Polyplexes assembled from self-peptides and regulatory nucleic acids blunt toll-like receptor signaling to combat autoimmunity

Abstract

Autoimmune diseases occur when the immune system incorrectly recognize self-molecules as foreign; in the case of multiple sclerosis (MS), myelin is attacked. Intriguingly, new studies reveal toll-like receptors (TLR), pathways usually involved in generating immune response again pathogens, play a significant role in driving autoimmune disease in both humans and animal models. We reasoned polyplexes formed from myelin self-antigen and regulatory TLR antagonists might limit TLR signaling during differentiation of myelin-specific T cells, inducing tolerance by biasing T cells away from inflammatory phenotypes. Complexes were formed by modifying myelin peptide with cationic amino acids to create peptides able to condense the anionic nucleic-acid based TLR antagonist. These immunological polyplexes eliminate synthetic polymers commonly used to condense polyplexes and do not rely on gene expression; however, the complexes mimic key features of traditional polyplexes such as tunable loading and co-delivery. Using these materials and classic polyplex analysis techniques, we demonstrate condensation of both immune signals, protection from enzymatic degradation, and tunable physicochemical properties. We show polyplexes reduce TLR-signaling, and in primary DC and T cell co-culture, reduce myelin-driven inflammation. During mouse models of MS, these tolerogenic polyplexes improve the progression, severity, and incidence of disease.

Graphical Abstract

1. Introduction

Failure to mount an immune response to an antigen is known as immunological tolerance.[1] When tolerance to self-antigens (e.g. host proteins) is not maintained, inflammation and autoimmune disease can develop. Autoimmune diseases such a multiple sclerosis (MS), type 1 diabetes, lupus, and rheumatoid arthritis, affect over 20 million Americans.[2, 3] In MS, the myelin sheath that insulates and protects the axons of neurons is recognized as a foreign antigen.[4, 5] Myelin derived antigens are now strongly implicated as the targets of malfunctioning self-reactive cells in MS,[6] inflammatory populations that infiltrate the central nervous system (CNS). In the CNS, the attack by these cells drives demyelination of neurons, while secretion of inflammatory cytokines recruits additional immune cells to the site.[7]

Cures for autoimmune diseases do not exist and treatment options are limited. Typical therapies rely on regular doses of immunosuppressive drugs or antibodies that are non-specific and can leave patients immunocompromised.[8] Thus, great interest has developed in new treatments that can durably block inflammatory responses against self-antigen without non-specific suppression. One exciting new strategy is co-delivery of an antigen, such as myelin, along with immunomodulatory molecules to redirect responses against self-antigen. Biomaterials offer very attractive features in this context, and, in addition to co-delivery, provide routes for efficiently targeting antigen presenting cells (APCs) and for controlled release.[9, 10] In particular, synthetic polymers have been used to co-deliver antigen with regulatory signals—small molecules, cytokines, and proteins—to polarize differentiating T cells away from inflammatory cells and toward regulatory T cells (T_REG).[11–19] These populations can control self-reactive effector cells (e.g., T_H1, T_H17) that drive autoimmune disease while limiting broad suppression.[1] Expansion and biasing of T cells toward regulatory response also reduces the absolute number of inflammatory cells and offers the potential for more durable treatments.[20, 21]

Toll-like receptors (TLRs) are a collection of signaling pathways that recognize pathogen-associated molecular patterns (PAMPs), resulting in secretion of inflammatory cytokines and activation of the immune cells needed to fight infection.[22] Interestingly, a developing body of new literature demonstrates that many TLRs are overexpressed in MS and other autoimmune diseases, as well as in animal models.[23–27] For example, in experimental autoimmune encephalomyelitis (EAE), a pre-clinical mouse model of MS, disease is significantly reduced in TLR9 knock-out mice.[23] This example highlights the importance of TLR9 signaling in driving disease. TLR9 typically activates innate immunity following recognition of a characteristic bacterial DNA sequence called CpG.[28] CpG DNA is a TLR agonist (TLRa) for TLR9, ultimately driving inflammation through the MyD88 pathway that leads to activation of dendritic cells (DCs), macrophages, monocytes, and B cells, along with secretion of inflammatory cytokines.[29, 30] For this reason, CpG has been intensively studies as a vaccine adjuvant.[29] In contrast to this common role of TLR9 as an adjuvant, one group has explored TLR9 antagonists to promote tolerance using GpG, an analog of CpG exhibiting a substitution of guanine for cytosine.[31, 32] Like CpG, GpG is unmethylated, single-stranded DNA with a phosphorothioate backbone that can bind TLR9. During cell studies, treatment with GpG suppressed proliferation of inflammatory T_H1 cells. In mice, repeated regular treatment with GpG attenuated EAE—a T_H1-mediated MS model— and further, enhanced induction of tolerance when GpG was mixed just prior to injection with plasmid DNA encoding myelin self-antigens. Thus, we reasoned juxtaposing GpG and self-antigen in polyplex-like nanoparticles might alter the inflammatory signaling associated with recognition of self-antigen, biasing differentiating T cells away from inflammatory phenotypes to help combat autoimmune disease.

One simple class of biomaterials particularly well-suited for the co-delivery strategy above is polyplexes, nano-structured complexes that spontaneously assemble due to electrostatic condensation when nucleic acid is mixed with a cationic polymer.[33–35] Polyplexes condense cargo to a high density that is easily internalized by cells and offer protection from enzymatic degradation. These and other advantages have been exploited in a variety of applications ranging from transfection to vaccination. For the former, great effort has been invested to develop cationic polymers that are non-toxic and offer features (e.g., proton-sponge capacity) that overcome specific barriers to DNA and RNA delivery such as endosomal escape. In the vaccine area—both prophylactic and therapeutic—the particulate nature of polyplexes promotes uptake by APCs, and further, polyplexes allow simple tuning of vaccine specificity by simply replacing a particular DNA plasmid or RNA molecule with different sequences. Despite these useful properties, only a handful of reports, all in the past 3–4 years, have explored polyplexes to modulate immune function or promote immunological tolerance. In one example, receptors involved in the development of diabetes were targeted in mice by condensing plasmid encoding a soluble ligand for this receptor.[36] In a second example, mice were treated during diabetes with polyplexes formed using chitosan to condense plasmids encoding interleukin 4 (IL-4) and interleukin 10 (IL-10), regulatory cytokines that suppress inflammatory cytokines.[37] The Mellor group has studied a system using poly(ethylenimine) (PEI) to condense plasmid DNA that is free of PAMPs, limiting activation of common inflammatory pathways and cytokines that are also often active during autoimmune diseases.[38–40] Although none of these approaches incorporate self-antigen, we hypothesized polyplexes including such components might allow induction of antigen-specific tolerance. Further, a number of reports demonstrate that commonly used biomaterials, such as chitosan, PLGA, and polystyrene, exhibit intrinsic immunogenicity that drives inflammation.[10, 41–45] This characteristic could be detrimental when developing therapies for autoimmune disease, as intrinsic inflammation from a delivery vehicle could exacerbate disease. Thus, therapeutics that eliminate carrier components but still offers features of biomaterials such as co-delivery and cargo protection, could offer simpler and lower-risk treatments.

To realize the possibilities discussed above, we designed polyplex-like structures assembled from GpG and a myelin peptide (myelin oligodendrocyte glycoprotein, MOG) modified with cationic arginine residues. This strategy is unique in four ways. First, these nanoparticles consist entirely of immune signals, and are free of all carriers or supports. The polyplexes thus mimic attractive features of traditional biomaterials, while eliminating the intrinsic immunogenicity described above. This is highly relevant for autoimmune and inflammatory conditions where such traits might worsen disease. Second, we are targeting suppression of TLR signaling to promote tolerance, a new idea arising in immunology research, and that is untapped in the biomaterials field. This is particularly important for autoimmune therapies owing to the intriguing recent studies revealing TLR signaling is overactive in human autoimmune diseases such as multiple sclerosis, diabetes, lupus, and rheumatoid arthritis. Third, because the polyplexes are built entirely from immune signals, they juxtapose both the self-antigen and the regulatory signal at very high densities. This feature provides highly efficient loading (i.e., 100%) relative to polymers or matrices used to load and deliver cargo, as well as the co-delivery needed to polarize T cells away from inflammatory phenotypes. Last, these immune polyplexes are novel because they include self-antigen for specificity, but do not rely on expression of any plasmid component; the nucleic acid cargo is an antagonistic TLR ligand that directly interacts with these receptors. Using this platform, below we show that MOG-GpG polyplexes are readily internalized by APCs, decrease TLR9 signaling, and are non-toxic. Treatment of DCs reduces activation and inflammatory cytokines, biasing myelin-specific T cell function away from inflammatory phenotypes during co-culture studies. In a mouse model of MS (EAE), polyplex treatment significantly reduces both the severity and incidence of disease.

2. Materials and Methods

2.1 Materials

GpG DNA (5’-TGA CTG TGA AGG TTA GAG ATG A-3’), CpG DNA (5’-TCC ATG ACG TTC CTG ACG TT-3’) and a control oligonucleotide (5’-TCC TGA GCT TGA AGT-3’) were purchased from IDT (Coralville, IA). MOG_35–55 (MEVGWYRSPFSRVVHLYRNGK) was synthesized by Genscript (Piscataway, NJ) with a FITC tag on the N-terminus and either one or two arginine (R) residues on the C-terminus (MOGR₁, MOGR₂). TE buffer and ethidium bromide were purchased from Amresco (Solon, OH). DNase I kits with 10X reaction buffer were purchased from New England Biolabs (Ipswich, MA). RPMI-1640 media and Molecular Biology Grade Water were purchased from Lonza (Allendale, NJ) and fetal bovine serum (FBS) was supplied by Corning (Tewksbury, MA). β-mercaptoethanol, ethylenediaminetetraacetic acid (EDTA) and bovine serum albumin (BSA) were purchased from Sigma Aldrich (St. Louis, MO). 20X PBS, HEPES, and non-essential amino acids were purchased from VWR (Radnor, PA). L-glutamine, penicillin-streptomycin, and DAPI were purchased from Thermo Fisher Scientific (Grand Island, NY). Spleen Dissociation Medium and CD4 negative selection kits were from STEMCELL Technologies (Vancouver, BC). CD11c microbeads were purchased from Miltenyi Biotec (Cambridge, MA). Cy5 nucleic acid labelling kits were from Mirus (Madison, WI). HEK-Blue™ TLR9 report cells, detection media, and antibiotics were supplied by Invivogen (San Diego, CA). Fluorescent antibody conjugates and enzyme-linked immunosorbent assay (ELISA) reagents were purchased from BD (San Jose, CA). Cell Proliferation Dye eFluor® 670 was from affymetrix eBioscience (San Diego, CA). Hooke Kits™ for EAE Induction were supplied by Hooke Laboratories (Lawrence, MA).

2.2 Cells and animals

All primary cells were harvested from female C57BL/6 mice (4–12 weeks, stock #000664) and male C57BL/6-Tg(Tcra2D2, Tcrb2D2)1Kuch/J (2D2) mice (10–16 weeks, stock #006912) purchased from Jackson Laboratories (Bar Harbor, ME). 2D2 mice have transgenic CD4⁺ T cell receptors specific for MOG. Female C57BL/6 mice (10–12 weeks, stock #000664) were used for EAE studies. All animals were cared for in compliance with federal, state, and local guidelines, and using protocols reviewed and approved by the University of Maryland's Institutional Animal Care and Use Committee (IACUC).

2.3 Complex formation

Complexes were formed by mixing aqueous solutions of GpG and MOG with MOGR_x (x=1 or 2). The components were mixed at defined mass ratios ranging from 1:20 to 40:1 MOGR_x:GpG. In these studies, the total mass of GpG was fixed at 25 µg while the mass of MOGR_x was varied to control the charge ratio. Prior to use in cell and animal studies, complexes were centrifuged for 90 minutes at 5,000 g in a Microfuge 22R (Beckman Coulter, Brea, CA). Supernatants were removed and the pellets were resuspended in water to remove any excess MOGR_x or GpG that was not complexed. Loading studies to determine the amount of complexed MOGR_x and GpG were carried out by measuring the fluorescence levels of FITC-MOGR_x and Cy5-GpG in the supernatant of the centrifuged complexes. Cy5 labeling was carried out using the manufacturer’s protocols.

2.4 Characterization of complex size and charge

Hydrodynamic diameter and zeta potential of complexes were measured in triplicate using samples prepared in molecular biology grade water and analyzed on a Zetasizer Nano ZS90 (Malvern Instruments Ltd, Westborough, MA). Stability analysis was completed by incubating MOGR_x-GpG in media with serum and measuring the hydrodynamic diameter at the indicated intervals.

2.5 Ethidium bromide exclusion assay

Ethidium bromide (EtBr) was added to GpG at a 1:5 mass ratio and allowed to equilibrate for one hour. MOGR₁ and MOGR₂ were then added at GpG:MOGR_x mass ratios ranging from 1:5 to 10:1. Fluorescence was measured using an excitation wavelength of 540 nm and an emission wavelength of 570 nm. The fluorescence of EtBr alone was subtracted from all samples and an intensity ratio was calculated by comparing of the fluorescence of the complex with EtBr to the fluorescence of DNA and EtBr alone.

2.6 Enzymatic degradation assays

Complexes were made from Cy5-GpG and MOGR_x and centrifuged as described above, then resuspended in 100 µL 1X DNA I reaction buffer. The fluorescence from the Cy5-GpG was then measured using an excitation wavelength of 640 nm and an emission wavelength of 670 nm. 2 units of DNase I were added to each sample after complex formation, then mixed and incubated for 30 minutes at 37°C. After incubation, 1 µL of 0.5M EDTA was added and the enzyme was heat inactivated at 75°C for 10 minutes per the manufacturer’s instructions. The fluorescence of the Cy5-GpG was then measured again to determine the extent of degradation compared to free Cy-5 GpG. Fluorescent measurement was chosen to quantify GpG amount over spectrophotometry because the absorbance of MOGRx (peak at 280nm) overlaps the absorbance of GpG (peak at 260 nm).[46, 47]

2.7 TLR9 reporter cell study

HEK-Blue™ mTLR9 cells are co-transfected with the murine TLR9 gene and an inducible secreted embryonic alkaline phosphatase (SEAP) reporter gene. Stimulation of the cells with an unmethylated CpG-ODN sequence (TLR-9 agonist) activates NF-κB and AP-1, which leads to the production of SEAP to allow colorimetric detection. Cells were first treated with CpG and then either complexes or free controls. After a 16 hour incubation period, the level of SEAP was measured by spectrophotometry at 650 nm.

2.8 Dendritic cell isolation and flow cytometry

CD11c⁺ DCs were isolated from the spleens of female C57BL/6 mice through positive selection with spleen dissociation media and CD11c Microbeads. DCs were plated in a 96 well plate at 100,000 cells per well and cultured in RPMI 1640 media supplemented with 10% FBS, 2 mM L-glutamine, 1X non-essential amino acids, 10 mM HEPES buffer, 1X penicillin and streptomycin, and 55 µM β-mercaptoethanol at 37°C and 5% CO₂. DCs were activated by adding CpG (TLR9 agonist) at a concentration of 5 µg/mL and treated with the each complex formulation, vehicle (water) only, or free controls of MOGR_x alone or GpG alone. For activation studies, cells were stained after 24 hours for viability (DAPI⁻) and for classic surface activation markers (CD80⁺, CD86⁺, CD40⁺) using fluorescent antibody conjugates. For uptake studies, complexes were made with FITC tagged MOGR₁or MOGR₂ and Cy5 labelled GpG. Cells were incubated with complexes for two hours prior to collection and staining for viability (DAPI⁻). For both sets of studies, cells were measured by flow cytometry (BD Cantoll, San Jose, CA) and data were analyzed with FlowJo v.10 (TreeStar, Ashland, OR).

2.9 Enzyme-linked immunosorbent assay

Supernatants from DCs treated for activation studies were collected. Cytokine secretion levels were analyzed via ELISA using mouse interferon gamma (IFN-γ) reagents. Briefly, 96-well plates were coated with an IFN-γ capture antibody and after an overnight incubation, the supernatant samples were added. An IFN-γ detection antibody and streptavidin-horseradish peroxidase conjugate mixture was then added to the wells for 1 hour. A tetramethylbenzidine and hydrogen peroxide mixture was added to each well for 30 minutes and the reaction was stopped by the addition of 1M phosphoric acid before reading the absorbance of each sample at 450 nm. IFN-γ concentrations were calculated from absorbance by comparing to a standard curve.

2.10 DC/T cell co-culture and flow cytometry

Isolated DCs were treated with CpG and the same complex formulations or controls described above for DC activation studies, then cultured for 24 hours. CD4⁺ T cells were next isolated from the spleens of 2D2 mice via a negative selection kit. For phenotypic studies, 300,000 T cells were added to the wells containing the treated DCs. After 48 hours, cells were stained for a T cell surface marker (CD4⁺) and transcription factors (Tbet⁺, RORγ⁺) indicative of T_H1 and T_H17 cells, respectively. For T cell proliferation studies, cells were labelled with eFlour 670 proliferation dye prior to incubation with DCs. After 48 hours, cells were collected and stained for viability (DAPI⁻) and for a T cell surface marker (CD4⁺). Cells were examined for transcription factor expression or proliferation dye dilution with flow cytometry and data were analyzed with FlowJo v.10.

2.11 EAE studies in mice

10 week old, C57BL/6 female mice were induced with EAE using a standard induction protocol provide by the reagent vendor (Hooke Laboratories). Briefly, mice were injected subcutaneously (s.c.) with 50 µL of an emulsion of MOG_35–55 and complete Freund’s adjuvant (CFA) at the upper and lower back on day 0. Injection of MOG/CFA emulsion primes inflammatory T cells specific for MOG. 100 µL of pertussis toxin (PTX) was also administered intraperitoneally (i.p.) 2 and 24 hours after MOG/CFA injection. PTX injection opens the blood brain barrier, allowing infiltration of inflammatory MOG-specific T cells. These cells demyelinate the axons of neurons leading to loss of motor function beginning around day 9 and complete paralysis occurring by day 16. Disease severity was monitored daily by measuring body weight loss and clinical score, a measurement of the extent of paralysis: 0 – no symptoms, 1 – tail paralysis, 2 – hind limb weakness, 3 – hind limb paralysis, 4 – partial front limb paralysis, 5 – moribund. The induced mice (N=8–10 per group) were either left untreated or treated s.c. at the tailbase using complexes prepared in molecular biology grade water at a 2:1 MOGR₂:GpG ratio (actual dose: 200 µg MOGR₂; 85.9 µg of GpG). As indicated in the text for each study, treatment regimens consisted of either a single complex injection on day 7, or three administrations of complexes on days 6, 12, and 18. In control studies, mice were treated with 50 µg free GpG on days 5, 10, and 15.

2.12 Statistical analysis

One-way ANOVA with a Tukey post-test was used to compare three or more groups for materials characterization and in vitro studies, with post-test corrections for multiple comparisons. Unpaired t-tests were used to compare mean clinical score and body weight between groups at each study day. Log-rank tests were used in analysis of disease incidence. For all tests, p values ≤ 0.05 were considered significant. For all figures: *p<0.05, **p<0.01, ***p<0.001, #p<0.0001, ns = not significant.

3. Results

3.1 MOG modified with cationic arginine residues binds GpG to form immunological polyplexes

Since GpG is a single stranded (ss) DNA molecule, and thus intrinsically anionic, we first tested if immunological polyplex-like structures could be formed using GpG and MOG modified with either one or two cationic arginine residues (MOGR₁, MOGR₂). In these studies, the mass of GpG was fixed while the mass of MOGR_x was varied to form complexes over the range of 1:20 – 40:1 MOGR_x:GpG. These formulations corresponded to a range of charge ratios spanning highly negative to highly positive values (Table 1). Dynamic light scattering confirmed the formation of complexes exhibiting nanoscale hydrodynamic diameters of 117.9 ± 6.5 nm to 199.2 ± 4.1 nm (Fig. 1A). These sizes were relatively uniform across the ratios tested, though for complexes formed from MOGR₂ at near-neutral charge ratios, the sizes increased slightly (e.g., 1:1, 2:1). Polyplex stability studies conducted by incubation in media with serum revealed that these sizes did not change appreciably over at least 24 hours (Fig. S1). Surface charge, however, was readily tunable as indicated by zeta potential measurements (Fig. 1B). As expected, complexes formed at lower MOGR_x:GpG ratios (e.g., 1:20) exhibited a negative zeta potential that became positive moving toward higher MOGR_x: GpG ratios (e.g., 40:1). The values of these measurements ranged from −42.5 ± 0.5 mV (1:5) to 33.4 ± 0.7 mV (40:1), with a shift in zeta potential observed approximately around the zone that charge ratio analysis predicted a charge inversion (Table 1). Corresponding to these changing physicochemical properties, measurements of the actual loading of each complex formulation ranged from 0.57 µg to 9.18 µg of MOGR_x and 2.18 µg to 4.88 µg of GpG as a function of ratio (Table S1).

Table 1. Charge characteristics of MOGR_x-GpG Polyplexes.

MOGRx:GpG Mass Ratio	MOGR1:GpG Charge Ratio	MOGR1-GpG Complex Charge	MOGR2:GpG Charge Ratio	MOGR2-GpG Complex Charge
1:20	0.025	−40	0.029	−35
1:10	0.051	−20	0.058	−17
1:5	0.10	−10	0.12	−8.5
1:2	0.25	−4	0.29	−3
1:1	0.51	−1.5	0.58	−1
2:1	1.0	0	1.2	+0.3
5:1	2.5	+2	2.9	+2.5
10:1	5.1	+5	5.8	+6
20:1	10.1	+10	11.4	+11
40:1	20.2	+20	22.9	+23

Figure 1.

MOGR_x and GpG formed complexes with controllable properties. (A) DLS measurements performed in triplicate showed relatively small diameters regardless of complex formulation. (B) Triplicate zeta potential measurements of complexes indicated controllable surface charge. (C) Binding of MOGR_x to GpG was measured by an EtBr assay. A reduction in fluorescent intensity relative to free GpG indicated displacement of EtBr by the peptide. (D) Protection of GpG from degradation by MOGR_x complexation was measured after incubation with DNase. *p<.05, **p<.01, ***p<.001, #p<.0001, ns = not significant. For Panels C and D, statistics are comparisons versus free GpG.

3.2 Strong binding between MOGR_x and GpG provides protection from enzymatic degradation

To determine the binding affinity between MOG and GpG, an ethidium bromide (EtBr) assay was performed using a subset of ratios (i.e., 1:5–10:1). For double-stranded DNA, EtBr intercalates base pairs, resulting in a significant increase in fluorescence. For ssDNA, EtBr can bind secondary and tertiary structural features where regions of local base pairing offer the stacked base pairs that support intercalation by dye molecules. [48, 49] Free (i.e., free) GpG controls incubated with EtBr led to a high level of fluorescence (Fig. 1C). Treating these mixtures with MOGR₁ or MOGR₂ at all ratios revealed that the peptides competed with and displaced the EtBr bound to GpG, decreasing the fluorescent measurements (Fig. 1C). As the mass of MOGR_x added to GpG increased (1:5 → 10:1), fluorescent intensity decreased (Fig. 1C), indicating increasing binding affinity as the surface charge became increasingly positive over this same range (Fig. 1B).

We next tested if condensation of GpG by MOGR_x protects GpG from enzymatic degradation. GpG was covalently labelled with Cy5 and the fluorescence was measured to quantify the amount of DNA prior to degradation. After incubation of free GpG with DNase for 30 minutes, less than 50% remained intact relative to the starting concentration of GpG (Fig. 1D). Complexation of GpG with MOGR₂, which generally exhibited a slight increase in positive charge versus MOGR₁ (Fig. 1B), significantly reduced GpG degradation at most ratios, with up to ~80% of the GpG remaining after incubation with enzyme (Fig. 1D). Complexation with MOGR₁ resulted in a trend of protection, but these effects were only significant against free GpG at one ratio.

3.3 GpG complexed with MOGR_x maintains the ability to restrain TLR9 signaling

We next confirmed GpG is able to blunt TLR9 signaling after complexation with MOGR_x. Cells were first stimulated with CpG, a strong TLR9 agonist, then treated with polyplexes or dose-matched free components (i.e., GpG or MOGR_x). These experiments confirmed high levels of TLR9 activity in cells receiving only CpG, while addition of GpG-containing complexes significantly reduced TLR9 signaling (Fig. 2A). Although in some cases the levels of reduction observed with complexes were not as high as treatment with equivalent doses of free GpG (Fig. 2A, purple bars), these experiments confirm GpG maintains the capacity to block TLR9 signaling after complexation with self-antigen. Although a small decrease in TLR9 activity was seen due to some free MOGR_x treatments (grey bars), these non-specific effects were not observed for most of the ratios. Building on the TLR signaling results, we assessed the toxicity of complexes by treating primary DCs (CD11c⁺) isolated from the spleens of mice with complexes or free controls. Flow cytometry analysis of DAPI⁻ cells indicated only a small decrease in cell viability (~10%), regardless of the specific complex formulation (Fig. 2B). These effects were also observed in wells treated with only GpG (purple bars), but were less common in wells treated with free MOGR_x (grey bars) controls (Fig. 2B).

Figure 2.

Complexes antagonized TLR9 without affecting viability. (A) The SEAP level induced by CpG and complex or free treatments correlates to TLR9 activity in a reporter cell line and was measured by spectrophotometry. (B) Viability following complex or free treatments was measured by incubation with DAPI and flow cytometry. *p<.05, **p<.01, ***p<.001, #p<.0001, ns = not significant. For Panel B, statistics correspond to comparisons against CpG only.

3.5 Primary DCs readily internalize MOGR_x-GpG complexes

Having confirmed cell compatibility and the ability of complexes to reduce TLR9 signaling, we studied the immunological processing and tolerogenic activity of complexes. To determine the ability of APCs to phagocytose these nanoparticle, DCs were incubated with complexes formed from FITC-tagged MOGR_x and Cy5-labeled GpG or free controls. Flow cytometry analysis revealed all complex formulations resulted in a significant increase in both FITC signal (MOGR_x; Fig. 3A,B) and Cy5 signal (GpG; Fig. 3C,D) compared to vehicle treatment. Interestingly, for MOGR_x peptides, uptake levels were similar regardless of whether the peptide was in a free or complexed form (Fig. 3A,B). In contrast, GpG uptake was significantly higher when complexed with MOGR₂, compared to both free GpG and MOGR₁-GpG polyplexes (Fig. 3C,D).

Figure 3.

MOGR_x-GpG complexes were readily taken up by DCs. Uptake of complexes by primary DCs was analyzed by flow cytometry following incubation with polyplexes formed from FITC-MOGR_x (A), (B) and Cy5-GpG (C), (D). *p<.05, ***p<.001, #p<.0001, ns = not significant

3.6 MOGR_x-GpG complex treatment deactivates DCs

To determine how uptake of complexes impacts DC activation, DCs were stimulated with CpG prior to incubation with MOGR_x-GpG complexes or dose-matched free controls. Flow cytometry analysis revealed stimulation with CpG induced high levels of classic surface activation markers (Fig. 4A,B,C). Strikingly, complexes containing GpG significantly decreased activation compared to equivalent free MOGR_x controls (grey bars) by as much as 54.1%, 53.6%, and 53.1% for CD80 (Fig. 4A), CD86 (Fig. 4B), and CD40 (Fig. 4C), respectively. However, these decreases were most evident for CD80 and CD86, with MOGR₂-GpG complexes generally causing greater reductions in DC activation compared with MOGR₁-GpG complexes (Fig. 4A–C, green vs. blue series). Measurement of MFI yielded similar results (Fig. S2). Importantly, DC deactivation was not seen when GpG was replaced with a control oligonucleotide (ODN), indicating the specificity of the TLR antagonist (i.e., GpG) to regulate inflammation (Fig. S3). In similar studies, we tested if complexes alter the levels at which DCs secrete IFN-γ, a key inflammatory cytokine. As expected, ELISA measurements confirmed CpG-induced secretion of IFN-γ (Fig 4D). Analogous to the results of surface activation staining, MOGR_x-GpG complexes significantly reduced IFN-γ secretion relative to MOGR_x free controls (Fig. 4D).

Figure 4.

Complex treatment deactivated primary DCs. DC activation was measured following incubation with stimulatory CpG and complex or free treatments by expression levels of costimulatory markers CD80 (A), CD86 (B), and CD40 (C), by staining with fluorescent antibody conjugates and analysis by flow cytometry. (D) DC deactivation was confirmed by ELISA to measure the secretion of inflammatory cytokine IFN-γ. **p<.01, ***p<.001, #p<.0001, ns = not significant

3.7 MOGR_x-GpG complexes decrease inflammatory T cell proliferation

The results above indicate that complexes containing GpG alter both the activation state of DCs and the secretion of inflammatory cytokines by these cells. Since these characteristics are critical in determining the type and magnitude of response to foreign or self-antigens, we investigated whether these complex-induced changes altered the function of myelin-specific T cells characteristic of disease in MS. In these studies, we cultured similarly-treated primary DCs with T cells isolated from 2D2 mice. T cells from 2D2 mice exhibit transgenic T cell receptors, and thus proliferate upon encounter of APCs displaying MOG in the major histocompatibility complex II (MHC-II). After co-culturing for 48 hours, flow cytometry analysis revealed a high level of CD4⁺ T cell proliferation in co-cultures stimulated with CpG and free MOGR_x, indicated by decreasing signal intensity as dye was diluted from cell proliferation (Fig. 5A). Interestingly, all MOGR_x-GpG complex formulations greatly decreased proliferation, in some cases more than 12-fold, compared to controls containing equivalent doses of MOGR_x peptide without GpG (Fig. 5B, Fig. S4). Further, since CpG alone did not cause any proliferation (Fig. 5B), our results suggest complexes limit proliferation of myelin-specific T cells.

Figure 5.

Complexes reduced the proliferation and inflammatory function of transgenic T cells displaying T cell receptors specific for MOG presented by DCs in the MHC-II complex. Primary DCs were incubated with stimulatory CpG and complex or free treatments prior to coculture with 2D2 T cells. T cells containing a dye were used to assess proliferation by dilution of the dye between daughter cells (A), (B). T cell phenotype was assessed by staining with T_H1 (C) and T_H17 fluorescent antibody conjugates and flow cytometry. **p<.01, #p<.0001, ns = not significant

To investigate if reduced proliferation also alters the phenotype of these T cells, we used flow cytometry to assess T_H1 (CD4⁺/Tbet⁺) and T_H17 (CD4⁺/RORγ⁺) levels, effector populations that drive disease in human MS and mouse models of MS. Incubation with free MOGR_x and CpG induced a high level of inflammatory T_H1 (Tbet⁺) (Fig. 5C) and T_H17 (RORγ⁺) cells (Fig. 5D). As expected, cell numbers and T_H1 levels were only detectable in treatments containing some form of myelin peptide since 2D2 cells are specific for myelin and do not respond to other stimulus. In agreement with our proliferation data, we discovered that treating CpG-stimulated cultures with MOGR_x-GpG complexes significantly reduced T_H1 subsets relative to cells stimulated with CpG and free peptide (i.e., without GpG). Complexes containing either MOGR₁ or MOGR₂ drove these reductions, but the effects were more robust in MOGR₂-GpG complexes where significant reductions in T_H1 levels were measured for all complex ratios (Fig. 5C). Regardless of peptide structure or complex ratio, no effects were observed on T_H17 levels compared with the free controls (Fig. 5D). Taken together, the data in Figures 1–6 demonstrate that complexes reduce TLR9 signaling and deactivate DCs, while limiting myelin-specific T cell proliferation and reducing T_H1 phenotypes associated with pathogenesis during disease.

Figure 6.

Complex treatment reduced disease severity in EAE. (A) In one study, mice were induced with EAE on days 0 and 1 and treated s.c. with complexes on day 7. (B) Disease progression was monitored by daily scoring of paralysis. (C) Comparison of disease severity was completed by averaging the maximum disease score of each mouse. (D) Disease onset was the first day a mouse showed symptoms. (E) In a similar study, mice were instead injected s.c. with complexes on days 6, 12, and 18. Disease progression (F), severity (G), and onset (H) were again measured. *p<.05, **p<.01, ***p<.001, ns = not significant

3.8 MOGR₂-GpG polyplexes improve disease progression and severity during a mouse model of MS (EAE)

Building on our in vitro results, we tested if complexes promote tolerance and control disease in a mouse model of MS (EAE). In these studies, mice received a single s.c. injection of MOGR₂-GpG complexes—since R₂ complexes were generally more potent than R₁ complexes during in vitro studies—on day 7 after EAE induction (Fig. 6A). These mice, along with untreated mice induced with EAE were assessed daily using an established clinical score system and by body weight. As shown in Fig. 6B, mice treated with complexes exhibited a modest, but statistically significant, improvement in disease progression. Additionally, disease severity was improved, as indicated by a significantly reduced mean maximum disease score compared to the untreated group (Fig. 6C). This single complex treatment regimen also delayed onset of symptoms, with a mean time to onset of 13.6 ± 0.5 days for treated mice compared 11.6 ± 0.3 days for untreated mice (Fig. 6D). We next tested if a multi-injection regimen could enhance these therapeutic effects by treating mice with complexes on days 6, 12, and 18 (Fig. 6E). Strikingly, this strategy substantially enhanced efficacy, as indicated by a much greater reduction in the progression and severity of disease in treated mice. The clinical score of treated mice was significantly lower than untreated mice throughout the course of the study (Fig. 6F), with more than a 2.0-point decrease in the mean maximum disease score (Fig. 6G). This three injections regimen also led to a significant decrease in weight loss (Fig. S5) and disease incidence, with only 60% of mice showing symptoms at the end of the study compared to 100% of untreated mice (Fig. 6H). Control studies using similar regimens, but involving three treatments of GpG alone, revealed no therapeutic effect relative to untreated mice (Fig. S6).

4. Discussion

Current treatment options for MS involve the regular injection of immunosuppressive drugs or antibodies that are broadly acting. These options are not curative and can leave patients immunocompromised,[8] highlighting the need for new, more selective therapies. Despite the simplicity and unique features of polyplexes, this general approach has only recently begun to be exploited to promote tolerance. We approached the current challenges to autoimmune therapies using principles from the polyplex field to enable co-delivery through self-assembly of myelin-derived peptide (MOG) and antagonistic ligands for TLRs—pathways increasingly implicated in both human and animal autoimmune disease.[23–27] Our strategy is unique in several ways. First, the reports highlighted in the introduction utilize plasmid DNA encoding regulatory signals,[36, 37] requiring internalization, endosomal escape, transcription, and translation, or DNA designed to expand T_REGS by avoiding PAMPs to support regulatory pathways (e.g., indoleamine 2,3-dioxygenase induction, IDO).[38–40] We have taken an alternative approach, incorporating peptide self-antigen (i.e., MOG) into polyplexes to create a direct route for generation of antigen-specific tolerance without need for gene expression. Second, we are directly targeting a new pathway, TLR signaling, to limit the pro-immune signaling during differentiation of self-reactive T cells to bias these cells away from inflammatory populations (e.g., T_H1). This is of particular relevance for the polyplexes we have designed since TLR9 is expressed in endosomes, eliminating the need for endosomal escape. [50] Along these same lines, our approach does not require transcription or translation of either a self-antigen (i.e., MOG) or a regulatory cue (i.e., GpG) since each of these components are already the active molecules. We propose that following polyplex uptake by APCs, GpG blunts inflammatory TLR9 signaling as myelin-derived peptide is being presented to naïve myelin-specific T cells (Fig. 7). This altered signaling may still allow proliferation of MOG-specific T cells, but polarize these cells away from inflammatory subsets (e.g., T_H1, T_H17) and toward T_REGS that can migrate to the CNS or other sites of disease to control inflammation (Fig. 7). Lastly, the polyplex-like structures we have developed eliminate all synthetic carrier components, and instead, are constructed entirely from immune signals. This removes the potential for intrinsic inflammatory activity associated with many common polymers that could make disease worse while also simplifying composition since the signals are also the carriers and condensation agents.

Figure 7.

In multiple sclerosis, myelin peptide is presented to naïve T cells by activated DCs with strong TLR9 signaling, leading to the proliferation of inflammatory MOG-reactive effector T cells. MOGR_x-GpG polyplexes are taken up by antigen presenting cells. GpG blunts inflammatory TLR9 signaling as MOG is presented to naïve T cells. The combination of immune signals induces the proliferation of MOG-specific T_REGS that promote tolerance.

Beyond polyplex-like materials, there are a number of exciting strategies developing at the interface of biomaterials and autoimmune therapy. One important question is how nanoparticles are processed compared with soluble signals. Shea, Miller, and colleagues have shown that conjugation of self-antigens to polymeric nanoparticles helps direct these particles to macrophage scavenger receptors typically involved in clearance of apoptotic cells debris, material which clearly needs to be recognized as self.[13, 51] Thus, this altered trafficking results in activation of downstream tolerance pathways including deletion, anergy, and T_REG expansion. The polyplex studies by the Mellor lab focus on induction of the regulatory IDO pathway through delivery of DNA that does not contain inflammatory cues typically present in pathogens.[38–40] Instead, these particles are surveyed by the stimulator of IFN genes (STING) pathway in DCs, leading to IDO expression and T_REG activation. In yet a different recent study, tolerance was induced simply by altering the surface charge of synthetic polymeric particles to exhibit negative surface charge.[51] Together these examples highlight both the potential of biomaterials for studying and treating autoimmunity, as well the breadth of questions that require further study—design guidelines, roles of self-antigen and regulatory cues, the frequency of delivery and durability of tolerance, to name a few.

In cell culture, nanoparticles formulated with self-antigen and plasmid encoding a lymphocyte attenuator have recently been used to condition DCs in vitro for promoting tolerance after adoptive transfer.[52] Motivated by the potential of this idea, but striving to eliminate the need for cell isolation or ex vivo manipulation, we began by applying classic polyplex design and characterization techniques to nanoparticles assembled entirely from MOG-derived peptides and regulatory TLR9 antagonists. Interestingly, we discovered that the polyplexes could be formed over a range of charge ratios and surface charges (Fig. 1B) without significantly altering the diameter (Fig. 1A). This finding revealed some interesting implications during our in vitro studies. First, MOG peptide was internalized at similar levels in both free and complexed form, whereas GpG was internalized at much higher levels when complexed with MOGR₂ (Fig. 3) This likely results from the divergent charges and molecular weights of these molecules, with MOGR₁ (MW = 3240.69) and MOGR₂ (MW = 3396.88) exhibiting total net charges of +4 and +5, respectively, while GpG (MW = 7215.8) has a charge of −22. Thus, while free GpG is highly negatively charged, the peptide is cationic and condensation results in complexes that are less anionic. This feature of MOGR_x is attractive because it supports uptake by endocytosis, and, In particular, TLR9 is expressed in endosomes. This idea also highlights the possibility of modulating immune function along another dimension by using the composition of the polyplexes to change the physicochemical properties.

In the classic polyplex field, polyplexes are often formed by mixing the components and assuming all of the molecules become complexed. In our work, we purified the polyplexes by centrifugation to allow a precise measurement of the relative composition (i.e., peptide vs. GpG content) after polyplex formulation. This methodology is important because changing the relative input of each component could lead to complexes with different compositions, subsequent changes in physicochemical properties, and ultimately, impact the types of immune responses that develop. The latter point is particularly relevant as high levels of self-antigen in isolation might create situations that could exacerbate autoimmune disease, while high doses of regulatory signals alone might drive broad suppression. In our current proof-of-principle report, the goal was to understand if self-assembled immune signals that mimic useful features of polyplexes can be used to target toll-like receptor signaling to promote tolerance. In future studies, however, teasing out the link between polyplex composition and the balance between autoimmunity and tolerance will be informative to designing therapies that are both effective and selective.

We also observed that polyplexes formed from GpG and MOG modified with R₂ generally performed better (i.e., greater uptake and DC deactivation, reduced T cell proliferation) than MOGR₁-GpG complexes (Fig. 3–5). This also might be a result of the decreased negative charge, a hypothesis that could be directly tested by studying peptides modified with additional arginine residues. For example, in the polyplex field, endocytosis is typically the mechanism of cell entry, while cell penetrating peptides can enter through penetration or other energy-independent mechanisms.[53–55] The latter molecules are characterized by short stretches of mostly cationic amino acids, so appending such sequences to peptides might allow enhanced uptake; perhaps these phenomena even occur in the case of MOGR₂, though mechanistic studies blocking specific internalization routes will be required to elucidate this possibility. Charge can also greatly affect the ability of polyplexes to escape endosomes following cell entry, with a classic motivation for studying polyplexes as gene delivery agents being the buffering capacity of polycations to exploit endosomal acidification and promote endosomal escape by the “proton sponge effect”.[56–58] Although TLR9 is expressed in endosomes, facilitating our current work, future studies could exploit these materials to target cytosolic receptors through endosomal escape or through cell penetration. Future studies will leverage controllable surface charge to compare uptake and trafficking to other organelles as a function of the properties of polyplexes assembled from immune signals. From an immunological or therapeutic perspective, the high and tunable levels of GpG carried into cells through MOGR_x complexation could be important in explaining the ability of complexes to control the inflammatory response to myelin, as a higher amount of GpG entering cells should have a greater effect on inhibiting TLR9 signaling during differentiation of spleen- or lymph node-resident T cells. Since surface charge appears to play a direct role in tolerance,[51] and certainly impacts particle uptake,[59, 60] tunable control over charge could also be useful in optimizing immunological processing or efficacy.

We also observed some interesting differences in how GpG behaved in complexes versus free form. During the TLR9 signaling studies, free GpG reduced TLR9 signaling more than complexed GpG (Fig. 2A). One potential explanation is the strong binding between MOGR_x and GpG (Fig. 1C), which could require a longer time for binding and processing of GpG to TLR9 receptors. Supporting this possibility, MOGR₂—which protected GpG from degradation better than MOGR₁ (Fig. 1D)—did not suppress TLR9 signaling as much as MOGR₁, though we only studied these effects at a single time point (Fig. 2A). Surprisingly, despite better protection from enzymes (Fig. 1D), the EtBr assay suggested similar binding levels (Fig. 1C), so more sensitive measurements of binding affinity might provide additional insight into some of the different effects we observed from GpG as a function of which MOGR_x structure was used to complex GpG. Despite the less efficient suppression of TLR9 signaling when GpG was complexed at some ratios, both complex formulations (i.e., using MOGR₁ or MOGR₂) greatly reduced TLR9 activity, and further, provide other benefits such as improved uptake and protection from enzyme. Importantly in our studies, treatment of cells with free MOGR_x or MOGR_x-ODN polyplexes (Fig. S3) generally had no effect, directly demonstrating the role of GpG and TLR9 signaling in the observed effects. Further, free GpG had no therapeutic effect when injected alone into mice induced with EAE (Fig. S6). Similarly, in co-culture studies, CpG alone did not cause proliferation, while proliferation was observed during treatment with free MOGR_x (Fig. 5B). When GpG was complexed with these peptides, proliferation was greatly diminished. This result indicates that MOGR_x-GpG complexes alter the function of myelin-reactive cells in a myelin-dependent manner, a feature that is important in therapies that might offer antigen-specific control of autoimmune disease. Taken together, these results support an important role for co-delivery of both the self-antigen and the regulatory signal in the polyplex to induce tolerance.

Our approach reveals several important follow-on questions. First, we prepared polyplexes containing a single antigen, but in human disease, multiple epitopes are attacked initially and through the process of epitope spreading. Future studies will need to investigate induction of tolerance against other self-antigens alone or in combination. Along these same lines, the EAE model itself is limited in the reactivity and range of epitopes attacked.[61] Therefore there is also motivation to test polyplexes and other therapeutics in multiple models of autoimmunity. Additionally, we are currently investigating the trafficking and stability of complexes upon cell entry as a function of polyplex composition and the number of arginine residues on the peptide, which could modulate the route of update. These concepts are particularly important in therapeutics involving TLRs at these receptors are differentially expressed in different cell components (e.g., endosomal, cytosolic). Lastly, from a materials perspective, our approach relies on electrostatic assembly, so some self-antigens may require appending charged amino acid anchors if they do not exhibit sufficient inherent charge.[62–64] Not-withstanding these limitations, there is great recent excitement for translation of antigen-specific therapies. For example, a recent clinical trial with human MS patients involved administration of a cocktail of seven myelin peptides coupled to autologous cells as a possible approach to promote myelin-specific tolerance.[5, 65]

5. Conclusion

In this report we demonstrate that immune signals assembled into polyplex-like structures exhibit many cardinal features of traditional polyplexes, but eliminate carrier components. Our results show that GpG suppresses TLR9 signaling during complexation and that GpG is more efficiently internalized when in a complexed form. This enhanced delivery deactivates DCs and limits cytokines (IFN-γ) and proliferation of myelin-specific cell populations (e.g., T_H1) implicated in human MS and animal models. In mice, these complexes improve disease progression, severity, and incidence. Future studies will elucidate the mechanisms of uptake and trafficking in immune cells, as well as the underlying mechanisms of tolerance. Ultimately, this simple approach could contribute to more specific treatment options based on self-assembly of immune signals into structures that offer attractive features of biomaterials, enabling co-delivery of self-antigens and regulatory ligands that block TLR signaling to bias differentiating T cells toward regulatory populations instead of inflammatory effector subsets.

Highlights.

• MOGR_x condenses GpG to form nanosized polyplexes comprised purely of immune signals

• Complexes blunt inflammatory TLR9 signaling and deactivate DCs after CpG stimulation

• Polyplexes control myelin-specific T cell proliferation and reduce T_H1 phenotypes

• Polyplexes promote tolerance and control disease severity in a mouse model of MS