Design of Polyelectrolyte Multilayers to Promote Immunological Tolerance.

Recent studies demonstrate that excess signaling through inflammatory pathways (e.g., toll-like receptors, TLRs) contributes to the pathogenesis of human autoimmune diseases, including lupus, diabetes, and multiple sclerosis (MS). We hypothesized that codelivery of a regulatory ligand of TLR9, GpG oligonucleotide, along with myelin-the "self" molecule attacked in MS-might restrain the pro-inflammatory signaling typically present during myelin presentation, redirecting T cell differentiation away from inflammatory populations and toward tolerogenic phenotypes such as regulatory T cells. Here we show that myelin peptide and GpG can be used as modular building blocks for co-assembly into immune polyelectrolyte multilayers (iPEMs). These nanostructured capsules mimic attractive features of biomaterials, including tunable cargo loading and codelivery, but eliminate all carriers and synthetic polymers, components that often exhibit intrinsic inflammatory properties that could exacerbate autoimmune disease. In vitro, iPEMs assembled from myelin and GpG oligonucleotide, but not myelin and a control oligonucleotide, restrain TLR9 signaling, reduce dendritic cell activation, and polarize myelin-specific T cells toward tolerogenic phenotype and function. In mice, iPEMs blunt myelin-triggered inflammatory responses, expand regulatory T cells, and eliminate disease in a common model of MS. Finally, in samples from human MS patients, iPEMs bias myelin-triggered immune cell function toward tolerance. This work represents a unique opportunity to use PEMs to regulate immune function and promote tolerance, supporting iPEMs as a carrier-free platform to alter TLR function to reduce inflammation and combat autoimmunity.

A utoimmune diseases such as multiple sclerosis (MS), type 1 diabetes, and lupus occur when self-molecules are incorrectly recognized as foreign and attacked by the immune system. New studies reveal that excess inflammation through toll-like receptors (TLR), a class of inflammatory pathways typically triggered by foreign pathogens, strongly contributes to the pathology of autoimmune diseases. 1−8 Recent reports also demonstrate that co-administration of regulatory signals and self-molecules attacked during autoimmunity, myelin in MS, for example, can promote tolerance. 9 Biomaterial carriers have been explored to facilitate this codelivery, 9−11 but these materials exhibit intrinsic features that can activate inflammatory pathways that could exacerbate autoimmune disease. 12−14 To address these challenges, here we designed immune polyelectrolyte multilayer (iPEM) capsules that promote tolerance by mimicking features of biomaterials, but that are self-assembled entirely from myelin and a regulatory TLR antagonist. 15,16 Existing therapies for autoimmune diseases are not curative and employ broad immunosuppression that often leaves patients immunocompromised. 17 These limitations have motivated the exploration of alternative approaches, including vaccine-like strategies that offer the potential for both efficacy and specificity by, for example, expanding regulatory T cells (T REGS ) specific for self-molecules attacked during MS or other autoimmune diseases. Importantly, the signaling milieu in which myelin is processed and presented by antigen presenting cells plays a major role in programming the balance between autoimmunity and tolerance. 10 Thus, strategies that allow precise control over the relative composition of self-antigen and therapeutic components could be transformative in enabling more specific treatments for autoimmune disease. Biomaterials offer attractive features to achieve this goal, including codelivery, and have recently been studied to deliver and target tolerogenic drugs, cytokines, and self-antigen to key immune tissues (e.g., lymph nodes, spleen). 10,11,18−21 PEMs are particularly well-positioned for this application; these materials are self-assembled through a layer-by-layer process that enables juxtaposition of multiple signals with tunable, stepwise control over the absolute and relative loading of several cargos. 22−24 However, despite ubiquitous application to drug delivery and vaccination, 24−30 PEMs have not been studied to regulate immune function or promote immunological tolerance. We recently described a layer-by-layer approach to assemble model antigens and TLR-based adjuvants into vaccines that drive strong, pro-immune T cell function. 31,32 These nanostructures, termed iPEMs, were assembled through electrostatic interactions using model antigen and a nucleic acid-based TLR3 agonist that served as a stimulatory adjuvant. Thus, iPEMs are composed entirely of immune signals (i.e., antigen, TLR agonist), in contrast to traditional PEMs integrating synthetic or natural polymers. 32 We hypothesized the iPEM platform would be advantageous for tolerance by enabling codelivery of self-antigens and tolerogenic immune cues at high concentrations, while eliminating carrier components, such as poly(lactide-co-galactide) (PLGA), that have recently been shown to activate inflammasomes and other pro-inflammatory pathways that could increase the severity of autoimmune disease. 12−14 During MS, myelin-specific pathogenic CD4 + T cells (e.g., T H 1, T H 17) and antibodies infiltrate the central nervous system (CNS) to drive inflammation and demyelination. Thus, to Representative fluorescence microscopy images of quartz substrates, following deposition of eight MOG-R 3 /GpG bilayers (green, FITC-MOG-R 3 ; red, Cy5-GpG; scale, 20 μm). A needle was used to remove a portion of the film to provide contrast (dotted white lines). Atomic force microscopy analysis of the surface morphology of (MOG-R 3 /GpG) 8 iPEMs (x−y scale, 3 μm; z scale, 400 μm). (C) Stepwise measurements of iPEM thickness with increasing numbers of MOG-R 3 /GpG bilayers deposited on silicon substrates, quantified by ellipsometry. Data represent mean ± SEM (n = 3). (D) Spectrophotometric analysis of relative loading of MOG-R 3 (500 nm) and GpG (260 nm) as a function of the number of bilayers deposited on quartz substrates. Data represent mean ± SEM (n = 3). (E) Representative images (top) and quantitative pixel analysis (bottom) of relative GpG loading with subsequent iPEM deposition cycles on a calcium carbonate microparticle core (red, Cy5-GpG; scale, 2 μm). (F) Spectrophotometric quantification of tunable relative composition of iPEMs as a function of the relative input mass of MOG-R 3 and GpG into iPEM synthesis. Data represent mean ± SEM (n = 4 for each input ratio). (G) Representative fluorescence microscopy images of hollow iPEM capsules composed entirely of either MOG-R 3 and GpG (top) or MOG-R 3 and a nonimmunoregulatory oligonucleotide (CTRL, bottom), following removal of the calcium carbonate core by incubation in EDTA (green, FITC-MOG-R 3 ; red, Cy5-GpG; magenta, Cy3-CTRL; scale, 2 μm).

ACS Nano
Article design iPEMs for tolerance, we selected a myelin peptide implicated in human MS and animal models, myelin oligodendrocyte glycoprotein (MOG). 33 Two discoveries motivated our approach to assemble this peptide with a nucleic acid-based regulatory ligand of TLR9, GpG ( Figure 1A). First, new research reveals signaling through TLRs plays an important role in driving disease during MS, lupus, and type 1 diabetes; 1−8 blunting this signaling during T cell expansion might reduce inflammation and expand T cell phenotypes (e.g., T REGS ) specific for myelin, but that selectively control disease. Second, seminal work by the Steinman lab demonstrates TLR9specific activity of GpG and that systemic administration helps promote tolerance. 15,16 While the mechanism of GpG has not been fully elucidated, these reports support a role for competitive binding with TLR9, polarization away from inflammatory cells, and perhaps in parallel, the promotion of a protective T H 2-like response to restrain inflammatory immune cell activity. 15 In contrast, the agonistic TLR9 ligand, CpG, is under intense study as a vaccine adjuvant for infectious disease and cancer. 34 Together, these discoveries suggest codelivery of self-antigens with signals to modulate TLR9 or other TLR pathways could efficiently promote tolerance. Thus, we formed iPEMs by exploiting the polyanionic character of GpGwhich is an oligonucleotide with a phosphorothiate backbone structurally similar to CpG, but in which cytosine residues are replaced by guanine residuesfor electrostatic assembly with MOG conjugated to arginine residues as cationic anchors. We hypothesized that juxtaposition of these tolerogenic signals in the nanostructure of iPEMs would mute TLR9 signaling during antigen presentation to bias differentiating T cells away from inflammatory function and, instead, promote regulatory T cells (T REGS ) to control autoimmunity. In vitro, we show iPEMs codeliver both cargos to dendritic cells (DCs), downregulate TLR9 signaling, restrain DC activation, and polarize antigen-specific T cells toward T REGS . In mice, iPEMs reduce inflammation, expand T REGS , and eliminate disease in a common mouse model of MS. Using samples from human MS patients, we discovered iPEMs bias the function of T cells toward tolerance. These results demonstrate that PEMs can be used to regulate immune function and promote tolerance.

RESULTS
iPEM Assembly and Characterization. iPEMs were first deposited on planar substrates to characterize the assembly of GpG with MOG conjugated to triarginine (MOG-R 3 ). Fluorescence microscopy was used to visualize the colocalization of both cargos during assembly of 16 layers to form (MOG-R 3 /GpG) 8 ( Figure 1B). Both fluorescence and atomic force microscopy ( Figure 1B) revealed surface topography with a root-mean-square roughness of 51.6 nm. Next we confirmed film thickness and cargo loading could be controlled by varying the number of deposition cycles. Ellipsometry revealed increasing iPEM thicknesses to a value of ∼150 nm after 8 bilayers ( Figure 1C). Similarly, spectrophotometric analysis revealed linearly increasing absorbance values at characteristic wavelengths for GpG (R 2 = 0.99) and fluorescently labeled MOG-R 3 (R 2 = 0.98) ( Figures 1D and S1). These results indicate the loading of each signal can be tuned through the number of bilayers deposited.
To facilitate cell and animal studies, we assembled iPEMs on sacrificial calcium carbonate templates 35,36 then used a chelator to dissolve the core and create support-free MOG-R 3 /GpG iPEM capsules. Consistent with our results on planar substrates, fluorescence microscopy ( Figure 1E, top) and quantitative analysis of pixel intensity ( Figure 1E, bottom) indicated increasing GpG loading as more layers were deposited. Zeta potential measurements revealed a corresponding oscillation in surface charge, indicative of electrostatically driven layer-bylayer assembly ( Figure S1). We next investigated whether the cargo loading of iPEMs could be tuned, as the combinations and relative doses of immune signals play a major role in determining the magnitude and polarization of antigen-specific response. 37 We discovered the relative loading of MOG-R 3 and GpG could be directly controlled over a range from 89.7 ± 0.4% MOG-R 3 and 10.3 ± 0.4% GpG to 28.4 ± 0.7% MOG-R 3

ACS Nano
Article and 71.6 ± 0.7% GpG ( Figure 1F, Table S1) by altering the relative mass input during adsorption. Another notable characteristic of this approach is that 100% of the PEM coating is comprised of MOG-R 3 and GpG; therefore, following incubation of (MOG-R 3 /GpG) 3 -coated templates in EDTA, hollow capsules comprised entirely of immune signals were formed ( Figure 1G, top) with negligible loss of either cargo following core removal ( Figure S1). Control iPEM capsules, (MOG-R 3 /CTRL) 3 , could also be assembled from myelin peptide and an inactive control oligonucleotide (CTRL) ( Figure 1G, bottom). These iPEMs exhibited myelin and oligonucleotide loading similar to that of (MOG-R 3 /GpG) 3 iPEMs ( Figure S1). Using these sets of iPEM architectures, we next tested if self-antigens and regulatory cues incorporated into iPEMs promote tolerogenic functions in DCs and T cells.
iPEMs Restrain TLR9 Signaling and DC Activation. We hypothesized iPEMs assembled from MOG and GpG might polarize T cells away from effector cells and toward T REGS by reducing TLR9 signaling during the differentiation of myelinspecific T cells being expanded by antigen presenting cells (e.g., DCs). To investigate this idea, we first tested if iPEMs codeliver both signals to DCs by culturing dual-labeled (MOG-R 3 / GpG) 3 iPEMs with primary splenic DCs. In these studies, flow cytometry revealed a dose-dependent uptake of capsules without toxicity ( Figure S2) and, interestingly, irrespective of iPEM dose, >80% of cells positive for at least one component were positive for both MOG-R 3 and GpG (Figure 2A,B). In contrast, incubation of DCs with admixed MOG-R 3 and GpG resulted in dramatically reduced (<45%) codelivery of cargos compared with matched doses formulated as iPEMs ( Figure  S3). Promoting codelivery is of particular interest for autoimmune therapy as the administration of regulatory signals (e.g., GpG) in the absence of self-antigen may drive nonspecific immunosuppression, while delivery of self-antigen alone creates a risk of triggering self-reactivity that exacerbates disease. 10 In separate studies, MOG-specific CD4 + T cells were cocultured with wild-type DCs, incubated with CpG and indicated iPEM forumations, and the frequency of T REGS , CD4 + /CD25 + Foxp3 + cells, was analyzed by flow cytometry using the gating scheme shown in (G), which was assigned using control samples treated with media alone ( Figure S5). Data in all panels indicate mean ± SEM for studies conducted in triplicate. In panels B and D−F, data were analyzed with one-way ANOVA with a Tukey post-test to correct for multiple comparisons. For clarity, only key comparisons are shown: In panels B, E, F, # markers indicate the comparison (# = P ≤ 0.0001) of each group to a strong positive control; wells treated CpG and a high dose of soluble cognate antigen, MOG-R 3 (60 μg, dashed lines). Comparisons between (MOG-R 3 /GpG) 3 and (MOG-R 3 /CTRL) 3 iPEMs are indicated by a bracket (**** P ≤ 0.0001). In panel D, # markers indicate the comparison (# = P ≤ 0.0001) of each group to a control of wells treated CpG and soluble OT-II cognate antigen, OVA 323−339 (dashed line). In panel H, statistics indicate the results of a two-tailed t test (* P ≤ 0.05).

ACS Nano
Article TLR9 typically senses pathogen-associated nucleic acids that are unmethylated and rich in cytosine and guanine residues (e.g., CpG). Interestingly, autoimmune disease in a common mouse model of MS is attenuated when induced in TLR9 knockout mice and completely eliminated in mice deficient in MyD88, a downstream activator of many TLRs. 1 Further underscoring the role of TLR9 in driving autoimmunity, administration of GpG, which consists of a nucleic acid sequence similar to CpG but with cytosine replaced by guanine, reduces self-antigen triggered T cell proliferation and helps control disease in mice. 15,16 Thus, we used TLR9 reporter cells to directly investigate the impact of GpG-containing iPEMs on TLR9 signaling. Treatment with CpG (TLR9 agonist), but not TLR3 or TLR4 agonists, drove high levels of TLR9 activity in reporter cells relative to untreated control wells ( Figure 2C). The addition of soluble GpG or (MOG-R 3 /GpG) 3 iPEMs to CpG-treated wells significantly reduced TLR9 signaling, while a control nucleotide, CTRL, in either soluble form or assembled into iPEMs with MOG-R 3 had no effect. To investigate the impact on DC function, splenic DCs were next isolated, activated with CpG, and treated with (MOG-R 3 /GpG) 3 iPEMs, (MOG-R 3 /CTRL) 3 iPEMs, or soluble components. Both soluble GpG and (MOG-R 3 /GpG) 3 iPEMs downregulated expression of prototypical DC activation markers, CD40 ( Figures 2D and S4) and CD86 ( Figures 2E and S4). In contrast, soluble MOG-R 3 , soluble CTRL, and (MOG-R 3 / CTRL) 3 iPEMs did not impact CpG-induced activation. Together our results in Figure 2C−E suggest that MOG and GpG can be assembled into iPEMs without supports or carriers and without impacting the selectivity of GpG. However, we noted that the dose of GpG in iPEMs (∼6 μg/well) did not reduce the level of TLR9 signaling or expression of surface markers of DC activation to the same degree as the high dose of soluble GpG (10 μg/well) included in the positive control reference sample. Thus, we conducted studies over a range of matched doses of GpG in soluble or iPEM form to enable direct comparison of potency. We discovered that soluble GpG exhibited more potent restraint of TLR9 signaling ( Figure S5) and DC activation ( Figure S5) compared with GpG-containing iPEMs. While future studies could test if this result may be due to an increase in the required processing time following electrostatic complexation of GpG, iPEMs improved codelivery to cells (Figures 2 and S3) and caused significant attenuation of inflammatory cell activity in vitro (Figures 2C−E and S5). Further iPEMs offer unique advantages, codelivery and tunable compositions, for example, relative to soluble components or simple mixtures, and these are features of particular importance for autoimmune therapies in animal models or human patients. Thus, we next sought to investigate whether cells can properly process and present self-antigen following incorporation into iPEMs and whether codelivery of myelin self-antigen and GpG polarizes myelin-specific T cell responses against this antigen toward tolerance.
iPEMs Polarize Antigen-Specific T Cell Function and Phenotype in vitro. To test whether MOG-R 3 promotes myelin-specific T cell interactions in iPEM form, fluorescently labeled CD4 + MOG-reactive transgenic T cells were added to DCs isolated from wild-type mice and treated with CpG, along with either iPEMs or free components. After 72 h, flow cytometry revealed high levels of proliferation (i.e., dilution of dye) in cultures treated with a positive control of CpG and cognate antigen, soluble MOG-R 3 . Significant proliferation was also observed in wells treated with CpG and either (MOG-R 3 / GpG) iPEMs or (MOG-R 3 /CTRL) 3 iPEMs, but not in wells In similar studies, splenocytes were isolated (n = 4 for both groups) on day 13 and analyzed immediately for the frequency of CD4 + /CD25 + Foxp3 + cells. To assess therapeutic efficacy of iPEMs, mice were induced with EAE and either left untreated (n = 11) or administered iPEMs (n = 10) as in (A) and monitored for (G) mean clinical score, (H) disease-associated weight loss, and (I) incidence of disease. In panels B−D, data were analyzed with one-way ANOVA with a Tukey post-test. In panel F, statistics indicate the results of a two-tailed t test. In panels G and H, data were analyzed with multiple t tests, one at each time point, with a post-test correction for multiple comparisons. Disease incidence in (I) was analyzed with a log-rank test. (* = P ≤ 0.05; ** = P ≤ 0.01; *** = P ≤ 0.001; **** = P ≤ 0.0001).

ACS Nano
Article absent of MOG-R 3 ( Figures 3A,B and S6). Equivalent results were observed over a range of doses ( Figure S7). In contrast, during analogous studies with transgenic CD4 + T cells specific for an irrelevant epitope from ovalbumin (OVA 323−339 ), no proliferation was detected with MOG-R 3 in soluble or iPEM form ( Figures 3C,D and S6), indicating that iPEMs support antigen-specific T cell expansion. Strikingly, when supernatants from cocultures were analyzed by ELISA, we discovered that despite similar levels of proliferation ( Figure 3A,B), (MOG-R 3 / GpG) 3 iPEM-treated cells secreted dramatically lower levels of inflammatory IL-6 ( Figure 3E) and IFN-γ ( Figure 3F) compared with wells receiving (MOG-R 3 /CTRL) 3 iPEMs. Further, in similar coculture studies, (MOG-R 3 /GpG) 3 iPEMs caused a significant increase in the frequency of T REGS (CD4 + / CD25 + Foxp3 + ) compared with (MOG-R 3 /CTRL) 3 treatments ( Figure 3G,H). This polarization was observed in studies performed in the presence of self-antigen alone (i.e., without CpG; Figure S8) and when both self-antigen and a strong TLR9 agonist were present ( Figure 3G,H). Together, these differences in the processing of MOG-R 3 /GpG and MOG-R 3 / CTRL formulations indicate that iPEM-loaded MOG-R 3 triggers antigen-specific T cell proliferation, but that inclusion of GpG in iPEMs directs these cells toward T REGS and away from key inflammatory functions that drive autoimmune disease.
iPEMs Restrain Inflammatory Immune Cell Function and Halt a Model of MS in vivo. We next tested if iPEMs promote tolerance in mice. Lymph nodes and spleen are important tissues in this context, as these are the sites where differentiating T cells develop toward inflammatory or regulatory phenotypes. Recent studies, for example, reveal that distinct structural microdomains form in these tissues to promote tolerance whereas other domains form to support proimmune function, depending on the cells, structural elements, and immune signals present in each local microenvirnoment. 38 The ability to codeliver self-antigen and regulatory signals to these tissues without inflammatory components could create opportunities to promote tolerance by reprogramming the local signaling milieu and expanding T REGS that migrate to sites of disease (e.g., CNS) to control inflammation. Thus, we investigated if iPEMs deliver cargos to lymphoid tissues and effectively restrain self-antigen triggered inflammatory cytokine secretion at these sites. 48 h after subcutaneous injection, histological analysis revealed an accumulation of iPEMs in the subcapsular sinus of draining lymph nodes and distribution in the paracortex, while no signal was observed in lymph nodes from naïve mice ( Figure S9).
To test if iPEMs redirect inflammatory response to myelin self-antigen, mice were induced with a common myelin-driven model of MS, experimental autoimmune encephalomyelitis (EAE). 11,19,39 In this model, untreated mice develop severe paralysis over the course of several weeks as the CNS is attacked by infiltrating T cells. Mice were either left untreated or treated with (MOG-R 3 /GpG) 3 iPEMs 5 and 10 days after inducing EAE ( Figure 4A). Three days after the second treatment (i.e., day 13), splenocytes were isolated and pulsed with either MOG or irrelevant OVA 323−339 peptide. As expected, because myelin-specific T cells drive disease during EAE, in cells from untreated mice, MOG pulse increased the secretion of inflammatory cytokines IL-17 ( Figure 4B), IFN-γ ( Figure 4C), and IL-6 ( Figure 4D) compared with identical cultures pulsed with OVA. In contrast, restimulating cells from iPEM-treated mice with MOG peptide did not increase inflammatory cytokines and, instead, resulted in the baseline levels of secretion measured in cells from either group pulsed with OVA. Similar effects were also observed upon restimulation of cells isolated from axillary lymph nodes ( Figure S10). These results indicate MOG-R 3 /GpG capsules strongly blunt myelin-triggered inflammatory recall response.
To investigate whether differences in cytokine secretion were due to polarization of T cell phenotype, cells were isolated from treatment groups analogous to those above and stained immediately (i.e., without restimulation) for T REG markers. As with in vitro studies, iPEMs significantly increased CD4 + / CD25 + Foxp3 + T REGS ( Figure 4E,F), and we also observed a trend of increased frequency of CD4 + /CD25 − Foxp3 + cells ( Figure 4E), though this latter result was not statistically significant. These findings support the hypothesis that (MOG-R 3 /GpG) 3 iPEMs polarize T cells away from inflammatory subtypes and toward tolerogenic phenotypes. Next we assessed the functional impact of this T cell biasing by inducing mice with EAE and administering (MOG-R 3 /GpG) 3 iPEMs on days 5 and 10. Each cohort was then monitored for clinical disease symptoms. Untreated mice developed severe paralysis (mean clinical score = 2.95 ± 0.30; Figure 4G), experienced dramatic weight loss ( Figure 4H), and exhibited a high disease incidence of 87.5% ( Figure 4I). In stark contrast, iPEM treatment completely eliminated EAE, with (MOG-R 3 /GpG) 3 iPEMtreated mice remaining asymptomatic (i.e., clinical score = 0, disease incidence = 0%) for the duration of the study without evidence of indirect symptoms such as weight loss ( Figure 4G− I).
iPEMS Attenuate Inflammatory Response in Human MS Patient Samples. Building on our findings in mouse models of MS, we explored whether the tolerogenic effects of iPEMs would extend to human MS patient samples. Recent studies in MS patients confirm that signaling in TLR9 and other TLR pathways drives inflammation and disease via cytokines that promote T H 1 and T H 17 polarization. 2,3,6 Thus, modulating signaling through this pathway could restrain selfattack in human disease. To begin investigating this possibility, we drew on an approach used in recent clinical trials: ex vivo restimulation of peripheral blood mononuclear cells (PBMCs) from MS patients to test for a reduction in myelin-triggered recall responses conferred by experimental therapies. 33 In our studies, we tested this impact by incubating iPEMs with PBMCs collected from three randomly selected MS patients participating in the Veterans Affairs Longitudinal MS (VALOMS) observational study (Table S2). PBMCs were cultured in media (i.e., unstimulated) or with either (MOG-R 3 / GpG) 3 or (MOG-R 3 /CTRL) 3 iPEMs. MTT analysis revealed an increase in metabolic activity after iPEM treatment, irrespective of the sequence of the nucleic acid component ( Figure 5A). This result confirms the expected myelin-triggered increase in cell function associated with myelin-reactive immune cells that develop during MS in humans. However, similar to murine cocultures, despite equivalent levels of metabolic activity, (MOG-R 3 /GpG) 3 iPEMs polarized cytokine profiles away from pro-inflammatory function compared with (MOG-R 3 /CTRL) 3 iPEMs. Strikingly, in nearly every case, TNF-α ( Figure 5B), IL-6 ( Figure 5C), IL-10 ( Figure 5D), and IFN-γ ( Figure 5E) levels were lower when PBMCs were treated with GpG-containing iPEMs compared with CTRL-containing iPEMs, though the specific cytokines where these decreases were statistically significant varied across patients. For example, in Patient 1, a significant reduction of TNF-α, IFN-γ, and IL-10

ACS Nano
Article was observed. Although IL-6 secretion exhibited lower values, the differences were not significant relative to CTRL-containing iPEMs. Together, these results suggest that inclusion of GpG with myelin peptide can restrain inflammatory immune function typically triggered by encountering myelin-reactive cells from human patients. Further, while these samples were from patients with different stages of disease and treatment plans (Table S2), significant effects were observed in all three cases, highlighting the possibility of robust therapeutic effects enabled by iPEMs for future therapies.

DISCUSSION
The ability to assemble biological cargo into PEMs without impacting function, along with direct control of cargo loading and modularity, and the ability to coat onto substrates over a range of length scales are features that have motivated the use of PEMs in applications spanning electronics, 40 optics, 41 drug delivery, 24 and vaccines designed for traditional prophylactic applications (i.e., pro-immune). In the latter, PEMs have been used to encapsulate antigen or adjuvant within PEM shells composed of synthetic polymers, as surfaces to adsorb adjuvants or antigens, and to coat larger-scale vaccine substrates such as arrays of microneedles. 24−30 These directions all seek to amplify the immune response against infectious disease or cancer, but PEMs have never been used to regulate immune function or promote tolerance. This is an area that is particularly interesting for PEMs because one of the growing challenges in the vaccine and immunotherapy field is the increasing complexity of these formulations and the associated difficulties in producing, characterizing, and understanding the mechanism of increasing complex therapeutics. 42 The modular nature of PEMs could help support more rational design of tolerance-inducing therapies through self-assembly of several immune cues to form simpler, more-defined materials.
Modularity is also an important aspect for clinical translation because one of the current challenges facing development of more effective and specific autoimmune therapies is the diversity of antigens (i.e., peptide epitopes) that are attacked during MS and other autoimmune diseases. For example, in MS there are a variety of myelin-based antigens that are incorrectly recognized as foreign, but these vary from patient to patient and can also expand during disease progression through a phenomenon termed epitope spreading. 43 In humans, one recent clinical trial is exploring infusion of MS patients with cells coupled with combinations of myelin peptides to generate antigen-specific tolerance. 33 In the biomaterials field, several exciting approaches are investigating the challenge of epitope spreading in preclinical models of autoimmunity. The Santamaria lab has prepared iron oxide nanoparticles displaying self-peptide in immune protein complexes (major histocompatibility complex, MHC) to expand regulatory cell populations with the capacity to suppress responses against a broader set of self-antigens. 19,44 Miller and co-workers have used PLGA particles displaying self-epitopes to induce tolerance and have discovered that tolerance generated against particles displaying one myelin antigen can also help protect against epitope spreading in mice. 18 iPEMs offer unique features that can be exploited as tools to probe the progression of autoimmune disease and development of tolerance. In particular, iPEMs allow incorporation of one or multiple antigens and regulatory signals with control over the relative compositions by specifying the concentrations or number of layers of each component (Figure 1). Thus, while we selected a single epitope and ratio of myelin:GpG for our studies to investigate the therapeutic efficacy (i.e., 1:2 MOG-R 3 :GpG; Figure 1F and Table S1), this platform could be used to create libraries of materials that exhibit defined concentrations or combinations of epitopes and regulatory signals. This capability might be useful, for example, to investigate how such parameters impact epitope spreading. Relative to other biomaterials being studied in tolerance, iPEMs offer a composition that is simplified by elimination of supports, polymeric carriers, and stabilizers (e.g., poly(vinyl alcohol));

ACS Nano
Article components that often exhibit intrinsic inflammatory functions, as noted above. iPEMs also eliminate recombinant or higher order protein structures such as MHC molecules, instead juxtaposing self-antigen with signals (i.e., GpG) aimed at redirecting the response to a given self-antigen.
In our studies, we observed reduction in myelin-triggered inflammatory cytokines in mouse cells, mouse models of MS, and human patient samples (Figures 3−5). These molecules IL-6, IL-17, IFN-γ, for examplerepresent some of the key biomolecular drivers of inflammation during MS and other autoimmune diseases. In particular, seminal work indicates IL-6 expands T H 17 cells associated with disease in mouse models of MS and also in humans; this cytokine also inhibits development of the T REGS that can help control disease. 39 Similarly, IL-17 is a characteristic cytokine of the T H 17 subset, and IFN-γ is a broadly acting pro-inflammatory effector cytokine produced by T cells during infection and autoimmunity. Intriguingly, while these reductions were occurring, the level of myelin-specific T cell expansion (e.g., proliferation, metabolic function) was unchanged (Figures 3 and 5). Further, we observed polarization toward T REGS when measured in both coculture and mice (Figures 4 and 5). Data from our studies with MS patient PBMCs exhibited similar changes in cytokine profiles, though further studies will need to include T REGS measurements. Taken together, these current findings indicate MOG-R 3 /GpG iPEMs activate self-antigen-specific cells, but bias the phenotype and function of these populations toward tolerance.
Our experiments with PBMCs from MS patient samples also suggest the possibility of polarizing the function of cells from patients with a variety of disease stages and severities, though samples from larger patient cohorts will be needed to confirm this possibility. The idea of disease heterogeneity is a consideration of general importance because no cures exist for MS and existing treatments rely on nonspecific suppression, either of broad immune function or through targeting all instances of a specific cytokine or molecule. 43 Further, patient responses to treatments plans are variable. Lastly, while treatments for earlier stages of MS have progressed and provided important benefits to patients, very few options are available for patients in progressive and later stages of disease. Thus, treatments that provide therapeutic effects even during later stages of disease would greatly improve both patient outcome and quality of life.
Another important question in the autoimmune field is what types of components are required for tolerogenic therapies. In the recent biomaterials literature, for example, several reports demonstrate self-antigens must be codelivered with regulatory signals to reprogram the responses against these antigens. 10,11 However, other studies demonstrate changing the physical form of antigenfree peptide versus peptide displayed on a cell or particlealters the trafficking of these self-antigens, leading to activation of debris clearance pathways involved in tolerance. 18,20 And, finally, several studies demonstrate that myelin is already presented in the lymph nodes of mice and humans during MS, 45,46 motivating the possibility that even delivery of GpG or other TLR modulators alone might support antigenspecific tolerance. These possibilities are exciting because excess TLR signaling is associated with a number of autoimmune disease and numerous target TLR ligands are arising. 1−8 In our in vitro studies we observed each iPEM component maintained selectivity, but GpG was less potent relative to free GpG at the assessed time. We also observed that iPEMs provide more efficient codelivery of cargo compared with admixed formulations, thus one possibility is that the strong electrostatic interactions between iPEM components increase the time needed to process these signals in immune cells. Intracellular trafficking and processing studies could help elucidate this hypothesis, but many of the unique features of iPEMs are most relevant for the in vivo setting. As noted, for example, the ability to codeliver and control the relative doses of self-peptides and regulatory signals is important in developing new, more-specific therapies for autoimmunity. The modular nature of iPEMs could enable these design features through assembly of strategically selected combinations of disease-relevant or irrelevant antigens and inert or tolerizing signals. Future mechanistic studies will explore how the role and relative concentrations of GpG, myelin peptide, and other iPEM components (e.g., irrelevant peptide antigens, CTRL) impact inflammation and T cell function using knockout models and trafficking of transgenic reporter cells. In our platform, as well as in development of other nanomaterial platforms, linking efficacy with treatment-associated changes in pathology is also critical, for example, by demonstrating treatment reduces infiltration of myelin-reactive T cells to the brain and promotes remyelination during reversal of disease. In some of our other recent work, we showed iPEMs built from model immune signals promote codelivery of immunological cargos to draining lymph nodes. 32 In the current setting of tolerance, understanding how iPEMs built from regulatory signals impact the local microenvironment of lymph nodes is another fascinating question, one that will help reveal how iPEMs polarize T cell differentiation in these tissues and how specific the resulting tolerance is.

CONCLUSIONS
PEMs offer features that have sparked exploration in fields from energy to medicine. Here we use three experimental systems mouse cells, mouse models of MS, and human MS patient samplesto demonstrate that PEMs can also be exploited to promote immunological tolerance. The potency of the results in the restraint of EAE onset and progression suggests future studies to determine whether late stage treatment with iPEMs can reverse established disease and drive remyelination in the CNS. Similarly, studies with statistically relevant sizes of human patient sample sets will help reveal the utility and mechanisms underpinning the function of iPEMs. These future questions are catalyzed by the current work, which demonstrates the potential of PEMs to combat autoimmunity. More generally, our results demonstrate that regulating TLR signaling can be used to promote tolerance, an idea that can be extended to other biomaterials to study or treat autoimmune diseases and conditions such as allergies and asthma.

ACS Nano
Article Automation). Substrates were cleaned via sequential rinsing in acetone, ethanol, methanol, and water, dried under air, and charged with an oxygen plasma system (Jupiter III, March). Prepared substrates were then coated with base layers of strong polyelectrolytes, poly(ethylenimine) (PEI, Polysciences, Inc.) and poly(sodium 4styrenesulfonate) (SPS, Sigma-Aldrich), as previously reported. 32 Briefly, chips were incubated in 20 mM PEI with 50 mM NaCl and 5 mM HCl for 5 min, washed twice in water for 1 min, incubated in 20 mM SPS with 50 mM NaCl for 5 min, and washed two more times with water. This process was repeated for a total of 10 deposition cycles using a DR3 dipping robot (Riegler & Kirstein GmbH), and chips were dried under air and stored at room temperature until subsequent coating with iPEMs.
iPEM Assembly and Characterization on Planar Substrates. Cargo solutions were prepared by dissolving MOG-R 3 and GpG at 0.5 mg/mL in 1 × PBS. iPEMs were assembled by dipping baselayercoated substrates in MOG-R 3 for 5 min, washing twice in PBS for 1 min, incubating in GpG for 5 min, and washing two additional times in fresh aliquots of PBS. This process was repeated to deposit the desired number of MOG-R 3 /GpG bilayers. For experiments designed to monitor film thickness, iPEMs were deposited on silicon chips and, every two bilayers, dried under air and measured using a Stokes Ellipsometer (Gaertner Scientific). At each measurement step, the thickness at five locations on at least three separate substrates was recorded and averaged. In studies designed to quantify relative cargo loading, iPEMs were assembled on quartz chips, and UV−vis spectrophotometry was used to record the absorbance values from 250 to 600 nm every two bilayers, as above. Measurements were recorded at three locations on at least three separate substrates and averaged. Wavelengths of 260 and 500 nm were used to indicate the loading of GpG and FITC tagged MOG-R 3 , respectively. Following deposition of eight bilayers of FITC-labeled MOG-R 3 and Cy5-labeled GpG on a quartz substrate, fluorescence microscopy (Olympus IX-83) was used to confirm colocalization of both cargos in an iPEM assembly. In these experiments, a portion of the film was removed with a needle scratch to provide contrast. Atomic force microscopy images were obtained in tapping mode in air using a Digital Instruments (Veeco) multimode atomic force microscope with a Nanoscope III controller and 10 μm scanner. Images were analyzed and root-meansquare roughness was calculated using NanoScope Analysis software (v1.50).
iPEM Assembly and Characterization on Colloidal Substrates. Calcium carbonate microparticle (MP) templates were precipitated, as previously described. 32 Briefly, sodium carbonate (Alfa Aesar) and calcium chloride dihydrate (Sigma-Aldrich) were dissolved at 0.33 M in water. Equal volumes of these solutions were combined, under stirring (800 rpm), and allowed to mix for 5 min. For each batch of iPEMs, 500 μL of the MP solution was transferred to a 1.5 mL microcentrifuge tube, and aliquots were centrifuged (20 s, 1000g) to collect MPs and washed twice in water. To deposit iPEMS, MPs were then resuspended in 600 μL of MOG-R 3 cargo solution (0.5 mg/mL in water), incubated for 3 min, washed twice in water, incubated in 600 μL of GpG cargo solution (1 mg/mL in water) for 3 min, and washed two more times to complete one bilayer. As noted, between each coating and washing step, particles were collected with a short centrifugation (20 s, 1000g). In experiments to monitor iPEM growth, aliquots of MPs were collected after deposition of 1, 2, and 3 bilayers and imaged by fluorescence microscopy at a fixed exposure time. Pixel intensity was determined along line traces through the diameter of representative images using ImageJ. In separate studies, aliquots of iPEM-coated MPs were collected after deposition of each cargo layer, and surface charge was measured using a Malevern Zeta Sizer Nano ZS90. To measure iPEM composition, the concentrations of cargo solutions before and after coating, as well as the concentrations of cargos in wash solutions, were quantified using spectrophotometry, comparing absorbance values to standard curves of MOG-R 3 and GpG. These values were used to calculate the mass of cargos deposited on MPs via indirect loading measurement. To vary the input ratio of cargos, the mass of GpG was fixed at 600 μg/batch, as above, and the input mass of MOG-R 3 was titrated from 1200 μg to 18.75 μg, at the indicated ratios. iPEM composition for each ratio tested is reported in Table S1. To remove the calcium carbonate core and form iPEM capsules, calcium carbonate MPs were coated with three MOG-R 3 /GpG bilayers, as above, and then incubated in 1 mL of 0.1 M ethylenediaminetetraacetic acid (EDTA, Sigma-Aldrich), adjusted to pH 6 with 1 M HCl and 1 M NaOH solutions, for 30 min. Capsules were then centrifuged (3 min, 1000g), washed with 600 μL of 1X PBS, and resusupended in 1 × PBS for imaging and cell or animal studies. In some studies, iPEM capsules were formed from MOG-R 3 and CTRL, with or without a Cy3 label, following the same protocols above, for materials characterization and in vitro experiments.
Transgenic T Cell Proliferation. To characterize MOG-specific T cell interactions, DCs from naïve C57BL/6J mice were isolated, as above, and treated with CpG (1 μg) and either soluble MOG-R 3 (60 μg), GpG (10 μg), or CTRL (10 μg), or iPEM formulations (10 μg). After 16 h of culture, CD4 + T cells were isolated from the spleens of transgenic 2D2 mice (C57BL/6-Tg(Tcra2D2,Tcrb2D2)1Kuch/J, The Jackson Laboratory) with a magnetic isolation kit (StemCell Technologies), according to the manufacturer's instructions. Isolated CD4 + T cells were then incubated with a cell proliferation dye (eFluor 670, eBioscience) and washed, and 2.5 × 10 5 labeled cells were added to DC cultures. After 72 h of coculture, cells were collected, washed, and blocked, as above. Cells were then stained with anti-CD4 (BD Biosciences) for 20 min at room temperature, washed to remove unbound antibody, and resuspended in DAPI for analysis by flow cytometry. In separate studies, CD4 + T cells with receptors specific for OVA 323−339 were isolated from the spleens of transgenic OT-II mice (B6.Cg-Tg(TcraTcrb)425Cbn/J, The Jackson Laboratory), labeled with proliferation dye, and added to DC cultures, as above. In these experiments, control wells were treated with CpG (1 μg) and soluble OVA 323−339 (60 μg).

ACS Nano
Article ELISA. Supernatants from 2D2 cocultures were analyzed by ELISA according to manufacturer's instructions for the secretion of IFN-γ and IL-6 (BD Biosciences).
Transgenic T Cell Phenotype. To analyze the expression of markers for T REG phenotype, cocultures were prepared, as above, without fluorescent labeling of CD4 + 2D2 T cells. After 72 h of coculture, cells were collected, washed twice, and blocked as above. Cells were then incubated in antibodies against CD4 (BD Biosciences) and CD25 (BD Biosciences) for 45 min at 4°C, protected from light. Following staining for surface markers, cells were washed and then fixed and permeabilized using a Foxp3 Transcription Factor Staining Buffer Set, according to the manufacturer's instructions (eBioscience). Cells were then stained for the expression of Foxp3 overnight at 4°C, washed, and analyzed by flow cytometry.
Capsule Immunization. Mice were immunized with 400 μg of (MOG-R 3 /GpG) 3 iPEMs, administered as bilateral injections (2 × 25 μL injections, containing 200 μg of iPEMs each) subcutaneously at the tail base. Capsules were administered to either naïve mice for immunohistochemical analysis of iPEM drainage to lymph nodes or to mice induced with EAE on days 5 and 10 post induction.
Immunohistochemical Analysis. Two days after immunization, draining inguinal lymph nodes were excised from (FITC-MOG-R 3 / GpG) 3 iPEM-treated mice. Tissues were immersed in optimal cutting temperature medium (Tissue-Tek) and frozen. Blocks were sectioned at 6 μm thickness using a Microm HM 550 cryotstat. Sections were fixed in ice-cold acetone, dried, and washed in PBS. Sections were then blocked using appropriate serum and stained with primary antibodies for CD3e and B220 for 1 h at room temperature. After two 5 min washes, fluorescently labeled secondary antibodies were added for 45 min at room temperature. Sections were then washed, fixed with 4% paraformaldehyde, quenched with 1% glycerol, and mounted using ProLong Diamond Antifade Mountant (Thermo Fisher Scientific). Images were collected on an Olympus IX-83 fluoroescent microscope.
EAE Induction and Monitoring. EAE was induced in C57BL/6J mice with kits according to the manufacturer's instructions (Hooke Laboratories). Mice were monitored for body weight fluctuation and paralysis, which was assigned a clinical score (0, no symptoms; 1, limp tail; 2, hind limb weakness; 3, hind limb paralysis; 4, full hind limb and partial front limb paralysis; and 5, moribund). All animal care and experiments were carried out using protocols approved and overseen by the University of Maryland IACUC committee in compliance with local, state, and federal guidelines.
Tissue Collection and Processing. Three days after dosing with iPEMs (i.e., day 13 post induction), indicated tissues were collected and mechanically dissociated through 40 μm cell strainers. Spleen samples were resuspended in Ack lysing buffer (Invitrogen) to deplete red blood cells and then washed with PBS. Single cell suspensions were split to perform three analyses. First, cell counts were recorded by flow cytometry using counting beads according to the manufacturer's instructions (BD Biosciences). Second, using these cell counts, a uniform number of cells (5 × 10 5 ) from each tissue was plated in duplicate, with one well receiving a pulse of MOG peptide (25 μg/mL) and the other an equivalent dose of OVA 323−339 peptide. Restimulated cells were cultured for 72 h, and supernatants were analyzed for IFN-γ (BD Biosciences), IL-6 (BD Biosciences), and IL-17 (R & D Systems) secretion by ELISA, as described above. Third, the remaining aliquot of the single cell suspension was analyzed immediately for the expression of phenotypic markers of T REGS  CD4, CD25, and Foxp3as described above.
Human PBMC Samples. Human MS patient peripheral blood mononuclear cell samples were collected in conjunction with the IRBapproved VALOMS study protocol with informed signed consent. VALOMS is an observational study that has been initiated by the VA MS Center of Excellence-East to examine factors associated with disease progression among U.S. military veterans with MS. Frozen PBMC samples selected randomly from the patient sample repository were thawed and washed, and a Ficoll gradient was used to remove dead cells. Cells were then counted and plated in 96-well plates at 1.5 × 10 5 , 2.0 × 10 5 , or 5 × 10 4 cells/well for Patients 1, 2, and 3, respectively. Cells were left untreated or incubated with 30 μg of either (MOG-R 3 /GpG) 3 iPEMs or (MOG-R 3 /CTRL) 3 iPEMs. After 72 h of incubation, 20 μL of 5 mg/mL 3-(4,5-dimethlythiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) was added to each well, and cultures were subsequently incubated for 4 h at 37°C. Plates were then centrifuged at 2000 rpm for 10 min. Supernatants were collected for analysis of cytokine secretion, and then 150 μL dimethyl sulfoxide was added to each well. The absorbance at 570 nm, with a reference of 630 nm, was recorded. Cell culture supernatants were analyzed by Luminex Multianalyte System for secreted human TNF-α, IL-6, IL-10, and IFN-γ according to the manufacturer's instructions.