Engineering the lymph node environment promotes antigen-specific efficacy in type 1 diabetes and islet transplantation

Autoimmune diseases like type 1 diabetes (T1D) are currently managed with systemic immunosuppression or modulatory biologics that are not curative and can cause non-specific suppression, as even targeted monoclonal antibodies (for example, anti-CD3 teplizumab) do not discriminate healthy from autoreactive cells. Allogeneic islet transplantation restores insulin control but requires broad systemic immunosuppression to prevent rejection. Antigen-specific tolerance is an attractive alternative that could selectively restrain self-reactivity without compromising protective immunity by redirecting antigen-specific T cells toward regulatory phenotypes. Because T-cell fate is determined by integration of signals within lymph nodes (LNs), effective targeting and sustained presentation of antigen with tolerogenic cues in LNs is a key challenge. Intralymph node (iLN) injection has shown promise clinically in allergy, cancer, and T1D, but soluble cues rapidly drain. The authors previously developed iLN delivery of diffusion-limited, degradable microparticles (MPs) that retain within LNs and release cargo over time. This study applies that platform to T1D and allogeneic islet transplantation to test whether co-delivery of self- or allo-antigens with rapamycin can induce durable, antigen-specific tolerance, define the roles of each component, and characterize local and systemic LN responses and microenvironmental changes.

Clinical strategies for autoimmune disease rely on systemic immunosuppression or biologics (e.g., anti-CD3 teplizumab) that delay progression but are non-curative and non-specific. iLN injections of soluble vaccines have demonstrated potency and dose sparing in allergy, cancer, and T1D, but soluble agents rapidly drain from LNs. Prior work by the authors showed iLN delivery of adjuvant/antigen in MPs can locally condition LNs and promote tolerance in a multiple sclerosis model, associated with increased polyclonal Treg. Additional literature indicates that particulate antigen delivered systemically without regulatory cues can induce tolerance via tolerogenic processing by APCs in LNs/spleen/liver, causing deletion or exhaustion of effector T cells and Treg generation. mTOR inhibition is known to favor Treg differentiation and central memory phenotypes. The role of LN stromal laminins (lama4/lama5) in shaping tolerance versus inflammation has also been established, with higher lama4:lama5 ratios favoring Treg induction.

Design: Engineering of the LN microenvironment using direct iLN injection of PLGA microparticles co-encapsulating rapamycin (Rapa) and peptide antigens relevant to T1D or alloimmunity. Antigens: p31 (GAD65 mimotope), NRP-V7 (islet autoantigen mimotope), and Ea (I-Ed class II MHC peptide). MPs: PLGA MPs synthesized by double emulsion, formulated to be diffusion-limited and retained in LNs. Characterization included particle size, antigen and Rapa loading, and in vitro release kinetics (p31 and Rapa) by HPLC/UV-Vis; p31 released faster than Rapa. In vitro assays: Co-cultures of dendritic cells (DCs) with TCR-transgenic T cells: BDC2.5 (CD4, p31-specific), TEa (CD4, Ea-specific), and NOD8.3 (CD8, NRP-V7-specific). MPs (antigen alone or antigen+Rapa) were added to LPS-stimulated DCs; T-cell proliferation (CSFE/CellTrace dilution), Foxp3 expression, and cytokines (IFNγ, IL-17) were measured by flow cytometry. In vivo models: (1) Accelerated T1D model: Adoptive transfer of 15×10^6 ex vivo activated NOD8.3 CD8+ T cells into female NOD mice. iLN MP treatment was administered either prophylactically (day −3), therapeutically (day +1), or with delayed induction (day −10 treatment, then transfer). Blood glucose was monitored; diabetes defined as >250 mg/dL. (2) Allogeneic islet transplantation: Streptozotocin-induced diabetic C57BL/6 recipients received 400 BALB/c islets under the kidney capsule at diabetes onset (day 0) along with a single iLN MP treatment. Graft survival (glycemic control) was monitored; rejection defined by sustained hyperglycemia >300 mg/dL. Mechanistic studies: Antigen-specific priming and mTOR activity assessed after iLN p31/Rapa MPs with adoptive transfer of BDC2.5 cells labeled with CSFE or Thy1.1 marker; CD69 and phosphorylated S6 (ps6) measured in treated and untreated LNs (inguinal, axillary, popliteal, pancreatic). Antigen-specific Treg expansion (Foxp3+ among BDC2.5) quantified across LNs at days 2 and 7. Antigen biodistribution/presentation: DiR-labeled MPs to confirm depot retention; Y-Ae antibody staining to detect Ea peptide presented in recipient MHC-II in high endothelial venules (HEVs) and cortical ridges (CRs) of treated and distant LNs over time; FTY720 used to block APC egress to test dependence on APC migration versus lymphatic drainage. LN stromal remodeling: Immunofluorescence for laminin α4 and α5 in HEVs and CRs after Ea/Rapa or p31/Rapa MP treatment, with or without donor-specific splenocyte transfer (DST) to induce alloimmunity; quantification of lama4:lama5 ratio and localization of antigen-specific Foxp3+ Treg (Va4+ Foxp3+ for TEa). Memory phenotype: In vitro central memory markers (CD44, CD62L, CCR7) on BDC2.5 Treg after p31 versus p31/Rapa MPs; in vivo expression of CD44/CD62L, Bcl-2, and CD127 among p31-specific Foxp3+ Treg in treated LNs 7 days post-transfer. Statistics: One-way ANOVA with Tukey’s post hoc, Welch’s t-test for two-group comparisons, log-rank tests for survival; n sizes detailed per experiment.

- In vitro, co-delivery of antigen with Rapa in MPs shifted T-cell responses toward tolerance across CD4 and CD8 models: increased Foxp3+ Treg among BDC2.5 and TEa T cells, and reduced IFNγ and IL-17; for NOD8.3 CD8+ T cells, NRP-V7/Rapa MPs reduced proliferation and IFNγ compared to antigen MPs alone. - Prophylactic accelerated T1D model (NOD8.3 transfer): A single iLN treatment with p31/Rapa MPs or NRP-V7/Rapa MPs prevented diabetes in all animals; p31-only MPs had no benefit; Rapa-only MPs partially inhibited disease. iLN MPs outperformed intraperitoneal soluble p31+Rapa at matched timing. - Allogeneic islet transplant (full MHC mismatch): Median graft survival (days): Vehicle 9, Empty MP 12, Ea MP 11, Rapa MP 21, Ea/Rapa MP 56; Ea/Rapa MPs significantly prolonged graft function but did not fully prevent rejection, likely due to broad alloantigen responses beyond the single Ea disparity. - Antigen-specific priming and mTOR modulation: p31-containing MPs induced BDC2.5 activation (CD69 upregulation) in treated and untreated LNs; inclusion of Rapa reduced ps6 among activated cells, indicating mTOR inhibition. Antigen presentation and T-cell activation were detected in distant LNs. - Expansion of antigen-specific Treg: p31/Rapa MPs expanded Foxp3+ BDC2.5 Treg in treated and early-draining LNs by day 7, whereas antigen-alone MPs did not increase Treg despite robust proliferation. Antigen is required for expansion; Rapa is required for polarization to Treg. - Antigen dissemination independent of APC migration: DiR-MPs were retained only in treated LNs. Ea presentation (Y-Ae) was detected in treated and distant LNs; blocking APC egress with FTY720 did not reduce presentation, supporting lymphatic drainage of released antigen as the mechanism. - LN stromal remodeling to tolerogenic microdomains: Ea/Rapa or p31/Rapa MPs increased laminin α4:α5 ratios in HEVs and CRs, a signature favoring Treg induction. Ea/Rapa MPs promoted accumulation of antigen-specific Foxp3+ T cells in HEVs and CRs during an alloimmune response; Rapa alone increased laminin ratios but did not increase antigen-specific Treg in the absence of co-delivered antigen. - Therapeutic regimen and durability: When treated one day after NOD8.3 transfer, both Rapa-only MPs and antigen/Rapa MPs prevented disease, indicating strong short-term suppression by Rapa. With a 10-day delay between MP treatment and disease induction, durable protection required co-encapsulation of antigen with Rapa (p31/Rapa or NRP-V7/Rapa doubled disease-free survival rate compared to Rapa-only MPs). - Memory-like Treg phenotype: Antigen/Rapa MPs increased central memory markers in vitro (CD44+CD62L+, CCR7) among BDC2.5 Treg. In vivo, p31/Rapa MPs yielded higher numbers and frequencies of antigen-specific Treg expressing CD44/CD62L and survival markers Bcl-2 and CD127, consistent with enhanced persistence.

The study demonstrates that engineering the LN microenvironment via iLN depots co-delivering antigen with rapamycin induces antigen-specific tolerance in T1D and prolongs allograft survival in islet transplantation. Antigen provides specificity and primes cognate T cells locally and in distant LNs, while Rapa suppresses mTOR signaling during priming to polarize these cells toward regulatory phenotypes. The depot’s retention in the treated LN yields prolonged local release, while antigen (and likely some Rapa) disseminates through lymphatic drainage to untreated LNs, enabling systemic yet antigen-dependent tolerance without broad suppression. The approach promotes tolerogenic stromal microdomains (increased laminin α4:α5 in HEVs/CRs) that favor Treg induction and accumulation. Importantly, durable tolerance in T1D required co-delivery of antigen with Rapa, aligning with enhanced memory-like features (CD44/CD62L, Bcl-2, CD127) among antigen-specific Treg, whereas Rapa alone provided only transient protection when disease induction was close in time to treatment. In the stringent full MHC-mismatch transplant model, targeting a single class II mismatch with Ea/Rapa MPs markedly extended graft survival but did not fully prevent rejection, suggesting that including multiple donor alloantigens may be needed to broadly suppress diverse alloimmune responses. Overall, localized LN engineering enables dose sparing and selective immune modulation while minimizing systemic exposure, offering a pathway to safer, antigen-specific therapies for autoimmunity and transplantation.

Direct iLN delivery of degradable PLGA MPs co-encapsulating disease-relevant antigens and rapamycin induces antigen-specific regulatory immunity, prevents T1D in a CD8-driven adoptive transfer model, and significantly prolongs allogeneic islet graft survival. Efficacy depends on co-delivery of antigen and Rapa: antigen is necessary for specificity and expansion, and Rapa drives tolerogenic polarization, LN stromal remodeling, and memory-like Treg features that support durable tolerance. Antigen presentation occurs in treated and untreated LNs via lymphatic drainage from the depot, enabling systemic tolerance without systemic drug exposure. This modular platform can be adapted to different antigens and potentially multiple alloantigens to expand breadth of tolerance, with clinical feasibility supported by non-surgical LN access. Future directions include optimizing release kinetics, defining biodistribution of Rapa and antigen in distant LNs, testing combinations of multiple alloantigens for transplantation, evaluating spontaneous NOD models and long-term Treg persistence/function, and assessing the impact of LN selection and tissue-drainage context on efficacy.

- In the full MHC-mismatched islet transplant model, co-delivery of a single alloantigen (Ea) with Rapa prolonged but did not fully prevent rejection, indicating limited breadth against diverse alloantigens. - Biodistribution and concentration of Rapa and antigen in untreated LNs were not fully quantified; mechanistic contributions at distant sites need clarification. - The exact origin and fate of antigen-specific Treg (maintenance/expansion of natural or peripheral Treg versus de novo conversion) remain to be elucidated. - Most experiments used adoptive transfer models; spontaneous NOD disease was proposed for future testing. - Only female mice were used; generalizability across sexes was not assessed. - LN selection and drainage patterns may influence outcomes; this was not systematically compared. - Sample sizes in some imaging/histology timepoints were small, and longer-term functional assessments of Treg durability beyond the tested windows are needed.