Local control of T cell fate in lymph nodes safely and durably reverses myelin-driven autoimmunity

Significance

Treatments for multiple sclerosis (MS)—an inflammatory autoimmune disease—rely on systemic dosing, are noncurative, and leave patients vulnerable to infection. Here, we advance a lymph node targeting strategy that exploits local dosing and retention to guide T cell fates. We achieve durable efficacy with a single treatment in multiple preclinical models and show favorable profiles for safety in non-human primates and for chemistry and manufacturing controls (CMC). Mechanistic findings reveal how antigen-specific, tolerogenic T cells are reprogrammed, demonstrating a paradigm to overcome hurdles facing systemic tolerance induction strategies (e.g., antigen-coupled myeloid cells, mRNA vaccines, genetic engineering). Together, these results progress the technology toward human translation as a potential therapy to treat autoimmune disease with antigen-specific selectivity.

Abstract

Development of tolerogenic, antigen-specific immunotherapies could overcome limitations of existing treatments for inflammatory autoimmune diseases by achieving potent, durable remission without impacting healthy immune surveillance. Here, we deliver diffusion-limited polymer depots to lymph nodes to locally guide T cell fates for the treatment of multiple sclerosis (MS), an autoimmune disease that occurs when the immune system mistakenly attacks myelin. In preclinical MS models, depots loaded with myelin self-antigen and tolerizing cues mediate localized retention of activated CD4 T cells, promote myelin-specific regulatory T cells, and reshape inflammation in the central nervous system (CNS) to eliminate lesions. Selective disease reversal is achieved with a single treatment that induces long-lasting remission without hindering healthy responses to vaccine challenge with foreign antigen. Furthermore, depots offer favorable chemistry and manufacturing control features and are well tolerated in non-human primates. This work supports a clinically feasible concept for inducing safe, effective, antigen-specific tolerance without systemic or repeated dosing.

Multiple sclerosis (MS) is an autoimmune disorder impacting 2.8 million individuals worldwide (1). MS occurs when proinflammatory lymphocytes mistakenly target myelin, the protective sheath insulating axons of the central nervous system (CNS) (2). Demyelination results in severe disabilities including pain, motor deficits, and cognitive deterioration (1). Although there are over 20 approved disease-modifying therapies (DMTs), none are curative and many leave patients vulnerable to opportunistic infections or interfere with vaccination (3). Thus, next-generation MS therapies that restore peripheral tolerance to myelin without impacting healthy immunity must be both potent and selective.

Many experimental approaches to antigen-specific tolerance have thus far centered on ex vivo generation of regulatory T cells (T_regs) for suppression of autoreactive cells (4). However, strategies including conditioning (5, 6) and genetic engineering (7–10) are limited by major histocompatibility complex (MHC) restriction, potential concerns with lentiviral-based technologies, high cost, and challenges associated with low T_reg frequency in peripheral blood (11). These obstacles have motivated exploration of immunotherapies that instead generate antigen-specific T_regs in vivo (12–14). We posited that directly programming the LN niche could further transform the MS treatment landscape by local guidance of myelin-specific T cells in lymphoid tissue without systemic drug infusion or cell therapies. We have previously reported in proof-of-principle studies intra-LN (i.ln.) injection concentrates and retains immunomodulatory cues when coupled with diffusion-limited microparticles (MPs) designed to be too large to drain from treated tissue (15). Underscoring this strategy, efficacy does not occur after injection of depots via other routes, or following local LN injection of soluble cargo (15). Further understanding the role of myelin-specific T cell differentiation and function in such approaches could enable clinical success in related human therapies to selectively treat MS.

Here, we address this gap by elucidating key mechanisms underlying our strategic approach. We demonstrate accumulation of clonally distinct T cells in individual LNs containing MP-derived regulatory cues. Diverse antigens drive de novo polarization of T cells to stable, antigen-specific T_regs that are highly suppressive. Transcriptional analysis reveals treatment alters signaling pathways that directly control trafficking of T cells. These effects correlate with elimination of spinal cord lesions and disruptive, long-lasting efficacy in multiple autoimmune models following a single treatment. Furthermore, we show that this vaccine-like therapy does not interfere with healthy T and B cell responses during simultaneous challenge with foreign antigen or impede peripheral tolerance. Finally, LN depots offer favorable safety outcomes in non-human primates (NHPs) and an attractive profile for chemistry and manufacturing controls (CMC). This work supports a clinically feasible concept for inducing safe, potent, and myelin-specific tolerance without systemic or repeated dosing.

Results

I.ln. MPs Create Local Niches that Drive Antigen-Specific T Cells and Promote T_regs.

We hypothesized i.ln. MPs would generate T cell niches in an antigen-specific manner and that inclusion of the regulatory cue—rapamycin (Rapa)—would promote T_regs among these populations. To test this, we treated mice with MPs coencapsulating Rapa and peptide antigen derived either from myelin oligodendrocyte protein (MOG) or a model protein, ovalbumin (OVA) via direct injection into inguinal LNs. Twenty-four hours after treatment, all cohorts received intravenous (i.v.) infusions of a 1:1 mixture of 2D2 (MOG-specific) and OT-II (OVA-specific) transgenic donor CD4 T cells (Fig. 1A). Dyes with distinct spectral profiles were used to track donor cell frequencies (SI Appendix, Fig. S1 A and B) and proliferated cells in treated and untreated lymph nodes (Fig. 1B and SI Appendix, Fig. S1 C and D). Strikingly, 2D2 (MOG-specific) cells proliferated in LNs treated with MOG/Rapa MPs three days after transfer, while 2D2 cells isolated from OVA/Rapa-treated LNs in the same mice exhibited minimal proliferation (Fig. 1 B and C). Likewise, OVA-specific T cells proliferated in LNs treated with OVA/Rapa MPs but not MOG/Rapa MPs. Excitingly, when we administered MOG/Rapa MPs and OVA/Rapa MPs to the same LN, both populations of transgenic T cells divided, but proliferating cells were absent in the sham-injected contralateral LNs (Fig. 1 B and C). Likewise, divided cells were not recovered in distal, untreated cervical LNs three days after adoptive transfer (SI Appendix, Fig. S1B). Thus, i.ln. MPs create local niches that control clonally distinct T cells in response to diverse antigens.

Fig. 1.

MPs drive local programming of T_regs in LNs in an antigen-dependent manner. (A) Schematic illustrating design for transfer studies (panels B–E). (B) Representative plots showing proliferation among transferred cells in mice differentially treated in left (L) and right (R) LNs. (C) Frequency of divided cells in LNs among MOG-specific 2D2 (Left panel) or OVA-specific OT-II T cells (Right panel), three days after transfer. Single data points represent individual LNs. (D) Frequency of pS6+ 2D2 T cells in left (L) and right (R) LNs, 24h after transfer. (E) Quantification of T_regs in LNs among MOG-specific 2D2 (Left panel) or OVA-specific OT-II T cells (Right panel), six days after transfer. (F) Schematic illustrating design of transfer studies (panel G). (G) Number of Foxp3/GFP+ 2D2 T cells among T_reg-depleted donor cells, six days after transfer. Error bars represent mean ± SD. Dotted vertical lines separate i.ln. treatment cohorts for individual mice. ns = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by one-way ANOVA Tukey−Kramer multiple comparisons test.

Given this antigen-specific, highly localized proliferation, we tested whether Rapa also locally alters T cell signaling by quantifying phosphorylation of ribosomal protein S6 (pS6)—a downstream target of the mechanistic target of rapamycin (mTOR) complex 1 (mTORc1) that is inhibited by rapamycin (16). Cohorts were treated in left LNs with MOG or MOG/Rapa MPs and right LNs with sham injections, or sham injections in both LNs, followed by adoptive transfer of 2D2 T cells (SI Appendix, Fig. S2A). Twenty-four hours later, pS6+ donor cells recovered from LNs treated with MOG/Rapa MPs decreased relative to MOG MPs (Fig. 1D and SI Appendix, Fig. S2B) included as a baseline control. These data indicate Rapa in MPs inhibits mTORc1 activity in antigen-specific cells in treated LNs.

We next investigated how localized, i.ln. MP treatment precisely impacts antigen-specific T_regs. First, mice were treated with MOG/Rapa MPs (left LN) and OVA/Rapa MPs (right LN), or both MP forms in a single (left) LN and a contralateral sham injection (right LN). Six days after adoptive transfer, we observed an increase in the frequency (Fig. 1E) and number (SI Appendix, Fig. S3A) of T_regs among 2D2 cells in MOG/Rapa-treated LNs, compared with 2D2 cells in OVA/Rapa-treated contralateral LNs within the same animal. Likewise, T_regs among OT-II T cells were higher in OVA/Rapa-treated LNs compared with MOG/Rapa MPs. T_regs among 2D2 and OT-II cells were increased in LNs treated with both MOG/Rapa and OVA/Rapa MP formulations (Fig. 1E and SI Appendix, Fig. S3A). No changes were observed in CD4+ T cell frequencies among non-antigen-specific, recipient cells in MOG/Rapa-treated mice (SI Appendix, Fig. S3B), and no differences were observed in T_reg frequencies in distal, untreated cervical LNs following adoptive transfer (SI Appendix, Fig. S3C), supporting the specificity and localized nature of the programming paradigm. Importantly, the frequency of recovered, antigen-specific donor T_regs was greater than the proportion of T_regs transferred into recipients (SI Appendix, Fig. S3D), indicating MPs simultaneously support distinct, antigen-specific T_reg populations within a single LN.

Induced T_regs have been shown to re-establish tolerance when T_regs are absent or ineffective (17, 18). To assess whether MOG/Rapa MPs generate induced T_regs from naive CD4 T cells in the periphery, we sorted GFP-, naive CD4+ T cells from 2D2 Foxp3/GFP mice (SI Appendix, Fig. S3D), and adoptively transferred T_reg-depleted cells into MP-treated, congenic recipients (Fig. 1F). While Rapa MPs increased the frequency of T_regs among transferred cells (SI Appendix, Fig. S3E), only MOG/Rapa MP-treated recipients exhibited a concomitant increase in numbers of GFP+ donor cells (Fig. 1G). These data highlight the need for codelivery of antigen and Rapa and show i.ln. treatment promotes T_regs in treated tissues.

Antigen-Specific T_regs are Suppressive, Long-Lasting, and Provide Durable Protection Against Autoimmunity.

To investigate whether T_regs are functionally suppressive, we sorted GFP+ CD4 T cells from LNs of untreated or MOG/Rapa MP-treated Foxp3/GFP mice for coculture in a suppression assay with naive, effector CD4 T cells stained with proliferative dye (19). T_regs from MOG/Rapa-treated LNs exhibited enhanced suppression of effector T cell proliferation compared with T_regs isolated from untreated animals (Fig. 2A and SI Appendix, Fig. S4A). Thus, MOG/Rapa MPs increase both the number of antigen-specific T_regs and their suppressive capacity on a per cell basis.

Fig. 2.

Antigen-specific T_regs are suppressive, long-lasting, and provide durable protection against autoimmunity. (A) Frequency of divided CD4 T cells at indicated ratios in suppression assays. (B) Schematic illustrating design for T_reg persistence studies (panels C and D). (C) Representative plots for antigen-specific T_regs recovered from i.ln. treatment cohorts. (D) Quantification of Foxp3/GFP+ T cells recovered from LNs. (E) Timeline for durability studies (panels F–H). Clinical score in EAE (F) or RR-EAE mice (G) following day −28 i.ln. treatment (Tx). (H) Relapse incidence in RR-EAE mice. Error bars represent mean ± SD. Dotted vertical lines separate i.ln. treatment cohorts for individual mice. Clinical scores represent mean ± SEM. sham n = 16, MOG/Rapa MPs n = 8, PLP/Rapa MPs, n = 15. ns = not significant, *P < 0.05, **P < 0.01, ****P < 0.0001 by Steel–Dwass test (clinical scores), two-way ANOVA with Tukey–Kramer post hoc test (weight loss), Wilcoxon test for Kaplan–Meier analyses, or one-way ANOVA Tukey−Kramer multiple comparisons test.

One uncertainty faced by T_reg-based therapeutic strategies to treat autoimmunity is whether T_regs are maintained to enable durable tolerance (20, 21). Thus, we examined whether MOG-specific T_regs in MP-treated LNs are long-lasting. Mice received a single injection of MOG/Rapa MPs in the left LN only, followed by i.v. infusion of CD4 T cells isolated from 2D2 Foxp3/GFP mice (Fig. 2B). One month after transfer, GFP+ cells were recovered from MOG/Rapa MP-treated LNs, but were absent or minimal in sham-injected LNs (Fig. 2 C and D and SI Appendix, Fig. S4B). This indicates that myelin-specific T_regs are locally and stably maintained in treated LNs.

To examine whether long-lived MOG-specific T_regs correlate with lasting protection in a preclinical MS model, experimental autoimmune encephalomyelitis (EAE), we treated mice once with i.ln. MOG/Rapa MPs 28 d before EAE induction (Fig. 2E). Animals were protected from paralysis measured by increases in clinical score indicative of disease severity (Fig. 2F and SI Appendix, Fig. S4C). Likewise, in a model of relapsing-remitting (RR) MS that mimics aspects of cyclic disease observed in MS patients, near-complete protection was observed in mice treated with MPs containing Rapa and proteolipid protein (PLP)—the disease-associated antigen in RR-EAE (Fig. 2G and SI Appendix, Fig. S4D). Efficacy in RR-EAE was durable with no relapse after initial disease onset during a 55-d monitoring period, despite mice receiving only a single treatment 83 d earlier. (Fig. 2H). These data indicate peptide/Rapa MPs promote myelin-specific, suppressive T_regs that are durable and correlate with long-lasting, vaccine-like protection in progressive and RR MS models.

A Single MP Treatment Confers Efficacy at Disease Onset, Reverses Paralysis in Established Disease, and Stabilizes Chronic Disease.

To test whether a single i.ln. MP treatment provides efficacy across the RR disease course, mice were first treated with PLP/Rapa MPs (supp. Table 1) seven days after RR-EAE induction, before onset of symptoms (Fig. 3A). Only MPs coloaded with both PLP and Rapa completely protected against disease (Fig. 3B and SI Appendix, Fig. S5A) and minimized weight loss (Fig. 3C). Notably, no paralysis was observed in PLP/Rapa-treated animals only, defined as a clinical score ≥3 (Fig. 3D). Then, 79% (11/14) of animals treated with PLP/Rapa MPs were asymptomatic throughout the disease course (SI Appendix, Fig. S5B). In contrast, 100% of untreated (14/14) or Rapa MP-treated (14/14) animals and 64% (9/14) of PLP MP-treated animals developed symptoms, indicated by scores > 0 (SI Appendix, Fig. S5B). This result is consistent with studies reporting that peptide only impacts severity of the disease course (22–25). Notably however, all mice in the PLP/Rapa MP, but not PLP or Rapa MP cohorts, fully recovered from onset of initial symptoms and were resistant to relapse throughout the study (Fig. 3B).

Fig. 3.

Single-dose i.ln. MP treatment provides protection against RR-EAE in both early and chronic therapeutic settings. (A) Timeline for early therapeutic MP treatment (panels B–D). Clinical score (B), weight loss (C), and paralysis incidence (D) in RR-EAE mice treated on day 7. (E) Timeline for peak disease MP treatment (panels F–H). Clinical score (F), weight loss (G), and relapse incidence (H) in RR-EAE mice treated on day 14. (I) Timeline for chronic/late disease MP treatment (panels J–L). Clinical score (J), weight loss (K), and relapse incidence (L) in RR-EAE mice treated on day 30. n=12 to 16 mice/treatment group. Error bars represent mean ± SEM. ns = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by Steel–Dwass test (clinical scores), two-way ANOVA with Tukey–Kramer post hoc test (weight loss), or Wilcoxon test for Kaplan–Meier analyses.

Next, we determined whether MPs could control established disease using a more challenging scenario: a single MP treatment at peak disease (day 14) (Fig. 3E). PLP/Rapa MPs prevented relapse, and excitingly, reversed disease and weight loss to nearly asymptomatic levels (Fig. 3 F and G). All mice in other treatment cohorts relapsed (Fig. 3H) with no significant differences in disease severity between groups (SI Appendix, Fig. S5C).

Finally, we tested efficacy using the most challenging regimen: a single MP treatment during established chronic/late-stage disease (day 30) (Fig. 3I). PLP/Rapa MPs treatment stabilized disease (Fig. 3 J and K); 69% (11/16) of sham-treated mice exhibited a second relapse, whereas only 25% (4/16) of PLP/Rapa-treated animals relapsed. Relapse in PLP/Rapa-treated mice was also delayed (Fig. 3L) and disease less severe (SI Appendix, Fig. S5D). Together these data show potent single-dose efficacy across onset, established, and chronic disease, demonstrating potential as a therapeutic vaccine for myelin-specific immunotherapy in MS.

Efficacy Correlates with Elimination of CNS Lesions.

To connect efficacy to changes in CNS lesions—the root cause of paralysis in patients—spinal cords were harvested and examined via immunofluorescence microscopy. Following an early therapeutic regimen (day 7) (Fig. 4A and SI Appendix, Fig. S6A), PLP/Rapa MP-treated mice exhibited fewer lesions (Fig. 4 B and C) and lower lesion-to-myelin ratio relative to sham (Fig. 4D). Next, we treated mice at the peak of RR-EAE (day 14) and measured lesions at day 28 (relapse phase) (Fig. 4E and SI Appendix, Fig. S6B). During this more challenging regimen, treated animals again exhibited significant reductions in lesions (Fig. 4 F and G) and the lesion/myelin ratio (SI Appendix, Fig. S6C). Remarkably, lesion levels in mice treated at peak disease was statistically undistinguishable to healthy mice never induced with RR-EAE.

Fig. 4.

Localized programming in LNs is accompanied by reduced lesions and altered T cell infiltration in the CNS. (A) Timeline for analysis of spinal cords following early therapeutic treatment (panels B–D). (B) Representative images of sectioned spinal cords harvested from healthy (Left), sham-injected RR-EAE (Middle), or PLP/Rapa MP-treated RR-EAE mice (Right). Healthy, n = 2 mice. Sham/RR-EAE, n = 8 mice, PLP/Rapa MP/RR-EAE, n = 8. Average number of lesions (C) and lesion/myelin ratio per FOV per mouse (D). (E) Timeline for analysis of spinal cords following peak disease treatment (panels F–K). (F) Representative images of sectioned spinal cords harvested from sham-injected RR-EAE (Left) or PLP/Rapa MP-treated RR-EAE mice (Right). Sham/RR-EAE, n = 8 mice, PLP/Rapa MP/RR-EAE, n = 7. (G) Average number of lesions per FOV per mouse. (H) Quantification of CD4 T cells in spinal cords. (I) Quantification of Rorγt+ CD4 T cells (T_H17s) in spinal cords, number (Left) or frequency (Right). (J) Quantification of CD25hi Foxp3+ CD4 T cells (T_regs) in spinal cords, number (Left) or frequency (Right). (K) Ratio T_reg:T_H17 in spinal cords. Sham, n = 9. PLP/Rapa, n = 10. Boxes in images denote insets. (Scale bars, 50 µm.) Error bars represent mean ± SD. ns = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by Student’s t test or one-way ANOVA Tukey−Kramer multiple comparisons test.

I.ln. MP Treatment Alters the Inflammatory-To-Regulatory T Cell Balance in the CNS.

Lesion formation is partially mediated by infiltrating autoreactive inflammatory T cells (e.g., T_H17) (26). Thus, we hypothesized local reprogramming of LNs alters T cell populations in the CNS. A single dose of i.ln. PLP/Rapa MPs decreased the number of CD45hi cells (SI Appendix, Fig. S7 A–C), a marker of infiltrating leukocytes, and the number of CD4 T cells in spinal cords when administered as an early intervention (SI Appendix, Fig. S7D) or during peak disease (Fig. 4H and SI Appendix, Fig. S7E). Among CD4 T cells, peak disease treatment (day 14) reduced the frequency and number of T_H17s in spinal cords during relapse (Fig. 4I and SI Appendix, Fig. S7F). While we observed a decrease in the total number of T_regs in treated cohorts—consistent with the overall reduction in CD4 T cell infiltrates in treated mice—the T_reg frequency was unchanged in mice that received PLP/Rapa MPs (Fig. 4J and SI Appendix, Fig. S7F). These data suggest that among the CD4 T cells that traffic to spinal cords in treated mice, albeit less, a larger proportion are regulatory in phenotype—indicated by an increase in a T_reg:T_H17 ratio that takes into account the overall reduction in cell infiltrates (Fig. 4K). Together, these data reveal that antigen-specific reprogramming in LNs alters T cell populations that reshape inflammation and eliminate lesions in distal CNS tissue sites relevant for MS, without systemic or repeated dosing.

CD4 T Cell Trafficking Is Altered by Rapa-Mediated Transcriptional Regulation in Treated LNs.

Given the overall reduction in the number of infiltrating CD4 T cells recovered from the CNS of treated mice (Fig. 4H), we next investigated whether i.ln. treatment impacts trafficking of antigen-specific cells. This hypothesis is motivated by work showing that mTORC1 signaling, which is inhibited by rapamycin, controls expression of homing receptors in activated T cells (27). To test this, we used a primary in vitro coculture system of splenic dendritic cells (DCs) and 2D2 T cells that allows examination of receptor expression in a closed environment. While MOG induced downregulation of CD62L in CD4 T cells, T cells pretreated with MOG/Rapa MPs maintained CD62L expression (Fig. 5A and SI Appendix, Fig. S8A), a LN homing molecule that is lost following T cell receptor engagement. Notably, CD44 expression was similar among treatment groups (Fig. 5A and SI Appendix, Fig. S8B). These data suggest that Rapa in MPs could alter trafficking without impacting activation.

Fig. 5.

Trafficking of CD4 T cells is altered by Rapa-mediated transcriptional regulation in treated LNs. (A) Flow cytometric analyses of CD44 and CD62L surface marker expression, unstimulated or three days after start of DC:T cell in vitro cocultures. Quantification of (B) Klf2 mRNA and (C) gene expression analysis of 2D2 T cells from in vitro DC:T cocultures, expressed as relative to untreated. (D) Frequency of 2D2 T cells in treated, inguinal LNs six days post transfer into MP-treated recipients. (E) Frequency of 2D2 T cells in the blood. Connected points represent paired data from individual mice over time. (F) Frequency of 2D2 T cells in untreated tissues, six days post transfer; cervical LN (Left panel) and spleen (Right panel). (G) Gene expression analysis of sorted CD4 T cells pooled from inguinal LNs, expressed as relative to untreated. n = 4 to 5/treatment. Error bars represent mean ± SD, Tukey. ns = not significant, *P < 0.05, **P < 0.01, ****P < 0.0001 by Student’s t test or one-way ANOVA Tukey−Kramer multiple comparisons test.

CD62L expression is regulated by the transcription factor, Kruppel-like factor 2 (KLF2), that is suppressed by mTORC1 signaling (27, 28). To test whether MPs exert transcriptional control over KLF2, we performed real-time qPCR (RT-qPCR) analysis on CD4 T cells. Treatment with MOG/Rapa inhibited downregulation of Klf2 compared with antigen-only controls (Fig. 5B). Gene expression profiling further revealed altered expression of KLF2 targets: Ccr7, and S1p1r that control egress of T cells from LNs (Fig. 5C) (29). These data indicate MOG/Rapa MPs transcriptionally regulate Klf2 and associated homing molecules in antigen-specific CD4 T cells.

To test whether i.ln. MPs alter trafficking of activated CD4 T cells in vivo, 2D2 T cells were adoptively transferred into congenic recipients i.ln.-treated with either MOG or MOG/Rapa MPs. While the frequency (Fig. 5D) and number (SI Appendix, Fig. S8C) of antigen-specific T cells in treated LNs were similar among groups six days after transfer, we measured significant differences in blood (Fig. 5E and SI Appendix, Fig. S8D) and untreated tissues (i.e., cervical LNs and spleen) (Fig. 5F). Gene expression profiling performed on CD4 T cells isolated from treated LNs revealed Rapa in MPs limited Klf2 and Sell downregulation. Correspondingly, reduced downregulation of Ccr7 expression and to a lesser extent, S1p1r, were observed (Fig. 5G). Although MOG/Rapa MPs limited proliferation of T cells relative to antigen-only i.ln. treatment on day 3 (SI Appendix, Fig. S8E), when dye could be reliably used to track division, these data indicate that differences in circulation are not solely due to changes in proliferative capacity. Thus, peptide/Rapa MPs exert local control over transcriptional programs that alter trafficking of activated T cells in treated tissues and potentially limit peripheral damage by autoreactive cells.

Healthy, Systemic Responses to Vaccination are Intact Following i.ln. MP Treatment.

Antigen-specific immunotherapies that preserve healthy immunity by limiting off-target effects would be disruptive for the MS therapeutics landscape. To test whether i.ln. MPs leave proinflammatory responses to foreign antigen intact, mice with EAE were simultaneously treated with i.ln. MPs and subcutaneously immunized using a vaccine composed of ovalbumin (OVA) protein admixed with CpG adjuvant (30) (Fig. 6A). The efficacy of MP treatment was unaffected by vaccination (Fig. 6B and SI Appendix, Fig. S9A) and was accompanied by a reduction in antigen-specific recall responses during MOG peptide restimulation of cells from treated LNs that is suggestive of tolerance (Fig. 6C). Importantly, MP treatment did not inhibit OVA-specific vaccine responses, measured by anti OVA IgG antibody production (Fig. 6D and SI Appendix, Fig. S9B) and tetramer+ CD8 T cell frequency (Fig. 6E and SI Appendix, Fig. S9C), or suppress IFNγ production in response to OVA peptide restimulation (Fig. 6F and SI Appendix, Fig. S9D). These data demonstrate local conditioning of LNs with MOG/Rapa MPs drives systemic, myelin-specific tolerance without hindering simultaneous protective, antigen-specific responses to vaccine challenge. Technologies offering this selectivity would address important risks faced by patients receiving existing autoimmune therapies.

Fig. 6.

I.ln. MP treatment does not impact healthy immune responses to vaccination. (A) Timeline for treatment and analysis of immune responses in treated, vaccinated, EAE mice. (B) Clinical score. (C) IFNγ secretion by cells recovered from inguinal (Left) or axillary/brachial (Right) LNs following MOG_35-55 restimulation. (D) Anti OVA IgG in serum over time. (E) Representative gating strategy (Left) and quantification (Right) of the CD8 SIINFEKL tetramer+ cells over time. (F) IFNγ secretion by cells recovered from inguinal (Left) or axillary/brachial (Right) LNs following OVA_323-339 restimulation. Sham, n = 9. MOG/Rapa MPs, n=10. Clinical scores represent mean ± SEM. Error bars represent mean ± SD. ns = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by Steel–Dwass test (clinical score), two-way ANOVA with Tukey–Kramer post hoc or Student’s t test.

Lyophilized, Stored MPs Retain Molecular and Immunological Fidelity.

CMC considerations are vital to ensure therapeutics are well-defined and intact during synthesis, storage, and administration. We lyophilized and stored MOG/Rapa MPs for 90 d before reconstitution (31), then assessed the biophysical and biochemical properties of MP cargo. Comparison of functional groups using NMR (SI Appendix, Fig. S10 A–C) and mass spectrometry (SI Appendix, Fig. S11 A and B) revealed molecular equivalency between free, soluble MOG and Rapa standards and their counterparts recovered from MPs after lyophilization, storage, and reconstitution (SI Appendix, Table S2). Likewise, equivalent efficacy in EAE was observed between reconstituted and freshly prepared MPs using early (SI Appendix, Fig. S12A) or peak disease (SI Appendix, Fig. S12B) interventions. These studies support favorable CMC characteristics for patient translation.

I.ln. Delivery of MPs Exhibits a Favorable Safety Profile in NHPs.

Finally, we tested whether i.ln. MPs avoid safety and tolerability concerns associated with systemic therapies [e.g., mild, transient hepatoxicity and nephrotoxicity observed with mTOR inhibitors, attributable to high trough drug levels (32) or hyperinsulinemic hypoglycemia (33, 34)]. NHPs were treated i.ln. with MPs coloaded with Rapa and either OVA or keyhole limpet hemocyanin (KLH)—a prototypic T-helper-dependent immunogen commonly used in NHP and human studies (35) (Fig. 7A and SI Appendix, Table S3). Single-dose treatment did not alter weight (Fig. 7B), liver (ALT, AST; Fig. 7C), or kidney function (BUN, creatinine; Fig. 7D), suggesting that i.ln. MP treatment exhibits favorable safety in NHPs. We also tested whether introduction of antigen to LNs causes unwanted proinflammatory responses by challenging MP-treated monkeys intramuscularly with matched antigen in adjuvant (Fig. 7A). Hematocrit, platelets, and white blood cell count (WBC) remained within normal ranges (Fig. 7E). No gross perturbations indicative of a systemic inflammatory response were detected, as circulating neutrophils, monocytes, lymphocytes (Fig. 7F), and lymphocyte subsets (SI Appendix, Fig. S13 A–D) were unchanged. Additionally, intact initial and secondary antibody responses to vaccination were observed (SI Appendix, Fig. S13 E and F).

Fig. 7.

I.ln. delivery of MPs exhibits a favorable safety profile in non-human primates. (A) Timeline for cynomolgus macaque safety study. (B) Relative change in weight over time. (C) Liver function parameters—alanine transaminase (ALT) (Top) and aspartate aminotransferase (AST) (Bottom). (D) Kidney function parameters—blood urea nitrogen (BUN) (Top) and creatinine (Bottom). (E) Total percent hematocrit (Top), platelets (Middle), and white blood cells (WBC) (Bottom). (F) Total circulating neutrophils (Left), monocytes (Middle), and lymphocytes (Right). Untreated, n = 4 to 6; KLH/Rapa MPs, n = 2; OVA/Rapa MPs, n = 4.

Interestingly, one animal immunized with KLH/Rapa MPs exhibited an IgM response following treatment but before vaccination (SI Appendix, Fig. S13E). However, the number of observations is insufficient to assess sensitizing effects. Thus, we examined this possibility of sensitization by assessing B cell responses in mice. To test this, healthy mice were treated i.ln. with MPs assembled with Rapa and either OVA_323-339 peptide or whole OVA protein. MPs did not induce anti OVA IgG, compared with healthy, untreated mice or control mice vaccinated with OVA plus CpG using a prime/boost strategy that produces detectable circulating antibodies (SI Appendix, Fig. S13G). Collectively, these data suggest i.ln. MPs appear safe and unlikely to trigger pathogenic autoimmunity (36).

Discussion

Our data suggest a model (SI Appendix, Fig. S14) where antigen-specific T cells in MP-treated LNs undergo TCR engagement in a localized manner. Rapa creates tolerizing conditions that favor T_reg polarization (37–40) and impact trafficking of activated T cells, limiting damage of peripheral tissue. Together, these findings provide insight into the immunological mechanisms contributing to the durable, potent efficacy observed in preclinical models of MS following single-dose treatment. This approach, which relies on LN depots to circumvent systemic dosing that suffers from dose-limited efficacy, off-target effects, and hepatic clearance (41, 42), advances an alternative paradigm to existing immunotherapies that could offer correspondingly disruptive outcomes in patients relative to past clinical trials seeking antigen-specific tolerance through systemic dosing.

Emerging strategies to induction of T_reg-mediated tolerance for MS treatment (43, 44) have shown promise in Phase 1 trials (23, 45, 46), but less success in Phase II (36, 47) and III (48). I.ln. delivery provides unique flexibility for precise, coordinated administration of multiple signals considered classical drivers of tolerance—local display of (self) antigen in the absence of inflammation or in the presence of regulatory milieus (49). Here, we establish these localized signals not only support existing T_regs but generate antigen-specific T_regs in vivo that exhibit enhanced suppressive function. Persistent T_regs are also associated with long-lasting effects in distal, disease-relevant tissues. Future pharmacokinetic (PK) studies examining biodistribution and release rates of tolerogenic signals from MPs in LNs could help inform dose selection and treatment regimen. This information, coupled with potential roles of memory T_reg induced by treatment could help inform the durability of therapeutic effects and if retreatment might be beneficial. Notably, induction of tolerogenic memory is an exciting area of growing interest that i.ln depots may be able to exploit in pursuit of durable remission (50–52).

In addition, we show that localizing cues made possible by i.ln. MPs lead to altered trafficking of activated CD4 T cells that would also mitigate destruction of peripheral tissues. The relevance of this mechanism is evidenced by approved DMTs that inhibit T cell migration in MS [e.g., fingolimod (53, 54)]. Our data suggest that altered mTORc1 signaling by Rapa releases inhibition of KLF2, resulting in maintenance of CD62L and reduced downregulation of CCR7 (27) that promote LN homing and retention. It is not known at this time whether the contribution of CCR7 expression to LN homing in our system outweighs the effects of S1PR1 (55), which promotes T cell egress and is qualitatively less affected. Importantly, the localized retention of T cells mediated by MP depots does not completely abrogate trafficking, and mice exhibit healthy immunity following vaccination, unlike select systemic interventions that globally prevent lymphocyte egress and leave patients immunocompromised.

Our studies demonstrate i.ln. treatment is safe in NHPs. A simple surgical procedure was utilized to access inguinal LNs for injections; clinically, this could be replaced by image-guided injection enabling a minimally invasive approach using common technologies employed in biopsies (56). Clinical trials examining the safety and efficacy of intralymphatic immunotherapies (ILIT) for allergen sensitization (57, 58) and treatment of type 1 diabetes (59, 60) have defined target injection volumes (e.g., 0.1 to 0.4mL/LN) and dose ranges (e.g., 12ug GAD-alum/regimen). Injection site decisions will be driven by existing safety findings and accessibility of nodes. Notably, our data have shown that treatment of inguinal LNs in NHPs is safe and confers potent efficacy in mice—the most common set of nodes targeted by existing ILITs. While dose selection and frequency utilized for soluble therapies may not fully align with ILITs using intralymphatic MP depots, these clinical studies using soluble cargo with ILITs lay a solid foundation for future trials testing intra lymphatic MP depots to stop autoimmune disease.

One obstacle associated with existing MS therapeutics is loss of efficacy during severe disease activity, which patients and physicians report as the primary reason for switching DMTs (61). We observed beneficial effects of a single treatment at all intervention times, including during late/chronic disease. These data suggest that MP treatment may also impact remyelination mediated by innate cells that are critical for successful therapeutic outcomes in MS patients with progressive disease (62).

Finally, a persistent hurdle for antigen-specific MS immunotherapy is the identification of definitive target autoantigen(s). However, utilization of encephalitogenic peptides derived from myelin basic protein (MBP) (e.g., MBP_83-99), proteolipid protein (e.g., PLP_139-154), and myelin oligodendrocyte glycoprotein (e.g., MOG_1-20, MOG_35-55) exhibit immunodominance and have shown promise in MS clinical trials (23, 25, 63, 64). These peptides, and others that continue to surface [e.g., GDP L-fucose synthase (65), TSTA3 (66)], provide opportunities for targeting. Additionally, our data demonstrate that the i.ln. MP platform allows for facile exchange of peptides (and proteins) during assembly (e.g., MOG, PLP, OVA, KLH). Incorporation of multiple, diverse antigens could help to address the phenomenon of epitope spread in autoimmunity (29). Related to this, we have previously shown that Rapa MPs assembled with a CD4 peptide relevant for type 1 diabetes offer protection in a CD8-mediated mouse model of disease (49). Furthermore, i.ln. delivery of whole protein in MPs also represents a potential strategy for the treatment of autoimmune diseases where B cells and/or autoantibodies mediate disease (67, 68). Thus, diffusion-limited lymph node depots create a potential platform to induce specific and lasting-tolerance in MS or other autoimmune disease.

Materials and Methods

Materials were assembled as previously described [15, 31]; a detailed description of parameters for particle synthesis, characterization, and biochemical and biophysical analyses (e.g., LC-MS, ¹H-NMR) can be found in the Extended Materials and Methods sections of SI Appendix. Cell isolation and culture conditions, including protocols for ELISAs and RT-qPCR, are described in SI Appendix. Data from flow cytometric phenotyping and related assays (e.g., T_reg suppression, restimulation) were acquired on a BD FACSCelesta or CytoFlex (Beckman Coulter) and analyzed using FlowJo (BD). Cell sorting for adoptive transfers was performed on a BD FACSAria II. For immunofluorescence microscopy, spinal cords were sectioned, stained, and imaged using an Olympus IX83 microscope, as described in SI Appendix. All animal work (e.g., vaccination, induction/monitoring for EAE and RR-EAE preclinical MS mouse models, NHP studies) was performed using protocols approved by the Institutional Animal Care and Use Committee (IACUC) overseen by either the University of Maryland at College Park or the University of Maryland School of Medicine. Additional reagent lists and descriptions of statistical analyses are also included in subsections of the Extended Materials and Methods of SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.