Peptide conjugated scaffold fabrication

Antigen-conjugated scaffolds were fabricated using solvent casting and salt leaching. Ratios of low MW conjugated PLG were combined with high MW unconjugated PLG and dissolved in chloroform at a total polymer concentration of 13 % w/v. For blank scaffolds, 10 % of the total PLG mass consisted of low MW PLG, with the rest being high MW PLG. The polymer mixture was combined with 250–425 μm sieved sodium chloride (NaCl) at a ratio of 3.9 g NaCl/mL and cast into a cylindrical mold overnight to allow for solvent evaporation. The resulting disks (41.5 mm diameter and 3 mm thick) were leached in distilled water over the course of 8 h, with water changes every hour. The resulting bulk porous scaffolds were punched into individual scaffolds using a 5 mm biopsy punch to generate cylindrical scaffolds sized for all subsequent studies (5 mm diameter and 3 mm thick). Scaffolds were lyophilized and stored at −80 °C until use. Scaffold porosity and pore volume were calculated as previously described [16], with polymer density estimated to be similar to another commercially-available product (Polysciences, Cat # 26780–50-7). For in vitro cell culture and in vivo studies, scaffolds were sterilized in ethanol for one minute, followed by three sterile water washes for one minute each prior to use.

2.2.3. Peptide encapsulated and adsorbed scaffold fabrication

Antigen-encapsulated scaffolds were fabricated using solvent casting and salt leaching techniques as described above for blank scaffolds, with a few changes. Peptide antigen was first dissolved in dimethyl sulfoxide (DMSO) at 20 mg/mL, then added into the polymer chloroform solution to reach a total polymer concentration of 13 % w/v and antigen concentration of 4 μg antigen/mg PLG. Scaffolds were then cast, salt leached, biopsy punched, and lyophilized as described above and incubated in PBS for quantification of peptide release as described below. Antigen-adsorbed scaffolds were fabricated in a similar manner to previously published reports [24]. Blank scaffolds were fabricated as described above, then incubated in phosphate buffered saline (PBS) overnight at 37 °C. Pprior to overnight equilibration in PBS, scaffolds were weighed to determine the mass of antigen to adsorb to each scaffold to achieve a final concentration of 4 μg antigen/mg PLG. Scaffolds were then incubated in PBS containing peptide antigen at 37 °C for 2 h. Each sample was washed by gentle agitation in PBS, followed by incubation in PBS for quantification of peptide release as described below.

2.3. Characterization of antigen-conjugated scaffold uniformity and antigen release in vitro

2.3.1. Scanning Electron Microscopy (SEM)

Scaffolds were mounted on stubs with carbon tape and sputter coated with gold for 120 s using the SPI-Module Sputter Coater in the Michigan Center for Materials Characterization. Scaffolds were imaged using the TESCAN MIRA3 FEG SEM in the Michigan Center for Materials Characterization.

2.3.2. Scaffold peptide loading and release assay

Total antigen loading and antigen release over time were verified using a CBQCA Protein Quantification Kit according to the manufacturer’s directions (Thermo, Cat # C6667 and Cat # A66522). To quantify total antigen loading, scaffolds were degraded in 0.2 M sodium hydroxide with 0.01 % Tween 20 at 55 °C for 48 h. To assess antigen release over time in vitro, scaffolds were incubated in PBS at 37 °C and 5.0 % CO₂. Prior to incubation, scaffolds were weighed to estimate the starting mass of peptide in each individual scaffold. PBS was collected and replaced over the course of three weeks and cumulative antigen release over time was quantified. Percentage of antigen released was quantified by comparing the cumulative antigen released to theoretical total antigen loading for each individual scaffold. Blank scaffolds were incubated at the same conditions as antigen-loaded scaffolds and used to normalize for any signal derived from PLG. Any reads that were at or below zero were considered below the limit of detection and normalized to zero. Soluble peptide antigen incubated at the same conditions as samples were used for standard curves for each experiment.

2.3.3. Cell lines

A20 cells, a B-cell line, were obtained from ATCC, and DO-11.10 cells, an OVA-specific hybridoma T-cell line, were a gift from Dr. Philippa Marrack at the National Jewish Health Center [25,26]. Both cell lines were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) with 10 % fetal bovine serum (FBS) and penicillin/streptomycin at 37 °C in 5 % CO₂. Cells were passaged every 3 days to a concentration of approximately 200,000 cells/mL.

2.3.4. Scaffold co-culture with cell lines

For scaffold-conditioned media assays, scaffolds were incubated in 100 μL of DMEM in 96-well flat-bottom plates. After 48 h, conditioned media was collected, supplemented with fresh

DMEM up to 100 μL, and cultured with 15,000 A20 and 15,000 DO-11.10 cells for 24 h in 96-well flat-bottom plates. Both cell lines were cultured with 0.5 mg/mL soluble OVA_323–339 as a positive control. For scaffold co-cultures, 15,000 each of A20 and DO-11.10 cells were seeded on blank or antigen-conjugated scaffolds or cultured with 0.5 mg/mL soluble OVA_323–339 in 100 μL of DMEM for 24 h in a 96-well flat-bottom plate. For both experiments, media was collected after 24 h and frozen at −80 °C until further analysis. Antigen-specific T-cell activation and subsequent cytokine secretion in cell culture supernatants was measured using an IL-2 ELISA (Thermo) [27].

2.4. Characterization of antigen-specific cell proliferation in vitro and trafficking in vivo

2.4.1. Mice

Female OT-II mice (B6.Cg-Tg(TcraTcrb)425Cbn/J), transgenic mice with OVA-specific CD4 TCRs [28], aged 5–7 weeks, female green fluorescent protein (GFP) mice (C57BL/6-Tg(UBC-GFP)30Scha/J) aged 5–7 weeks [29], female C57BL/6J mice aged 5–6 weeks, and female SJL/J mice aged 6 weeks were purchased from Jackson Laboratory. All procedures with mice were performed in accordance with the regulations approved by the Animal Care and Use Committee at the University of Michigan.

2.4.2. OT-II splenocytes harvest and CFSE-labeling

OT-II mice were euthanized via CO₂ asphyxiation and cervical dislocation, and spleens were harvested. Splenocytes were passed through a 70 μm filter, red blood cells were lysed with ACK buffer (Thermo), and cells were washed with PBS to generate a single-cell suspension. For in vivo histological studies, single-cell suspensions of OT-II splenocytes were prepared for adoptive transfer as described below. For in vitro proliferation studies, cell suspensions at a concentration of 10 × 10⁶ cells/mL were stained with 5 μM carboxyfluorescein succinimidyl ester (CFSE) dye (Biolegend) [30,31]. Cells were stained for 5 min, then washed with Rosewell Park Memorial Institute Medium (RPMI) 1640 media supplemented with 10 % FBS and penicillin/streptomycin to quench the stain. CFSE-labeled splenocytes were then cultured as described below.

2.4.3. In vitro proliferation

CFSE-stained OT-II splenocytes were seeded on blank or antigen-conjugated scaffolds or cultured with 0.5 mg/mL soluble OVA_323–339 in RPMI 1640 media supplemented with 10 % FBS and penicillin/streptomycin at 100,000 cells per well in 96-well flat-bottom plates. After 72 h, scaffolds were washed with media and wash media and supernatant were collected for flow cytometric characterization of antigen-specific T-cell proliferation as described below.

2.4.4. GFP+ immunization

Female GFP mice were anesthetized via inhalation of 2.5–4 % isoflurane (MWI) and immunized with an emulsion containing 4 mg/mL OVA_323–339 mixed 1:1 with Complete Freund’s Adjuvant (Freund’s Incomplete Adjuvant (BD Difco) with 4 mg/mL heat-killed M. tuberculosis H37 Ra (BD Difco)). Mice were injected with 33 μL of immunization solution in three subcutaneous locations, at the base of the tail and hind legs, totaling 100 μL per mouse. After 10 days, mice were euthanized via CO₂ asphyxiation and cervical dislocation and spleens were harvested. Spleens were passed through a 70-μm cell strainer and red blood cells lysed with ACK buffer (Thermo) to generate single-cell suspensions. Isolated splenocytes were cultured at 2–3 × 10⁶ cells/cm² in RPMI 1640 media supplemented with 10 % FBS, penicillin/streptomycin, 4 mM l-glutamine, 1× Minimum Essential Media (MEM) non-essential amino acids, 1× sodium pyruvate, 10 mM N-2-hydroxyethylpiperazine-N-2-ethane sulfonic acid (HEPES), 50 μg/mL OVA_323–339, and 25 ng/mL rIL-12 (Peprotech) for 72 h at 37 °C. After 72 h, non-adherent cells were collected, counted, and prepared for adoptive transfer as described below.

2.4.5. Adoptive transfer of T-cells

For histological studies and cell trafficking studies, harvested non-fluorescent OT-II splenocytes and antigen-specific GFP-expressing donor cells, respectively, were resuspended at 9 × 10⁶ cells in 200 μL of sterile PBS. Recipient female C57BL/6J mice received adoptive transfers of either cell suspension via lateral tail vein injection. In all studies, mice received adoptive transfers and subcutaneous scaffold implants on the same day.

2.4.6. Scaffold subcutaneous implantation

Following adoptive transfer, female C57BL/6J mice were anesthetized via inhalation of 2.5–4 % isoflurane (MWI) and administered ophthalmic lubricant (MWI) and 5 mg/kg carprofen (MWI) subcutaneously as a preemptive analgesic. The dorsa were shaved and prepared for surgery with povidone iodine and ethanol. Mice received an incision measuring approximately 1 cm along the midline of the dorsum and subcutaneous pockets for each of the six implants was made by blunt dissection. The incision was closed using surgical clips (Relex 7 mm, Durect Corporation). Mice received another dose of 5 mg/kg carprofen one day after surgery and were monitored daily for signs of infection or pain.

2.4.7. Tissue isolation and preparation

Seven days post scaffold implantation, mice were euthanized via CO₂ asphyxiation and cervical dislocation and spleens and scaffolds were harvested and placed in PBS on ice. Spleens were passed through a 70-μm cell strainer and red blood cells were lysed with ACK buffer (Thermo) to generate single-cell suspensions. Scaffolds were harvested, minced with a scalpel, enzymatically degraded with 0.26 wU/mL liberase (Sigma) for 30 min at 37 °C while agitating, and filtered through a 70-μm cell strainer to generate single-cell suspensions.

2.4.8. Hematoxylin and eosin staining and cellular infiltration quantification

Scaffolds were harvested 7 days after implantation with attached skin intact for spatial orientation when staining. After harvest, scaffolds were fixed in 10 % zinc-buffered formalin at room temperature for 24 h, washed in 70 % ethanol, bisected, and loaded into tissue cassettes. The Histology Core at the University of Michigan Dental School paraffin embedded, sectioned, and hematoxylin and eosin (H&E) stained tissues. Slides were imaged using a Biotek Cytation 5 Imaging Reader (Agilent). For automated cell counting, images depicting broken scaffolds or scaffolds covered by debris were not used. Cell counts were quantified using QuPath. Scaffold area was manually selected, and cells were detected using the following cell detection settings: the minimum and maximum detected areas of the nucleus were 10 pixels² (px²) and 100 px², respectively, a 0.125 threshold for detecting nuclear objects, and a sigma value of 2.5 pixels. The software was manually trained to exclude noncellular objects using the Train Object Classifier feature of QuPath.

2.4.9. Flow cytometry

All isolated cell suspensions were stained with the Zombie UV fixable viability kit (Biolegend), blocked with anti-CD16/32 (clone 93, Biolegend), and stained for various immune cell markers (Supplemental Table 1). For studies where only surface staining was completed, samples were fixed in 1 % paraformaldehyde prior to analysis. For studies that included intracellular staining, samples were fixed and permeabilized using the true-nuclear transcription factor buffer set (Biolegend) according to manufacturer’s directions. Intracellular staining was completed in cell staining buffer (Biolegend). Prepared samples were run on either the Bio-Rad ZE5 Cell Analyzer or the Aurora EVO Spectral Cytometer in the University of Michigan Flow Cytometry Core. All flow cytometry data was gated and analyzed using the FlowJo software package (BD Biosciences). For CFSE studies, antigen-specific CD4 T-cell populations were identified as CD45+ CD4+ CFSE+ cells, and for GFP studies, antigen-specific CD4 T-cells were identified as CD45+ CD4+ GFP+.

2.5. Characterization of autoantigen-conjugated scaffolds

2.5.1. Scaffold Co-culture with splenocytes

Female SJL/J mice were anesthetized via inhalation of 2.5–4 % isoflurane (MWI) and immunized with an emulsion containing 2 mg/mL PLP_139–151 mixed 1:1 with Complete Freund’s Adjuvant (Freund’s Incomplete Adjuvant (BD Difco) with 4 mg/mL heat-killed M. tuberculosis H37 Ra (BD Difco)). Mice were injected with 33 μL of immunization solution in three subcutaneous locations, at the base of the tail and hind legs, totaling 100 μL per mouse. After 10 days and 14 days, mice were euthanized and spleens and lymph nodes (inguinal, axillary, and brachial) were harvested, passed through a 70-μm filter and red blood cells lysed with ACK buffer (Thermo) to generate single-cell suspensions. Blank scaffolds, antigen-conjugated scaffolds, or 0.25 mg/mL soluble PLP_139–151 were co-cultured with 1 × 10⁶ cells/well in 200 μL RPMI 1640 for 24 h in a 96-well flat-bottom plate. After 24 h, media was collected, and cytokine secretion was quantified using IL-2 ELISAs (Thermo).

2.6. Statistical analysis

All statistical analysis was completed using GraphPad Prism software. For in vitro co-culture experiments with conditioned media, cell lines, and primary splenocytes, one-way ANOVA with Dunnett’s multiple comparison was used to compare IL-2 secretion in response to antigen-conjugated scaffolds and soluble antigen to control (scaffolds loaded with off-target antigen). For GFP adoptive transfer studies in vivo, either a one-way ANOVA with Dunnett’s multiple comparison or a two-way ANOVA with Šídák’s multiple comparison were used to compare cellular trafficking to antigen-conjugated implants to control (blank scaffolds). For all studies, the presence of outliers was determined using Grubb’s test. For all conditions, p < 0.05 was considered statistically significant.

3. Results

3.1. Antigen-conjugated scaffolds enable precise antigen loading with a variety of target antigens

Carbodiimide crosslinking chemistry was used to chemically conjugate C-terminally amidated peptide antigens to PLG, an FDA-approved biocompatible and biodegradable polymer [32]. A panel of OVA peptide antigens were explored: OVA_257–264, OVA_321–339, and OVA_323–339. As determined by ¹H NMR, carbodiimide crosslinking yielded coupling efficiencies of 91.67 %, 72.06 %, 97.22 %, for OVA_257–264, OVA_321–339, and OVA_323–339, respectively, to PLG (Fig. 1A-B, Supplemental Fig. 1-2). Further confirmation of peptide conjugation was obtained using DOSY that indicated a successful conjugation due to equal diffusion coefficients for both PLG and OVA_323–339 (Supplemental Fig. 3).

Fig. 1.

Synthesis of PLG-OVA_323–339 conjugates. (A) ¹H NMR spectrum of PLG, OVA_323–339, and PLG-OVA_323–339 measured in dimethyl sulfoxide-d₆ (referenced at 2.5 ppm). (B) Carbodiimide crosslinking yielded high coupling efficiencies for OVA_257–264, OVA_321–339, and OVA_323–339 peptides.

Porous antigen-conjugated scaffolds were fabricated using solvent casting and salt leaching techniques, yielding individual antigen-conjugated scaffolds (5 mm diameter and 3 mm thick) (Fig. 2A). This scaffold manufacturing process yielded uniform, porous scaffolds with consistent porosity and pore volume, independent of the type and concentration of peptide used (Fig. 2B-D, Supplemental Fig. 4, Supplemental Table 2). By ratiometrically combining conjugated and non-conjugated PLG, we fabricated scaffolds loaded with precise concentrations of antigens. To demonstrate the tunability of the system, we fabricated scaffolds loaded with a range of antigen concentrations: 0.04 μg antigen/mg PLG, 0.4 μg antigen/mg PLG, 2.2 μg antigen/mg PLG, 4 μg antigen/mg PLG, 20 μg antigen/mg PLG, and 40 μg antigen/mg PLG, hereafter denoted as 0.04, 0.4, 2.2, 4, 20, and 40, respectively for brevity. The total peptide loading for each antigen at each scaffold concentration was quantified using CBQCA assays, a fluorescence-based assay that detects primary amines present on peptides through interactions with CBQCA in the presence of cyanide [33]. Consistent with previous findings, the presence of PLG polymer did not interfere with the CBQCA assay [34], and the fluorescent output increased as the concentration of peptide antigen in each scaffold increased (Fig. 2E). At lower antigen concentrations, 0.04, 0.4, and 2.2, the assay signal was low for all three OVA peptides, but the signal increased at higher antigen concentrations, 4, 20, and 40.

Fig. 2.

Antigen-conjugated scaffolds are modular and allow for the fabrication of uniform, porous scaffolds loaded with a variety of peptide antigens. (A) Antigen-conjugated scaffolds are fabricated using solvent casting and salt leaching techniques. Created using BioRender.com (B,C) A SEM image of a 2.2 OVA_323–339 scaffold shows a cylindrical scaffold with pores throughout. Scale bars = 1 mm. (D) Percent porosity and pore volume were quantified for both blank scaffolds and antigen-conjugated scaffolds. (E) CBQCA assays were used to quantify the total antigen loading for each OVA peptide across a range of antigen concentrations. Antigen concentrations are denoted in terms of μg antigen/mg PLG.

3.2. Antigen-conjugated scaffolds exhibit sustained release of peptide that retains biological activity

Antigen-conjugated scaffolds were incubated in PBS, and CBQCA assays were used to quantify peptide released into supernatants over the course of three weeks. Consistent with previous findings, the signal of the CBQCA assay for OVA_323–339 release was low [34], so we were unable to generate a release curve for this peptide. By swapping out the peptide of interest to one that included a lysine, OVA_257–264 or OVA_321–339, the sensitivity of the assay increased, and we were able to detect peptide release (Fig. 3A-B). The overall ability to detect peptide released at the 1 h, 6 h, 1 day, and 3 day time points was limited for both peptides. At later time points, we were able to detect peptide released for both peptides at all scaffold concentrations. The release profiles between antigens were similar, with OVA_257–264-conjugated scaffolds releasing slightly more antigen at the 20 and 40 doses than the OVA_323–339-conjugated scaffolds. The lower concentration scaffolds, 0.04 and 0.4, were near the lower limit of detection, resulting in limited signal and inflated percentage released values (Supplemental Fig. 5A-B). The higher concentration scaffolds, 2.2, 4, 20, and 40, exhibited sustained release over time, with less than 10 % of antigen released after 3 weeks for both peptides (Supplemental Fig. 5A-C).

Fig. 3.

Carbodiimide crosslinking allows for sustained release of antigen and activation of antigen-specific T-cell lines. (A,B) CBQCA assays were used to quantify antigen release for OVA_257–264-conjugated scaffolds and OVA_321–339-conjugated scaffolds for 3 weeks. (C) A20 (B-cell line) and DO-11.10 (OVA-reactive T-cell line) cells were cultured in scaffold-conditioned media for 24 h, followed by an IL-2 ELISA on the supernatant to quantify antigen-specific T-cell activation (*p < 0.05, ****p < 0.0001). Schematic created using BioRender.com. Antigen concentrations are denoted in terms of μg antigen/mg PLG.

Next, we confirmed that antigen released from these scaffolds retained its biological activity. Given the similar release profiles between peptides, we selected to explore OVA_323–339-conjugated scaffolds for all experiments moving forward. This specific epitope of the OVA protein is MHC II-restricted and therefore allows for the study of CD4 T-cell responses with previously developed OVA-reactive cell lines (DO-11.10 T-cells) and commercially available transgenic mouse strains (OTII mice) [25,28,35]. For these experiments, we utilized an epitope of proteolipid protein (PLP), PLP_139–151, as our negative control antigen. PLP is a known target of autoreactive T-cells in EAE/MS, but has no relevance for these cell lines/mouse strains, and should not elicit antigen-specific recognition [36]. We incubated scaffolds in media for 48 h, then cultured A20 B-cells and DO-11.10 T-cells with the conditioned media for 24 h, followed by quantification of IL-2 cytokine release in the supernatant. Biologically active OVA released into the media should undergo antigen presentation and be recognized by the DO-11.10 cells, causing these cells to be stimulated and secrete IL-2 [27]. After analyzing the media, we found that media conditioned with OVA_323–339-conjugated scaffolds or soluble OVA_323–339 peptide led to significantly more IL-2 production than media conditioned with PLP_139–151-conjugated scaffolds, indicating that the peptide that was released retained biological activity (Fig. 3C).

3.3. Antigen-conjugated scaffolds activate antigen-specific T-cells and induce proliferation

After concluding that peptide activity was retained after carbodiimide crosslinking, we simulated how antigen-specific cells would respond in vivo when in direct contact with antigen-conjugated scaffolds. When A20 and DO-11.10 cells were directly co-cultured with antigen-conjugated scaffolds or soluble OVA_323–339 for 24 h, we found that cells were activated by antigen loading (Fig. 4A). As the concentration of OVA_323–339 loaded in each scaffold increased, the concentration of IL-2 secreted increased indicating dose dependency. Additionally, cells cultured with soluble OVA_323–339 or 4, 20, or 40 OVA_323–339-conjugated scaffolds led to significantly greater IL-2 production than control scaffolds. Due to the low levels of stimulation in response to 0.04 and 0.4 OVA_323–339-conjugated scaffolds, these conditions were excluded from the rest of the study.

Fig. 4.

Antigen-conjugated scaffolds induce dose-dependent antigen-specific activation and proliferation in vitro. (A) A20 and DO-11.10 cells were cultured with antigen-conjugated scaffolds or soluble antigen for 24 h, followed by IL-2 ELISA to quantify antigen-specific T-cell activation (***p < 0.001, ****p < 0.0001). (B) CFSE-labeled OT-II splenocytes were cultured with scaffolds or soluble antigen for 72 h. Antigen-specific proliferation of CD4+ T-cells was measured with flow cytometry via CFSE dye dilution, with the vertical line indicating the end of the parent population signal. Schematics created using BioRender.com. Antigen concentrations are denoted in terms of μg antigen/mg PLG.

We then looked to further characterize this antigen-specific T-cell activation and see whether these cells were proliferating in response to antigen-conjugated scaffolds. Here, we co-cultured CFSE-stained splenocytes from OT-II mice with antigen-conjugated scaffolds or soluble antigen. After 48 h, cell supernatants were collected and CFSE dye dilution was quantified using flow cytometry (Supplemental Fig. 6). We showed that all OVA_323–339 conditions induced a higher degree of proliferation than control, with the 20 and 40 concentrations inducing the most proliferation (Fig. 4B).

3.4. Antigen-specific T-cells become enriched in antigen-conjugated scaffolds in vivo

We then looked to characterize cell trafficking to scaffolds in vivo in healthy mice (Fig. 5A). For these experiments, 9 × 10⁶ OT-II splenocytes were adoptively transferred into naïve C57BL/6J mice via lateral tail vein injection. Following adoptive transfers, mice received 6 implants: 2.2, 4, 20, and 40 OVA_323–339-conjugated scaffolds, a 40 PLP_139–151 scaffold, and a blank scaffold. To account for local inflammation or fibrosis occurring at the scaffold site, mice receive both negative control implants (blank and PLP scaffolds) and scaffolds loaded with various doses of cognate antigen. After 7 days, scaffolds were isolated and cellular infiltration was quantified using histological staining. H&E staining of the scaffolds after explant showed cellular invasion across all scaffold conditions, with greater cell counts occurring in scaffolds loaded with OVA_323–339 as compared to the control scaffolds (Fig. 5B-C, Supplemental Fig. 7). We used QuPath software to quantify total cellular infiltration across all implant conditions. We observed similar cell counts across 2.2 and 4 OVA_323–339-conjugated scaffolds, 40 PLP_139–151-conjugated scaffolds, and blank scaffolds, with an increase in infiltration in the 20 and 40 OVA_323–339-conjugated scaffolds (Fig. 5D).

Fig. 5.

Antigen-conjugated scaffolds promote immune cell invasion and antigen-specific T-cell enrichment in vivo. (A) Naïve C57BL/6J mice received adoptive transfers of OT-II splenocytes, followed by subcutaneous scaffold implantation for 7 days. Created using BioRender.com. (B,C) Hematoxylin and eosin staining of blank and 20 OVA scaffolds. Arrows point to scaffold edges. Scale bars = 200 μm. (D) Total cellular infiltration observed across scaffold sections was quantified using QuPath. (E) To monitor antigen-specific T-cell trafficking in vivo, naïve C57BL/6J mice received adoptive transfers of GFP-expressing OVA-reactive splenocytes, followed by subcutaneous scaffold implantation for 7 days. Created using BioRender.com. (F) The number of antigen-specific CD4 T-cells across implants. (*p < 0.05, **p < 0.01). Antigen concentrations are denoted in terms of μg antigen/mg PLG.

Given these results, we confirmed that the FBR to each implant allowed for cellular invasion, with higher concentrations of antigen increasing infiltration.

After observing an increase in total cell infiltration in high concentration antigen-conjugated scaffolds, we then looked to specifically explore trafficking of antigen-specific cell populations. For these experiments, GFP mice were immunized to generate GFP-expressing OVA_323–339-reactive splenocytes. Spleens were isolated, stimulated in culture with soluble OVA_323–339 and rIL-12 to promote antigen-specific Th1 cell expansion [37], and adoptively transferred into naïve C57BL/6J mice via lateral tail vein injection. Following adoptive transfer of 9 × 10⁶ OVA_323–339-reactive T-cells, mice received 4 randomly assigned implants from the 6 conditions explored: 2.2, 4, 20, and 40 OVA_323–339-conjugated scaffolds, a 40 PLP_139–151 scaffold, and a blank scaffold. Like above, mice receive negative control implants (blank and PLP scaffolds) and scaffolds loaded with various doses of target antigen to normalize for local inflammation at the scaffold site. After 7 days, spleens and scaffolds were isolated, and antigen-specific cellular trafficking was quantified using flow cytometry (Fig. 5E, Supplemental Fig. 8). As determined via flow cytometry, the percentage of live cells was not significantly different across conditions, however a slight decrease in viability at the 40 OVA_323–339-conjugated scaffold dose was noted (Supplemental Fig. 9B). Additionally, total numbers and percentages of antigen-specific B-cells and CD8 T-cells did not increase significantly between OVA_323–339-conjugated scaffolds and controls (Supplemental Fig. 9C-H).

The number of antigen-specific CD4 T-cells was significantly greater in the 4 and 20 OVA_323–339-conjugated scaffolds than in the blank scaffold (Fig. 5F). While trending higher than the blank scaffold, the enrichment was lower for the 40 OVA_323–339-conjugated concentration than in the 4 and 20 OVA_323–339-conjugated scaffold concentrations (Fig. 5F, Supplemental Fig. 9B).

3.5. Characterization of antigen-presenting cells and CD4 T-cell populations in antigen-conjugated scaffolds in vivo

After observing complex, dose-dependent enrichment of antigen-specific CD4 T-cells in vivo, we sought to further characterize the myeloid and T-cell populations present at the site of the scaffold. Specifically, we investigated the presence of antigen-presenting cells, exhausted T-cells, and a variety of CD4 T-cell phenotypes to see whether antigen delivery had an impact on these populations. Because we observed the greatest enrichment of antigen-specific CD4 T-cells in 4 μg/mg OVA_323–339-conjugated scaffolds, we elected to further explore this dose in vivo. For this study, we followed the same procedure as described above with adoptive transfers of GFP-expressing OVA_323–339-reactive splenocytes into C57BL/6J mice (Fig. 5E). Here, mice received 3 of each type of implant for a total of 6 implants: 4 μg/mg OVA_323–339-conjugated and blank scaffolds. After 7 days, spleens and pooled scaffolds were isolated, and myeloid and lymphoid cell populations were characterized via flow cytometry (Supplemental Fig. 10). Due to limited cell numbers, phenotypes of antigen-specific CD4 T-cells could not be accurately determined and analysis was limited characterization of CD4 T-cell populations broadly. Both the spleen and scaffolds contained a variety of macrophage and dendritic cell populations, with few type-1 conventional dendritic cells (cDC1s) and tolerogenic dendritic cells, and no significant differences in APC populations between blank and OVA_323–339-conjugated scaffolds (Supplemental Fig. 11A, Fig. 6A). We observed higher expression of T-cell exhaustion markers PD-1, TIM-3, and LAG-3 in scaffolds compared to the spleen, with a significantly lower percentage of CD4 T-cells expressing PD-1 in OVA_323–339-conjugated scaffolds compared to blank (Supplemental Fig. 11B, Fig. 6B) [38]. Finally, we quantified the presence of Th1, Th2, Th17, and regulatory (Treg) CD4 T-cells across the spleen and scaffolds. Within the total CD4 T-cell population, we observed an increase in all phenotypes in scaffolds, with no differences between the two types of scaffolds, in agreement with a more inflammatory tone expected from biomaterial implantation (Supplemental Fig. 11C, Fig. 6C).

Fig. 6.

Antigen-conjugated scaffolds recruit a variety of myeloid and lymphoid cell populations, while avoiding massive T-cell exhaustion in vivo. Naïve C57BL/6J mice received subcutaneous scaffold implants and adoptive transfers of GFP+ OVA-reactive splenocytes, followed by explant on day 7. (A) The percentage of unpolarized/M0 macrophages (CD45+ CD11b+ F4/80+ CD206− CD86−), pro-inflammatory/M1 macrophages (CD45+ CD11b+ F4/80+ CD206− CD86+), pro-regenerative/M2 macrophages (CD45+ CD11b+ F4/80+ CD206+ CD86−), type-1 conventional dendritic cells/cDC1 (CD45+ CD11c+ MHC II+ CD11b− CD8a+), type-2 conventional dendritic cells/cDC2 (CD45+ CD11c+ MHC II+ CD11b+ CD8a−), and tolerogenic dendritic cells (CD45+ CD11c+ MHC II+ CD11b+ CD8a− CD80^low CD86^low), out of the total CD45+ immune cells. (B) The percentage of CD4 T-cells isolated from implants expressing T-cell exhaustion markers. (C) The percentage of T-bet+ (Th1), GATA3+ (Th2), ROR gamma (t)+ (Th17), and FoxP3+ (Treg) cells out of all CD4 T-cells. (*p < 0.05, **p < 0.01). Antigen concentration is denoted in terms of μg antigen/mg PLG.

3.6. Antigen-conjugated scaffolds target autoimmune disease-relevant T-cells in vitro

After developing the antigen-conjugated scaffold platform with OVA peptide antigens, we looked to apply this system to autoimmune disease-relevant peptides. We selected an autoantigen prevalent in multiple sclerosis, PLP_139–151, as the target antigen, with OVA_323–339 as the negative control antigen. PLP_139–151-conjugated scaffolds were fabricated at the same concentrations with the same techniques as in previous experiments, yielding a coupling efficiency of 91.14 % (Supplemental Fig. 12-13). Like the OVA-conjugated scaffolds, PLP_139–151-conjugated scaffolds were porous and yielded uniform porosity and pore volume measures independent of the concentration of peptide used (Fig. 7A, Supplemental Table 3). These scaffolds were loaded with precise concentrations of antigen and peptide release was quantified for each scaffold concentration through three weeks (Fig. 7B-C). Like the OVA peptides, the amount of antigen released increased as the antigen concentration increased. The percentage of antigen released after three weeks for each concentration scaffold was comparable to results shown above for the 2.2, 4, 20, and 40 OVA concentrations, with 10 % or less released for each concentration (Supplemental Fig. 14). After observing sustained release from antigen-conjugated scaffolds using both model and disease-relevant antigens, we sought to compare antigen release kinetics from our system to other commonly used drug delivery methods like encapsulation and adsorption. When comparing scaffolds loaded with the 4 μg/mg PLP dose via conjugation, encapsulation, or adsorption, we observed burst release followed by a plateau in release from both adsorbed and encapsulated scaffolds, whereas conjugated scaffolds exhibited sustained and prolonged release through 5 weeks (Supplemental Fig. 15).

Fig. 7.

Scaffolds loaded with autoantigens stimulate disease-relevant T-cells in vitro. (A) SEM image of a 4 PLP scaffold, scale bar = 1 mm. (B,C) CBQCA assays were used to quantify total PLP_139–151 antigen loading and antigen release over time. (D) PLP-reactive splenocytes from immunized SJL/J mice were cultured with scaffolds or soluble peptide for 24 h, followed by IL-2 ELISA to quantify antigen-specific T-cell activation (**p < 0.01, ****p < 0.0001). Schematic created using BioRender.com. Antigen concentrations are denoted in terms of μg antigen/mg PLG.

To evaluate cellular responses to myelin antigen-loaded scaffolds, we seeded antigen-specific primary cells on PLP_139–151-conjugated scaffolds. Here, SJL/J mice were immunized to generate PLP-reactive splenocytes, followed by isolation of spleens and lymph nodes. Isolated PLP-reactive cells were co-cultured with PLP_139–151-conjugated scaffolds, soluble PLP_139–151, or blank scaffolds for 24 h. For this experiment, we chose to test a range of PLP_139–151-conjugated scaffold doses, while excluding doses that resulted in little stimulation (0.04 and 2.2) or low antigen-specific cell enrichment (40) in previous experiments. We observed that 4 and 20 PLP_139–151-conjugated scaffolds induced significantly more T-cell activation than the blank scaffold, as evidenced by an increase in IL-2 secretion (Fig. 7D).

4. Discussion

T-cells are key mediators for orchestrating immune functions and exist as clonal populations expressing TCRs for specific antigens. Our ability to study antigen-specific T-cell clones, however, is limited. Efforts to collect these T-cell clones rely on the use of blood draws and ex vivo expansion, often resulting in altered phenotypes, slow expansion rates, non-functioning T-cells, and costly manufacturing procedures [5,7]. The development of a tool for the local enrichment of antigen-specific T-cell populations would provide an opportunity for the collection of these T-cell clones for immunological studies, as well as longitudinal disease detection and monitoring. Previous work has shown that the FBR to implanted biomaterials results in a vascularized niche of immune and stromal cells that are reflective of those found at sites of disease [15-17]. Immune cells within these niches are primarily innate immune cells, however, and the lack of incorporation of antigens prevents the study of antigen-specific T-cells using this approach.

Antigen delivery is a promising strategy for increasing local enrichment of adaptive immune cells, but previous efforts have resulted in uncontrolled loading, burst release of antigen, and subsequent T-cell exhaustion [22]. While previous antigen delivery efforts are limited for examining autoantigens and T-cell function, peptide antigen delivery using biomaterials has been widely explored for vaccines [39-41]. Additionally, peptides have been delivered for therapeutic applications such as the treatment of ischemic tissues [42], melanoma [43], osteochondral tissue defects [44], and diabetogenic T-cells [45]. Alternatively, antigen-loaded scaffolds implanted in the subcutaneous space allow for the generation of local depots of antigen-specific immune cells, while reducing risks of systemic activation by restricting antigen delivery to a local site. Within this space, Thelin et al. developed a scaffold with adsorbed antigens and was able to successfully enrich disease-relevant T-cell populations, however, it is difficult to control antigen loading with this method [1]. Antigen-loaded nanoparticles have shown promise in therapeutic development for autoimmunity, and previous work employed direct conjugation of peptide antigens to polymer for precisely fabricating antigen-loaded nanoparticles [23]. Inspired by this approach, we hypothesized we could apply a similar strategy to load antigen in implantable scaffolds for local monitoring of antigen-specific populations. In this report, we detail the development of a modular antigen-conjugated scaffold platform that allows for precise loading of peptide antigens, sustained release, and antigen-specific cell activation and enrichment.

In this work, we utilized carbodiimide crosslinking followed by solvent casting and salt leaching to fabricate porous PLG scaffold loaded with varying concentrations of OVA peptide antigens (Figs. 1, 2). We first demonstrated the modularity of the system by fabricating 0.04, 0.4, 2.2, 4, 20, and 40 concentrations for three distinct peptides: OVA_257–264, OVA_321–339, and OVA_323–339. Antigen concentrations were selected based on previous work completed by Thelin et al., and while our system is able to load higher concentrations of antigen, we sought to mimic prior work in initial studies [1]. Due to the wide range of peptide concentrations studied, we utilized CBQCA assays (working range = 10 ng to 150 μg) to determine total antigen loading and antigen release over time. We determined that the 0.04, 0.4, and 2.2 antigen concentrations were at the lower limit of detection for this assay, hindering our ability to quantify peptide loading for all three peptides. For the OVA_323–339 peptide in particular, the sensitivity of the assay was low and resulted in inaccurate peptide mass readings, aligning with previous findings (Fig. 2E) [34]. The CBQCA assay detects peptides through primary amine interactions with CBQCA in the presence of cyanide, so a strategy for improving assay resolution is the inclusion of additional primary amines in target peptides through the addition of lysines. In this case, OVA_323–339 does not contain any lysines, but OVA_257–264 and OVA_321–339 both contain one lysine each, contributing to enhanced sensitivity and detection of these peptides. Peptides are easy to modify, so altering the sequence to include additional lysines for accurate detection is a factor to consider when optimizing these controlled delivery systems or applying them to other applications. For example, OVA_321–339 includes an additional lysine compared to OVA_323–339, while still including the same epitope and allowing for antigen recognition. Additionally, peptides could be altered to complement other conjugation chemistries of interest such as cysteine/maleimide chemistries [18]. When exploring other loading methods, it is important to consider the impact that differences in conjugation efficiency between chemistries would have on antigen loading efficiency, antigen release, and subsequent T-cell engagement. In our studies, the versatility of carbodiimide crosslinking and high sensitivity of the CBQCA assay allows for the fabrication of scaffolds that are precisely loaded with any peptide antigen of interest.

Following chemical conjugation, antigen release is dependent upon PLG backbone degradation via hydrolyzation, contributing to a more sustained release profile compared to systems where antigen is encapsulated [22,32,46]. For this study, 75:25 (lactic acid/glycolic acid) PLG was used, but extensive prior work demonstrates that PLG composition and molecular weight can be tuned to modify degradation rates, which would alter peptide delivery [47]. At early time points (1 h, 6 h, 1 day, and 3 day), we were not able to detect antigen release, but at later time points, we detected peptide release for both OVA_257–264 and OVA_321–339 and all scaffold concentrations (Fig. 3A-B). Release profiles did not differ significantly between the sequence of peptide antigen conjugated, only the concentration of antigen loaded. The 0.04 and 0.4 concentrations were at the lower limit of detection for both peptides, so our ability to accurately detect release was limited using this assay (Supplemental Fig. 5A-B). The higher concentration scaffolds, 2.2, 4, 20, and 40, allowed for sustained antigen release, with less than 10 % of total antigen loaded released by three weeks for each concentration (Supplemental Fig. 5A, C). As the concentration of antigen loaded increased, the cumulative mass released increased as well (Fig. 3A-B). Consistent with previous reports, we hypothesize this is due to two factors: (1) an increase in the initial release of surface-associated antigen and (2) an increase in antigen loaded near peripheral pores of the scaffold, allowing for faster release due to the shortened path length [48-50]. When applying this system to other applications, it is also important to consider other physiochemical properties of peptides that may alter the release kinetics such as size, hydrophobicity, and charge [19,51,52]. Here, we have demonstrated release from scaffolds with antigen distributed homogenously throughout the scaffold, but future studies could look to investigate scaffolds with antigen density varying by region. Such studies would allow for analysis of the spatial distribution of cells and lymphoid structure formation in scaffolds. While this work explored the use of peptide antigens, it is also possible that full length proteins could be conjugated to PLG in a similar manner as demonstrated by Skoumal et al. [53]. The incorporation of full-length protein would allow for the delivery of multiple epitopes at once from a single antigen, however, further studies would need to be conducted to assess protein stability. PLG degradation products may impact the pH of the local environment, which could impact protein stability and subsequent antigen-presentation and T-cell activation [32]. Not to mention, scaffold fabrication via solvent casting involves the use of chloroform, which also has the potential to impact protein stability. Ultimately, our studies suggest that chemical conjugation of antigen allows for controlled loading and sustained release profiles.

We utilized OVA_323–339-conjugated scaffolds for further studies, but the ability to swap out which antigen is conjugated allows for the study of a variety of other epitopes from a single antigen of interest [54]. For example, future studies could include OVA_257–264-conjugated scaffolds, as the OVA_257–264 epitope is MHC-I restricted and allows for the study of CD8 T-cell function [55]. In contrast to other biomaterial scaffold platforms that incorporate peptide-loaded MHC complexes, costimulatory molecules, cytokines, or chemokines for direct presentation to antigen-specific T-cells [56-58], our platform requires antigen processing and presentation by APCs for antigen-specific T-cell activation. Direct presentation removes the need for APCs, however, peptide-loaded MHC platforms require complex manufacturing processes and host matching of MHC, and often result in unstable complexes with improper peptide loading [59]. By leveraging conjugation strategies, we have overcome these limitations and streamlined the fabrication process to generate scaffolds that can be widely applied for prolonged delivery of antigen. When antigen-specific cell lines were cultured in OVA_323–339-conjugated scaffold-conditioned media, we observed an increase in IL-2 secretion compared to media conditioned with control antigen (Fig. 3C). This increase in IL-2 production indicated that the scaffold fabrication process did not hinder the ability of the OVA_323–339-conjugated scaffolds to degrade and release antigen that retained biological function.

To further understand interactions between antigen-specific T-cells and antigen-conjugated scaffolds, we seeded cells directly on scaffolds. When antigen-specific cell lines were seeded on scaffolds, we observed not only an antigen-specific increase in IL-2 production, but a dose-dependent increase in IL-2 production as well (Fig. 4A). Additionally, when antigen-specific primary cells were cultured with scaffolds, we observed an increase in proliferation in response to increasing concentrations of target antigen (Fig. 4B). Taken together, these experiments confirm that antigen-conjugated scaffolds induce both activation and proliferation of antigen-specific T-cells in a dose-dependent, antigen-specific manner.

Next, we were interested to see if systemically circulating antigen-specific immune cells would traffic to and become enriched in antigen-conjugated scaffolds. Here, we utilized adoptive transfer models in mice for histological and flow cytometric studies (Fig. 5A,E, Supplemental Fig. 8). To account for local inflammation or fibrosis at the scaffold, mice received negative control implants in addition to implants loaded with varying concentrations of cognate antigen. Due to the high porosity of the implants (Fig. 2D), we were expecting to observe cellular infiltration across all implants. Additionally, because we saw an increase in antigen released (Fig. 3) and subsequent T-cell activation (Fig. 4) as the antigen concentration increased, we expected to observe an increase in infiltration of antigen-specific CD4 T-cells as the concentration of target antigen loaded increased. While we observed an increase in total cell infiltration as the concentration of OVA_323–339 increased (Fig. 5B-D), this was not the case for antigen-specific CD4 T-cells. Instead, we observed an increase in antigen-specific T-cell enrichment from the 2.2 to the 4 concentrations, followed by decreased enrichment in the 20 and 40 concentrations (Fig. 5F). One potential cause for the decrease in antigen-specific trafficking is AICD, previously reported by Griffen et al. at high antigen concentrations [22]. In other words, it is possible that high dose scaffolds are releasing too much antigen and overstimulating T-cells of interest. While this phenomenon is interesting from a therapeutic standpoint, it is less helpful for when looking to learn more about specific T-cell clone phenotypes in healthy states vs diseased states. Because our system allows for precise loading of peptide, an advantage of the system is the ability to readily tailor the concentration of antigen for specific studies [1,22]. Ultimately, our studies showed that the delivery of cognate antigen increased enrichment of antigen-specific CD4 T-cells in a dose dependent manner, suggesting that it is the specific antigen delivery that matters.

After observing this complex pattern in enrichment of antigen-specific CD4 T-cells in vivo, we sought to further characterize the types of immune cells that were present at the scaffold site and any impact antigen delivery had on these populations. We repeated our adoptive transfer model in healthy mice, followed by flow cytometry to quantify the presence of myeloid and lymphoid cell populations (Supplemental Fig. 10). Within the spleens and scaffolds, we detected a variety of macrophage and dendritic cell (DC) populations, with naïve (M0) macrophages and type-2 conventional dendritic cell (cDC2) populations, potent CD4 T-cell activators, being the most abundant of the two subsets (Supplemental Fig. 11A, Fig. 6A) [60]. We also observed similar percentages of pro-regenerative (M2) macrophages and pro-inflammatory (M1) macrophages in both types of scaffolds. Additionally, we saw a small percentage of tolerogenic DCs in both types of scaffolds, and negligible percentages of type-1 conventional DCs (cDC1), CD8 T-cell activators [60]. While we observed an increase in the percentage of total CD4 T-cells expressing exhaustion markers PD-1, TIM-3, and LAG-3 in scaffolds compared to the spleen, the expression between blank and antigen-loaded scaffolds were largely similar, with CD4 T-cells in OVA_323–339-conjugated scaffolds expressing less PD-1 (Fig. 6B, Supplemental Fig. 11B). Finally, we looked to characterize the specific phenotypes of CD4 T-cells that were present in the spleens and scaffolds. Th1, Th2, Th17, and Treg cells were present in both the scaffolds and the spleen (Fig. 6C, Supplemental Fig. 11C). Out of the total CD4 T-cells, scaffolds recruited higher percentages of all phenotypes compared to the spleen, with no differences between antigen-conjugated and blank scaffolds. In this study, limited cell numbers prevented the phenotyping of antigen-specific CD4 T-cells, but we successfully characterized the overall CD4 T-cell populations and ultimately determined that the presence of antigen did not skew phenotypes or exhaustion levels. Although these studies examined T-cell recruitment to the scaffolds and characterized phenotype/exhausted states to some degree, future work should further explore T-cell states in long-term studies (e.g. investigating memory T-cells).

Finally, we applied the antigen-conjugated scaffold platform to measure autoreactive T-cell responses using MS-relevant myelin autoantigens. Due to the modularity of the conjugation chemistry, this system can be applied to a variety of disease models in which the target antigens have been identified. In our case, autoreactive CD4 T-cells play a role in the pathogenesis of MS, and the autoantigen PLP_139–151 has been implicated in both MS and the mouse model, EAE [61-64]. Once again, we confirmed our ability to generate uniform, porous scaffolds loaded with precise concentrations of antigen (Fig. 7A-B). Similar to the OVA-conjugated scaffolds, we saw sustained release profiles at all concentrations, with faster release at the 40 concentration (Fig. 6C, Supplemental Fig. 14). When compared to scaffolds loaded with antigen via adsorption or encapsulation, antigen-conjugated scaffolds prevented burst release and enabled continuous delivery of antigen through 5 weeks (Supplemental Fig. 15). Adsorbed and encapsulated scaffolds exhibited major and minor burst release, respectively, and both exhibited a plateau in release after the 5-week period, indicating that the antigen payload had been exhausted. This comparison further highlights the uncontrolled loading and burst release limitations associated with adsorption and encapsulation and provides further evidence for the utility of our antigen-conjugated platform for longer release studies [1,22]. We then used co-culture experiments to determine if these implants could interact with disease-relevant T-cells. Though we only studied a subset of the antigen concentrations that we studied previously, we saw the same results: as the concentration of target antigen increased, the concentration of IL-2 secreted increased (Fig. 7D). Taken together, these experiments provided further evidence that the antigen-conjugated scaffold platform can be applied to disease-relevant models and stimulate disease-relevant T-cell populations in a dose-dependent manner. Future directions for this work include in vivo studies investigating the impact that antigen delivery has on local APC and T-cell phenotypes during early and late-stage disease progression in models of autoimmunity.

In conclusion, our results demonstrate the development of an antigen-conjugated scaffold for enrichment of antigen-specific CD4 T-cells. Due to the rarity of these T-cell populations, large quantities of blood are needed to study these cells using currently available tools. In small animal models, a blood draw of such necessary volume would normally require the animals to be euthanized. By using this system, in contrast, rare cell populations can traffic to and be retained at the site of the scaffold for further study after biopsy or explant. Our platform obviates the need for euthanasia in animal studies of antigen-specific T-cells, potentially permitting longitudinal studies using individual mice. In humans, these scaffolds could be implanted using a trocar into the subcutaneous space under local anesthesia. While this minimally invasive procedure may cause more discomfort than a blood draw, the potential for cellular information gained from the scaffold is far greater. The concentration of antigen loaded and size of the implant can be optimized for human applications by simply adjusting the ratio of conjugated to non-conjugated polymer and altering the size of the mold the polymer solution is cast in, respectively. Finally, our technique removes the need for ex vivo expansion, enabling facile examination of T-cells immediately upon removal. In the long run, this system could be used for early detection of diseased T-cells or to identify antigens useful for precision therapies.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jconrel.2025.114564.

Data availability

Data will be made available on request.