* Department of Chemical Engineering and Materials Science, Wayne State University, 5050 Anthony Wayne Drive, Detroit, Michigan 48202, United States.
Find articles by Meng Li† Division of Rheumatology, Allergy, and Clinical Immunology, University of California, Davis, Davis, California, 95616 United States.
Find articles by Arata Itoh* Department of Chemical Engineering and Materials Science, Wayne State University, 5050 Anthony Wayne Drive, Detroit, Michigan 48202, United States.
Find articles by Jingchao Xi* Department of Chemical Engineering and Materials Science, Wayne State University, 5050 Anthony Wayne Drive, Detroit, Michigan 48202, United States.
Find articles by Chunsong Yu† Division of Rheumatology, Allergy, and Clinical Immunology, University of California, Davis, Davis, California, 95616 United States.
Find articles by Yuehong Wu† Division of Rheumatology, Allergy, and Clinical Immunology, University of California, Davis, Davis, California, 95616 United States.
Find articles by William M. Ridgway* Department of Chemical Engineering and Materials Science, Wayne State University, 5050 Anthony Wayne Drive, Detroit, Michigan 48202, United States.
‡ Department of Oncology, Wayne State University, Detroit, Michigan 48201, United States.
§ Tumor Biology and Microenvironment Program, Barbara Ann Karmanos Cancer Institute, Detroit, Michigan 48201, United States.
Find articles by Haipeng Liu* Department of Chemical Engineering and Materials Science, Wayne State University, 5050 Anthony Wayne Drive, Detroit, Michigan 48202, United States.
† Division of Rheumatology, Allergy, and Clinical Immunology, University of California, Davis, Davis, California, 95616 United States.
‡ Department of Oncology, Wayne State University, Detroit, Michigan 48201, United States.§ Tumor Biology and Microenvironment Program, Barbara Ann Karmanos Cancer Institute, Detroit, Michigan 48201, United States.
1 Current address: College of Pharmaceutical Sciences, Hebei Medical University, Shijiazhuang, 050017, China.
Address correspondence and reprint requests to: Dr. Haipeng Liu, Department of Chemical Engineering and Materials Science, Wayne State University, 5050 Anthony Wayne Drive, Detroit, Michigan 48202, United States. ude.enyaw@uil.gnepiah.
The publisher's final edited version of this article is available at J ImmunolAntigen specific immunotherapy to restore immune tolerance to self-antigens, without global immune suppression, is a long-standing goal in the treatment of autoimmune disorders such as type 1 diabetes (T1D). However, vaccination with autoantigens such as insulin or glutamic acid decarboxylase (GAD) have largely failed in human T1D trials. Induction and maintenance of peripheral tolerance by vaccination requires efficient autoantigen presentation by antigen presenting cells (APCs). Here, we show that a lipophilic modification at the N-terminal end of CD4 + epitopes (lipo-peptides) dramatically improves peptide antigen presentation. We designed amphiphilic lipo-peptides to efficiently target APCs in the lymph nodes by binding and trafficking with endogenous albumin. Additionally, we show that lipophilic modification anchors the peptide into the membranes of APCs, enabling a bivalent cell-surface antigen presentation. Subcutaneously injected lipo-peptide accumulates in the APCs in the lymph node, enhances the potency and duration of peptide antigen presentation by APCs, and induces antigen-specific immune tolerance that controls both T-cell- and B-cell-mediated immunity. Immunization with an amphiphilic insulin B chain 9–23 peptide (Lipo-B9–23), an immunodominant CD4 + T cell epitope in NOD mice significantly suppresses the activation of T cells, increases inhibitory cytokine production, induces regulatory T cells (Tregs), and delays the onset and lowers the incidence of T1D. Importantly, treatment with a lipo-beta cell peptide cocktail delays progression to end-stage diabetes in acutely diabetic NOD mice, while the same doses of standard soluble peptides were not effective. Amphiphilic modification effectively enhances antigen presentation for peptide-based immune regulation of autoimmune diseases.
Keywords: antigen presentation, type 1 diabetes, immune tolerance, amphiphilic peptide, albuminType I diabetes (T1D) is a chronic autoimmune disorder characterized by destruction of insulin-producing beta-cells within pancreatic islets, and affects nearly 1.3 million people in the United States. (1) Current therapies for T1D alleviate the symptoms of hyperglycemia but do not cure the disease. (2, 3) Insulin replacement therapy does not stop the autoimmune attack on beta-cells and has limitations including serious complications that lead to significant mobility and mortality. (3) Transplantation of islet cells from healthy donors is an alternative therapy, but suffers from drawbacks including limited availability of islets cells and the requirement for lifelong immunosuppression. (4) Antigen specific immunotherapy (ASI) has been the paradigmatic goal for treating T1D for several decades, as it strives to restore self-tolerance by targeting only the autoreactive T cells in the adaptive immune repertoire while leaving responses to foreign antigens intact. (3, 5–8) ASI relies on the presentation of self-antigens and immune modulators (tolerogenic vaccines) to autoreactive T cells in order to down-regulate the autoaggressive immune responses. ASI acts either directly on effector T cells (via clonal deletion or anergy) or on antigen specific regulatory T cells that down-modulate immune response by a number of mechanisms. (3, 5–8) Specificity is the main advantage of ASI: it allows for selectively targeting of disease-relevant T cells, while leaving the immune system intact. Despite intensive research in this field, to date, no FDA approved ASI effectively treats T1D patients. Multiple trials of ASI in human T1D have essentially failed. Induction of antigen-specific tolerance to dominant immune responses driving autoimmunity remains an unmet challenge.
In ASI, autoantigen presentation by antigen presenting cells (APCs) is essential for the induction and maintenance of peripheral tolerance. (9) In the steady state characterized by low level of expression of co-stimulatory molecules, antigen localization, quantity/quality, and kinetics determine the fates of T cells in the secondary lymphoid tissues. (10–15) While low levels of auto-antigen presentation in the lymph nodes (LNs) leads to immunological ‘ignorance’, a passive mechanism that fails to regulate auto-reactive T cells, efficient antigen-presentation leads to active tolerance, whereby clonal deletion, anergy or conversion to regulatory T cells (Tregs) prevail. (9–15) Current strategies of ASI in T1D face major challenges in antigen presentation. First, autoantigens introduced parenterally fail to reach LNs, (16) where the dominant immunological decisions are being made. Second, the major autoantigens (e.g., insulin) are poorly presented to T cells. In nonobese diabetic (NOD) mice, the initiating diabetogenic epitope insulin B chain 9–23 (B9–23) peptide binds poorly (with micromolar affinity and rapid dissociation rate) to IA g7 . (17–20) Though diabetogenic T cells respond to wildtype insulin B9–23 peptides presented by APCs in vitro, the levels of stimulation are low and inefficient. (21–23) Thus, quantitatively and qualitatively insufficient antigen presentation results in insufficient T cell tolerization to autoantigens, and represents one of the major hurdles in current ASI.
Several approaches are currently being tested to enhance autoantigen presentation to induce antigen-specific tolerance, for example autoantigens delivered by nanoparticles, (24–27) or antibodies (28, 29) which target dendritic cells (DCs), to enhance the antigen uptake, processing, and presentation. Efficient delivery of antigens to DCs (the most efficient antigen presenting cells) in the absence of costimulatory stimuli promoted tolerance induction in murine model of T1D. (25) Antigen co-delivered with small-molecule drugs which modulate DC function is another popular approach for prophylactic and therapeutic vaccines that can drive antigenic tolerance. (27) Recently, targeting DCs in the LNs and modulating DC-T cell interactions restored T1D tolerance. (30) Lymph nodes house abundant DCs/lymphocytes and are the primary anatomic sites where the inflammatory/regulatory fate of T cell responses is determined. (31) For example, intralymphatic injection of glutamic acid decaroboxylase (GAD65), another major autoantigen in T1D, led to dramatically prolonged preservation of beta-cell function as compared to subcutaneous injection in a small clinical study, likely through the enhanced antigen presentation to T cells in the lymph nodes. (30) Another intriguing approach to overcome the low antigen presentation is to use altered antigen ligands or neoantigens. For example, strong agonist epitopes obtained by post translational modifications of insulin peptide on the C-terminal (neoantigen) induced antigen-specific tolerance in NOD mice. (32, 33) These modifications likely affect the C-terminus MHC anchor residue, resulting in stable binding of peptide in a register that is otherwise unstable. (33) Likewise, infusion of small amounts of a strong agonistic insulin B9–23 mimetope modifying the MHC anchor residue at position 9 (R22E) completely prevented the onset of T1D in NOD mice. (23) These important studies demonstrated that efficient antigen presentation under subimmunogenic conditions (low level of co-stimulation) is indispensable for tolerance induction in vivo.
Here, we report a molecular approach to enhance the CD4 + peptide presentation by simultaneously controlling the localization, quantity, and quality of antigen presentation in vivo. We achieve this by conjugating peptide antigen with a diacyl lipid linked by a solubility-promoting polyethylene glycol linker. The lipid moiety tagged on self-peptide fulfills a dual role: 1) it efficiently delivers peptide antigens to DCs in the LNs by hitchhiking on albumin protein (albumin-hitchhiking) and 2) it enhances peptide/MHC presentation by anchoring peptide antigen on the DC surface, resulting in a bivalent peptide agonist simultaneously improving the potency and duration of antigen presentation. Amphiphilic modification of antigenic peptide provides a broadly applicable approach for immune modulation. Its application goes beyond the treatment of autoimmunity to other therapies such as cancer, infection, and organ transplantation, which require efficient antigen presentation.
All chemicals were purchased from Sigma-Aldrich unless noted otherwise. Chemicals were used without further purification. All peptides were custom synthesized by GenScript (Piscataway, NJ). OVA peptide (323–339) consisted of amino acids CISQAVHAAHAEINEAGR, 5-FAM (5-Carboxyfluorescein) labeled OVA peptide (323–339) consisted of amino acids 5-FAM-Ahx-CISQAVHAAHAEINEAGR. Other peptides with cysteine-tagged N termini included: OVA328–339: CHAAHAEINEAGR, B9–23: CSHLVEALYLVCGERG, ZnT8345–359: LTIQIESAADQDPSC, GAD65524–543: CSRLSKVAPVIKARMMEYGTT. 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[maleimide(polyethylene glycol)-2000] (DSPE-PEG2000-Maleimide) was obtained from Laysan Bio Inc. (Arab, AL). Lipo-CpG were synthesized in house using an ABI 394 DNA/RNA synthesizer. Murine MHC class I tetramers were obtained from MBL international Corporation (Woburn, MA). Antibodies were purchased from eBioscience (San Diego, CA) or BD Bioscience (San Jose, CA).
Animals were housed in the United States Department of Agriculture (USDA)-inspected Wayne State University or University of California, Davis animal facilities under federal, state, local and NIH guidelines for animal care. Female C57BL/6 mice (5–8 weeks), NOD/ShiLtJ mice (5 weeks) and OT-II mice were obtained from the Jackson Laboratory. Bone marrow dendritic cells (BMDCs) were generated as described by a modified protocol. (34) Cells were cultured in complete medium (MEM, 10% fetal bovine serum (Greiner Bio-one), 100 U/mL penicillin G sodium and 100 μg/mL streptomycin (Pen/Strep).
5 mg peptides were mixed with 2 equivalent DSPE-PEG2000-Maleimide in 1 mL Dimethylformamide (DMF). 3 μL triethylamine (TEA) was added and the solution was stirred in the dark at 25 °C for 24 hours. The solution was dried in a stream of air for 72 h to evaporate DMF. The remaining reaction residues were dissolved in 5 mL phosphate buffered saline (PBS) buffer (0.1 M, pH = 7.4) with sonication. Lipo-peptides were purified by reverse-phase HPLC with a C4 column (Thermo Scientific, 250 × 4.6 mm, 5 μm).
A gel electrophoresis mobility shift assay was used to detect albumin protein binding with lipo-peptide. The solution of free 5-Fam-Ahx labeled OVA323–339 and lipo-OVA323–339 were incubated with Fetal Bovine Serum (FBS) (estimate molar ratio 1:1) for 4 hours at 37°C. Samples were loaded for electrophoresis run under 75 V for 30 min through native 0.7% agarose gel. Images were recorded using a digital camera under UV illustration for fluorescein labeled peptide, or briefly stained with coomassie blue for protein.
The cell uptake of free OVA323–339 and lipo-OVA323–339 was examined in BMDCs by flow cytometry. Cells were seeded to 48-well plate (1 × 10 6 cells per well) and incubated at 37°C for 24 hours. The cell medium was removed and replaced with 5-FAM-Ahx labeled OVA323–339 and lipo-OVA323–339 at a final concentration of 1.0 μM for 2 hours. The cells were harvested and washed with 1 × PBS buffer three times and analyzed by flow cytometry using an Attune acoustic focusing cytometer (Applied Biosystems). Each assay was performed in triplicate. Cells were also examined by confocal microscopy. Images were captured by Zeiss confocal microscope (LSM 780).
Each well of a round-bottom 96-well plate was seeded with BMDCs (day 7). Cells were pulsed with different concentrations of OVA323–339 or lipo-OVA323–339 peptides for 2 hours. Cells were then washed and cocultured with freshly isolated OT-II CD4 + T cells (DC/T cell ratio 1:2) for 48 hours. Supernatant was transferred and stored at −80 °C prior to IL-2 quantification by ELISA assays. All experiments were performed in triplicate. In some cases, BMDCs were fixed with 1% PFA at room temperature for 30 min, or treated with NH4Cl (200 μM) at 37 °C in culture medium for 45 min.
Groups of C57BL/6 mice (n=3–4 LNs/group) were injected subcutaneously at the tail base with 3.3 nmol of FAM-labeled free or lipid-modified OVA323–339. After 24 h, animals were sacrificed and inguinal and axillary LNs were excised and digested with 1.5 mL freshly prepared enzyme mixture comprised of RPMI-1640, 0.8 mg/mL Collagenase/Dispase (Roche Diagnostics) and 0.1 mg/mL DNase (Roche Diagnostics). LNs cells were stained with antibodies against CD11c versus peptide fluorescence in viable cells. Percentages of peptide positive cells in the LNs were determined by flow cytometry.
On day 0, C57BL/6 mice (6–8 weeks; n=3–4 per group) were stimulated with ovalbumin protein (10 μg) and lipo-CpG (1.24 nmol) and tolerized OVA323–339 (10 μg) or lipo-OVA323–339 on days 7 and 17. On day 21, mice were challenged with Ovalbumin (10 μg) and lipo-CpG (1.24 nmol). On day 28, mice were sacrificed, and OVA-specific immune responses were evaluated. The volume of all vaccine injections was 100 μL/animal. All injections were performed s.c. (subcutaneously) at the base of the tail.
For T1D prevention in NOD mice, female NOD mice were randomly divided into three groups (n = 12/group). Experimental groups were as follows: B9–23; lipo-B9–23; and PBS control. Mice received two subcutaneous injections of soluble B9–23 peptide (40 μg peptide/injection, total volume 100 μL), or same equimolar lipo-B9–23 peptide, or PBS at 6 and 8 weeks of age. Blood glucose levels were measured once per week before 12-week of age, but twice per week after that with a OneTouch® Ultra® 2 blood glucose meter (Lifescan, Inc., Milpitas, CA, USA). If the glucose level was higher than 250 mg/dL for two consecutive days, the animal was considered to be diabetic.
For treatment of acute T1D, NOD mice were randomly assigned to PBS, soluble peptides, or lipo-peptides groups, and followed for onset of T1D as defined by 2 serial BG measurements of 250 mg/dL. New-onset diabetic NOD mice were treated with lipo-peptide cocktail (20 μg each peptide/injection), or soluble peptide cocktail, or PBS. Mice received 2 injections in the first week, followed by one injection/week for 9 weeks. Blood glucose levels were followed over time. Mice were sacrificed at endpoint or when BG exceeded 500 mg/dL.
Seven days after challenge, blood samples were collected and red blood cells were lysed by ACK lysing buffer. Cells were then blocked with Fc-blocker (anti-mouse CD16/CD32 monoclonal antibody) and stained with human papillomavirus HPV-16 E749–57 (RAHYNIVTF) loaded phycoerythrin-labeled tetramer (Beckman Coulter) and anti-CD8-APC (ebioscience) for 30 minutes at room temperature. Cells were washed twice, resuspended in FACS buffer, and analyzed on Attune Focus flow cytometer. Analysis typically gated on live, CD8 + , Tetramer positive cells.
Cells were plated in 96-well round-bottomed plates and pulsed with peptide antigen for 6 hr at 37°C in T-cell media (RPMI 1640, 10% FBS, 50 μM β-mecaptoethanol, 100 U/mL Penn/Strep, 1x Gibco® MEM Non-Essential Amino Acids Solution (Life Technologies), 1 mM Sodium pyruvate, 1 mM HEPES), followed by the addition of brefeldin A for 4 hours. Cells were stained with anti-CD4-APC and then fixed using Cytofix (BD biosciences) according to the manufacturer’s instructions. Cells were then washed and permeabilized. Intracellular staining for anti-IFN-γ-PE was performed according to the manufacturer’s instructions. Foxp3 staining was performed according to the manufacturer’s instructions (ebioscience) for fixation and permeabilization.
Serum or cell culture supernatant levels of antibodies and cytokines were determined by ELISA: 96-well plates were coated overnight with capture antibodies in PBS and blocked with 1% BSA in PBS. After incubation of serum samples for 1h at a series of dilutions, plates were washed with PBS/1% Tween 20. Goat anti-mouse IgG conjugated to Horseradish peroxidase (HRP) was added at 1 μg/ml for 30 min. Plates were washed with PBS/1% Tween 20 and ELISA was developed by (3,3’,5,5’-Tetramethylbenzidine) (TMB, ebioscience). The reaction was stopped by 1 M H2SO4 and the absorbance was read at 450 and 570 nm using a plate reader.
All plots represent mean values and error bars represent the standard error of the mean (SEM). Comparisons of mean values of two groups were performed using unpaired Student’s t-tests. One-way analysis of variance (ANOVA), followed by a Bonferroni post-test was used to compare >2 groups. *, p
We recently discovered an ‘albumin-hitchhiking’ approach which uniquely targets subunit vaccines to antigen presenting cells in the draining lymph nodes. (35, 36) The effectiveness of this vaccine approach relies on molecular design of lipophilic peptide (lipo-peptide) that enhances entry into the traffic pathway of endogenous albumin in the interstitial fluid. Following subcutaneous injection, lipo-peptide vaccines bind avidly to albumin, transport to draining LNs via lymph circulation, and efficiently accumulate in APCs including DCs. To test whether this approach can be translated to deliver a CD4 epitope for immune regulation, we conjugated OVA323–339, an ovalbumin derived, MHC II restricted peptide, to the amphiphilic polymer DSPE-PEG2000, following our previously published procedures. (36) The DSPE-PEG2000-OVA323–339 (referred as lipo-OVA323–339) was purified by reverse phase HPLC and characterized by mass spectometry. The interaction between lipo-peptide and albumin was characterized by electrophoretic mobility shift assay. (37) 5-Carboxyfluorescein-labeled lipo-OVA323–339 and OVA323–339 were incubated with fetal bovine serum (FBS) and analyzed by native gel electrophoresis. FBS showed a major albumin band visible after coomassie staining ( Fig. 1A , lane 5). Lipo-OVA323–339 bound to serum albumin, showing a slower-moving fluorescent band visible under UV, co-migrated with albumin as compared with lipo-OVA323–339 ( Fig. 1A , lane 3–5). In contrast, OVA323–339 incubated with FBS showed no decrease in relative mobility ( Fig. 1A , lane 1 and 2), indicating a lack of interaction with albumin.
(A) Fluorescein labeled lipo-OVA323–339, or OVA323–339 were incubated with FBS at 37 °C for 4 hours and analyzed by native gel electophoresis (0.8% agarose). The peptide bands were visualized by photograph under UV; protein bands were stained with commassie blue. Lane 1: OVA323–339; lane 2: OVA323–339 + FBS; lane 3: lipo-OVA323–339; lane 4: lipo-OVA323–339 + FBS; lane 5: FBS. The amount of FBS was titrated with lipo-peptide to determine the albumin concentrations. (B) Fluorescein labeled peptides were injected s.c. at the tail base of C57BL/6 mice (n = 3–4 LNs/group), inguinal and axillary nodes were isolated 24 h after injection and analyzed by flow cytometry. Data show the mean values ± SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001 by one-way ANOVA with Bonferroni post-test.
We next characterized LN accumulation of peptides after injection. Fluorescein labeled lipo-OVA323–339 and OVA323–339 were subcutaneously injected into C57BL/6 mice (n = 4 LNs per group). 24 h later, inguinal and axillary LNs were excised and analyzed by flow cytometry. Albumin-binding lipo-OVA323–339 exhibited dramatically increased accumulation in DCs in both inguinal nodes and axillary nodes after injection. By contrast, negligible amount of unmodified OVA323–339 was detected in DCs in the lymph nodes ( Fig. 1B and Supplemental Fig. 1).
Amphiphilic modification alters the physicochemical properties of peptide and might affect peptide uptake, processing, and presentation. We thus investigated the uptake and antigen presentation of amphiphilic peptide by GM-CSF (granulocyte-macrophage colony-stimulating factor) induced bone marrow dendritic cells (BMDCs). Unlike unmodified peptide, lipid-conjugated peptide exhibits high affinity towards the membrane bilayer. In the presence of albumin and cells, the lipid-modified peptide equilibrates between an albumin-binding state and a membrane-anchoring state. (35, 36, 38–40) In cell culture, compared with unmodified peptide, lipo-OVA323–339 exhibited a 10-fold increase in DC uptake, with a significant portion of the peptide anchored on the membrane surface 1 h after incubation ( Fig. 2A , ,B). B ). This observation suggests a high density of lipo-OVA323–339 anchors on the surface of DCs ( Fig. 2A ).
(A) and (B), Confocal images (A) and uptake quantifications (B) of BMDCs after 1h incubation with dye labeled OVA323–339 (A, left) or lipo-OVA323–339 (A, right) showing high density of membrane-anchored peptide on cell surface. (C), Schematic illustration showing CD4 lipo-peptide anchors on the cell surface and directly loads onto MHC-II. Unmodified peptide displays transient interaction (low affinity and short half-live) with MHC, resulting in insufficient presentation to T cells (upper panel). In contrast, membrane anchored peptide acts as an antigen reservoir, enhancing the binding and presentation via an additional association with cell membrane. Membrane-anchor enables re-bind as peptide and MHC disengage (lower panel). (D) BMDCs were treated with NH4Cl, or fixed with paraformaldehyde (PFA), washed, and pulsed with (2 μg/mL) lipo-OVA323–339 or unmodified OVA323–339 for 2h, OT-II T cells were then added and co-cultured for 48 h. T cell responses were quantified by ELISA measurement of IL-2 production. (E), BMDCs were pulsed with different concentrations of lipo-OVA323–339, or soluble OVA323–339, washed and incubated with OT-II T cells. IL-2 was measured at 48 h. (F), BMDCs were pulsed with 10 μg/mL lipid-OVA323–339 or unmodified OVA323–339 for 2h, washed, and cultured for the indicated times to allow peptide/MHC to decay. OT-II T cells were then added and co-cultured for 48 h. T cell responses to DCs loaded with CD4 epitope were quantified by ELISA measurement of IL-2 productions. Scale bar: 20 μm. Data show the mean values ± SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001 by unpaired student’s t-test.
The observation of direct membrane anchoring of lipo-OVA323–339 but not soluble OVA323–339 led us to investigate how the amphiphilic peptides were presented by DCs. Because both ends of MHC II are open, it is possible that the lipo-OVA323–339 interacts with MHC II by a hetero-bivalent interaction: the peptide moiety binds to MHC-II while the lipid tail anchors the peptide on DC membrane ( Fig. 2C ). To test this hypothesis, BMDCs were treated with NH4Cl, or fixed with paraformaldehyde (PFA), washed, and pulsed with 2 μg/mL lipo-OVA323–339 or same molar amount of unmodified OVA323–339 for 2 h. Both NH4Cl and PFA inhibit antigen processing but do not affect the surface peptide loading and presentation. (41) OT-II CD4 + T cells which recognize OVA323–339 in the context of I-A b were then added and co-cultured for 48 h. T cell responses were quantified by ELISA measurement of IL-2 production. As shown in Figure 2D , lipo-peptide can directly bind to the MHC-II pocket without being processed, since chemically inactivated, GM-CSF induced BMDCs loaded with low concentration (2 μg/mL, unsaturated) of lipo-OVA323–339 exhibited significantly enhanced activation of OT-II T cells compared with soluble OVA323–339 in vitro ( Fig. 2D ). At a higher peptide concentration (saturated, 10 μg/mL), both lipo-OVA323–339 and soluble OVA323–339 activated OT-II T cells at similar level (Supplemental Fig. 2A). Thus lipid modified CD4 epitope can be loaded directly onto MHC-II and can be presented intact to T cells without the need for enzymatic degradation and processing.
Several crucial parameters such as peptide/MHC ligand density, potency, and duration define a cumulative quantity of TCR stimulation which governs the initial induction and persistence of peripheral T cell tolerance. (42, 43) We next assessed the impact of lipid modification on the potency and duration of peptide presentation. BMDCs were pulsed with different concentrations of lipo-OVA323–339 vs. soluble OVA323–339, washed and incubated with OT-II T cells. Peptide stimulation was increased in response to low concentrations of lipo-OVA323–339, yielding an EC50 value which was 3-fold lower than that of unmodified OVA323–339 (0.5 vs. 1.5 μg/mL) ( Fig. 2E ). To test whether membrane-anchored lipo-OVA323–339 can prolong the antigen presentation, BMDCs were pulsed with 10 μg/mL (saturated) lipo-OVA323–339 or equal molar amounts of free OVA323–339, washed, and cultured for various time periods to allow peptide/MHC-II to decay. OT-II T cells were then added and antigen presentation was determined by quantification of IL-2 secretion following 48 h of co-culture. As shown in Figure 2F , lipo-OVA323–339 peptide markedly prolongs antigen presentation, showing 7-fold half-life extension of T-cell stimulation compared to DCs pulsed with soluble OVA323–339 in equal molar amounts. Pulsing with uncoupled OVA323–339 results 50% less IL-2 release by 5h, primarily due to peptide-MHC turnover/dissociation/degradation. At 24 h after peptide pulsing, OVA323–339 had lost > 95% of its ability to activate T cells ( Fig. 2F ). By contrast, the half-life of lipo-OVA323–339 peptide for T cell stimulation was 37 h. At 24 h after pulsing, lipo-peptide retained 70% antigen presentation capability ( Fig. 2F ). Thus, lipid modified antigen can enhance MHC-II presentation not only by delivering a high concentration of peptide antigen to DCs in the LNs, but also by providing a unique hetero-bivalent interaction with MHC-II on DCs: the peptide moiety binds to MHC-II while the lipid tail anchors the peptide on DC membrane. Even if the peptide dissociates from MHC molecule, the membrane anchor helps the peptide remain on the cell surface, increasing the local concentration of peptide and permitting the reengagement with MHC ( Fig. 2C ). Notably, OVA323–339 binds stably to IA d , the MHC II in C57BL/6 mice with a relative high affinity and long half-life. Since most autoantigens exhibit weak MHC binding, this observation based on OVA323–339 might not predict the antigen presentation with a low-stability peptide. To test whether enhanced presentation can be obtained by lipid conjugation of a low-stability peptide, a truncated OVA328–339 was used (OVA328–339). This variant of OVA323–339 is known to bind IA d with reduced stability. (44–46) Although the overall activation level was lower than wild type OVA323–339, OT-II T cell stimulation by lipo-OVA328–339 was significantly higher than its unmodified counterpart OVA328–339 (Supplemental Fig. 2B). Together, these data revealed that in vitro, lipid modification altered the physicochemical properties of peptides and subsequently improved their uptake, presentation, and downstream T cell activation.
Given the dramatically enhanced LN targeting, APC accumulation, and antigen presentation of lipo-peptide, we tested the ability of molecular vaccines consisting of lipo-peptide antigen in the absence of co-stimulation to induce antigen-specific T and B cell immunological tolerance. Our hypothesis is that efficient delivery of antigen to DCs in the lymph node under homeostatic conditions would promote effective tolerogenic antigen presentation and subsequently lead to the inhibition of self-reactive immunity. A major concern for antigen specific immunotherapy is the ability to induce and maintain tolerance in the presence of inflammatory stimuli caused by infection, tissue damage or autoimmunity. To investigate whether amphiphilic antigens could inhibit T-cell activation under inflammatory conditions, C57BL/6 mice were immunized with ovalbumin (OVA) protein, combined with a potent LN targeting CpG adjuvant (lipo-CpG) which stimulates toll-like receptor 9 ( Fig. 3A ). (35, 38) Mice were tolerized twice with low doses of peptide antigen before a final challenge. Unmodified OVA MHC-II peptide (OVA323–339) and N-terminal conjugated OVA peptide (lipo-OVA323–339) reduced the frequency of OVA-specific CD4 + T cells in blood by 2- and 3-fold, respectively ( Fig. 3B ). Mice treated with lipo-OVA323–339 showed a statistically significant inhibition of IFN-γ producing CD4 + T cells compared to mice treated with soluble OVA323–339 ( Fig. 3B ). An increase in the percentage of CD4 + CD25 + Foxp3 + T cells in the blood was evident after lipo-OVA323–339 treatment ( Fig. 3C ), indicating the induction of T regulatory cells by peptide vaccines. Similar inhibitory effect on CD4 + T cell activation was observed in an adoptive T cell transfer experiment, where lipo-OVA323–339 significantly reduced the frequencies of CFSE + OT-II T cells in the draining LN ( Fig. 3D ). These results suggest that administration of amphiphilic peptide leads to the reduction of effector T cells as well as induction of regulatory T-cells.
(A) C57BL/6 mice were immunized with OVA protein mixed with CpG and two subgroups were tolerized with 10 μg dominant MHC-II peptide (OVA323–339) vaccine, or lipo-OVA323–339. Blood was collected 7 days post challenge and assayed for effector CD4 + T cell (B), CD4 + CD25 + Foxp3 + Treg cell (C) and anti-OVA IgG responses (D). (E), C57BL6 mice were tolerized with two s.c. injections of 10 μg amph-OVA233–249 peptide. 7 days after final tolerization, mice were challenged with a HPV peptide or KLH protein mixed with CpG adjuvant. Blood was collected 7 days post challenge and assayed for HPV-specific CD8 + T cell (F), KLH-specific CD4 + T cell (G) and anti-KLH IgG responses (H). Data show the mean values ± SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001 by one-way ANOVA with Bonferroni post-test.
We next investigated the impact of lipo-OVA323–339 peptide treatment on the development of a humoral response. C57BL/6 mice were immunized as above, and tolerized with two doses of soluble peptide or lipo-peptide. Treatment with peptide vaccines significantly reduced the anti-OVA IgG titers in the blood compared to the non-tolerized treatment group ( Fig. 3E ). The strongest inhibition was observed in mice treated with the amphiphilic MHC class II-restricted OVA peptide 323–339 ( Fig. 3E , lipo-OVA323–339). To determine whether the tolerance induction was antigen specific, both MHC class I- and class II-restricted lipid conjugated non-OVA peptide epitopes were used to replace OVA323–339 peptide. None of the irrelevant peptides suppressed the anti-OVA immunity (Supplemental Fig. 3), suggesting the tolerance induction is antigen specific.
Studies have also highlighted the importance of the route of antigen administration in tolerance induction. (5) To investigate whether tolerance can be induced by systemic exposure of peptide antigen, mice were tolerized with intravenous (i.v.) administration of lipo-OVA323–339. We observed slightly lower levels of tolerance induction when lipo-peptide was injected i.v., as compared with s.c. route (Supplemental Fig. 4).
Global immunosuppression is an undesirable side effect in many immunotherapies. To test whether lipo-peptide vaccines compromise the global immune responses, animals were tolerized with OVA323–339 peptide and challenged with irrelevant antigens (HPV peptide or Keyhole limpet hemocyanin (KLH) protein) adjuvanted with CpG DNA ( Fig. 3F ). Compared with no treatment group, both T-cell proliferation and antibody production related to irrelevant antigens were not affected in lipo-OVA323–339 treated animals ( Fig. 3G – I ). These results support the notion that amphiphilic peptide vaccine suppresses the immunity in an antigen-specific manner without compromising the global immune responses.
We next tested the ability of lipo-peptide vaccine to prevent the development of type 1 diabetes in the NOD mice. Two injections (at 6 and 8 weeks of age) of amphiphilic insulin B chain 9–23 peptide (lipo-B9–23), an immunodominant T cell epitope in NOD mice (47, 48) delayed the onset and lowered the incidence of type 1 diabetes ( Fig. 4A ). Blood lymphocyte analysis at 10 weeks of age indicated that treatment with lipo-B9–23 significantly reduced the percentage of B9–23 reactive, IFN-γ producing CD4 + T lymphocytes after B9–23 restimulation, and increased the frequency of Foxp3 expressing regulatory T cells ( Fig. 4B ). Additionally, treatment with lipo-B9–23 significantly reduced inflammatory IL-12 production, whereas production of inhibitory transforming growth factor-beta (TGF-β) was elevated ( Fig. 4C ), compared to animals treated with PBS or unmodified B9–23 peptide. These results demonstrate that targeting peptide autoantigen to DCs in the draining LNs via amphiphilic modification supports the induction of tolerogenic mechanisms, delaying the onset and reducing the incidence of T1D in NOD mice.
NOD mice were treated with two (on week 6 and 8) s.c. injections of 20 μg B9–23 peptide, amphiphilic B9–23 peptide (lipo-B9–23) or PBS, blood glucose concentrations were monitored. (A) Percentage of diabetic mice (n=12 for each group, p values were shown by log-rank test). (B) At the age of week 10, percentage of IFN-γ + secreting CD4 + T cells and Foxp3 + CD25 + CD4 + T cells in blood. (C) At week 10, serum cytokine levels were measured by ELISA. Data show the mean values ± SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001 by one-way ANOVA with Bonferroni post-test.
Most T1D patients have significant β-cell destruction before clinical diagnosis, requiring a therapy which delays or halts progressive disease. (49) Additionally, the fact that multiple autoantibodies are detectable in T1D patients (50) strongly suggests epitope spreading has occurred, reducing effectiveness of tolerogenic vaccines targeting a single epitope. To overcome these issues, and determine whether LN-targeting vaccine strategy can ameliorate diabetes in newly diagnosed diabetic mice, we treated new-onset diabetic NOD mice (blood glucose level 250 +/− 20 mg/dL) within 1 week of diagnosis with a lipid-modified peptide cocktail comprising CD4 epitopes from insulin (B9–23, SHLVEALYLVCGERG), (47, 48) zinc transporter 8 (ZnT8345–359, LTIQIESAADQDPSC), (51) and glutamic acid decarboxylase 65 (GAD65524–543, SRLSKVAPVIKARMMEYGTT), (52, 53) or unmodified (soluble) peptide cocktail, or PBS. Mice received two s.c. injection/week (20 μg each peptide per injection) for the first week followed by one injection/week for 9 weeks. PBS treated mice rapidly progressed to end-stage T1D (≥ 500 mg/dL) ( Fig. 5A ). No statistically significant difference was observed between soluble peptides and PBS groups ( Fig. 5A ). In contrast, immunization with the lipo-peptide cocktail vaccine in new-onset diabetic NOD mice significantly delayed progression to endstage T1D ( Fig. 5A ). At the end of the treatment (140 d after the start of the treatment), 50% of the mice treated with soluble peptide cocktail were diabetic (> 250 mg/dL), in contrast to 25% in lipo-peptides treated group ( Fig. 5A ). The mean blood glucose levels measured at day 140 were significantly different between PBS and lipo-peptides groups (p = 0.003). In contrast, no statistically significant difference was observed between soluble peptides and PBS treated groups ( Fig. 5B ). In addition, dual immunofluorescence analysis (staining for insulin and glucagon) showed preservation of insulin producing β-cells only in the lipo-vaccine group ( Fig. 5C ).
(A), New-onset diabetic NOD mice were treated with lipo-peptide cocktail (20 μg each peptide/injection), or soluble peptide cocktail, or PBS. Mice received 2 injections for the first week and followed by one injection/week for 9 weeks, and then the mice were followed with no additional treatments. Blood glucose levels were followed over time. (B), Blood glucose measurements at the beginning (day 0) and the end of the treatment (140 d after treatment started). (C), Dual-immunofluorescent staining of frozen pancreatic sections from NOD islets at the end of the study. Tissue sections were stained with antibodies against insulin (red) and glucagon (green), as well as Hoechst dye (blue, nuclear staining). (Scale bar: 200 μm).
Induction of peripheral tolerance by targeting autoreactive T cells with autoantigens requires TCR-pMHC engagement to mediate T-cell deletion, anergy, or regulatory T cells conversion. The level of TCR stimulation by pMHC greatly influences the T cell activation and is governed by many variables including the affinity, avidity, TCR ligand density and duration of TCR-pMHC interaction. (42, 43) T cells are widely believed to be able to sense the subtle differences from these variables and amplify the subsequent signaling, leading to activation and differentiation. For example, antigenic stimulation can promote T cell activation, anergy or differentiation into regulatory T cells, depending on the strength, duration of stimulation and specific TCR-pMHC pair. Reports, however, have been conflicting regarding which variables dominate T cell-induced tolerance, in part because some of these variables are interchangeable toward a cumulative TCR stimulation. It is also likely that multiple biological and kinetic parameters conjointly determine the cumulative outcomes of T cell signaling. Thus, it is reasonable to hypothesize that T cell fates are governed by the cumulative level of TCR stimulation and that minimal stimulation thresholds must be satisfied for both T cell tolerance and activation, although it is thought the latter requires stronger and longer stimulation. Regardless of its mechanisms of action, under subimmunogenic conditions in which costimulation is minimal, tolerogenic APCs need to engage autoreactive T cells through auto-antigen presentation to mediate deletion, anergy, or Treg programming. In NOD mice, insulin β-chain peptide B9–23 is a primary autoantigen triggering spontaneous T1D. (54) Yet, presentation of B9–23 by APCs is weak, since this peptide has poor binding and a weak register on the major histocompatibility complex class II molecule IAg 7 . The majority of insulin-reactive T cells specifically recognize this weak peptide/MHC binding register. (17–20) Both structural and functional studies indicate that diabetogenic T cell clones isolated from NOD mice respond poorly to peptides derived from native antigens. (55) The poor stimulation is in part due to the weak binding between B9–23 and IA g7 . To compensate for poor presentation, high concentrations of peptide are required to observe a detectable T cell response by an insulin-reactive T cell clone. (22) An alternative approach is to use altered peptide ligands. For example, replacing the arginine residue (R22) in the natural B9–23 peptide with glutamic acid resulted in a strong agonist that binds tightly to MHC II. This MHC II stable insulin B9–23 mimetope (R22E) induced much stronger T cell stimulation in vitro and promoted antigen-specific tolerance in vivo. (23) Subsequent immunization with a low dose of B9–23 (R22E) mimetope completely prevented the onset of T1D in NOD mice. (23) Likewise, post-translational modifications of insulin peptide on the C-terminal (neoantigen) stabilized the MHC binding of peptide and protected mice from developing T1D. (32, 33) Although the optimal TCR stimulation remain to be determined, these important studies argue that in the periphery, stable peptide/IAg 7 presentation is needed to drive antigen-specific immune tolerance, and this may explain the poor outcomes of antigen-specific immunotherapy trials so far.
One of the prerequisites for the initiation and maintenance of antigen-specific tolerance is the efficient delivery of disease-relevant antigens to APCs. Targeting antigens to lymph node APCs is a viable approach since a large portion of APCs reside in the lymph nodes, the anatomic sites where immune responses are orchestrated. In clinical trials, autoantigens administered orally or parenterally fail to reach lymph nodes. (16, 56) We hypothesized that efficient delivery of autoantigen targeting APCs in the lymph nodes would substantially improve the antigen presentation. To test this hypothesis, an ‘albumin-hitchhiking’ approach which efficiently delivers peptide antigens to the APCs in the LNs was used. This approach relies on a molecular design which allows covalently modified antigens to bind tightly to endogenous albumin protein and traffic to the draining LNs following s.c. injection. We have previously demonstrated the structural requirements for LN accumulation: a long diacyl lipid (≥ 16 carbons) and a long PEG (≥ 36 ethylene glycol linkers) are needed for an efficient LN targeting. (38) These design rules also alleviated the problem associated with peptide solubility, as both the PEG moiety and albumin-binding increase the water solubility of peptide. In agreement with our prediction, subcutaneous injection of lipo-peptide showed a 10-fold increase in peptide accumulation in DCs in the draining LN, as compared with soluble peptide.
The process of clonal deletion eliminates high affinity T cells and only TCRs with low affinities for self-antigen can escape into the periphery and cause autoimmune diseases. One of the unique features of amphiphilic peptides is that they equilibrate between an albumin-binding state and membrane-anchoring state. (39, 40) Due to the amphiphilic nature of lipo-peptide, the lipid moiety spontaneously inserts into the bilayer structure of the membrane and anchors the peptide to the cell membrane. Indeed, incubation of cells with fluorescein labeled lipo-peptide in the presence of FBS (bovine albumin) resulted in a significant portion of the peptide associated with membrane. The extent of insertion is depended on the concentrations of FBS and density of the cells. The advantage of targeting a drug to a membrane to increase its binding toward membrane-bound receptors has long been recognized. (57, 58) Because MHC II is exclusively expressed on the cellular membrane, and both ends of MHC II are open, the membrane insertion property of lipo-peptide prompted us to investigate whether anchoring the peptide to a membrane can increase its binding to membrane-bound MHC II. Compared to low stability of native peptide, lipo-peptide showed a bivalent interaction on the cell surface: peptide binds to MHC II and lipid anchors on cell surface. One intriguing question is: can MHC II present lipid modified peptides to T cells without the need for enzymatic cleavage and processing? Peptide antigens can be loaded on MHC II by two distinct pathways: antigens are either endocytosed, digested in lysosomes, and the resulting peptide fragments are loaded onto MHC II prior to their migration to the cell surface, or peptides can be loaded directly on the empty cell-surface class II MHC proteins without intracellular processing. (59) Chemically inactivated DCs loaded with lipo-OVA323–339 potently activated OT-II T cells, suggesting intact lipo-peptide can indeed be presented without antigen processing. This observation suggests lipo-peptide can bypass the class II-like protein DM editing which removes peptides that form unstable complexes or have bound with a weak register. (60) Lipo-OVA323–339 loaded DCs efficiently activated OT-II T cells and lowered the EC50 value by approximately 3 fold, compared with unmodified OVA323–339. Importantly, lipid modified OVA323–339 peptide greatly prolonged the peptide presentation by DCs. During DC-T cell interaction, peptide degradation, rapid MHC turnover, and dissociation of peptide from MHC molecules over time reduce peptide/MHC density and duration, dampening the DC-T interaction ( Fig. 2C ). However, due to the bi-valent interaction of lipo-peptide on DC surface as well as the elevated amount of lipo-peptide inserted on the DCs surface acting as antigen reservoir, peptide presentation can be enhanced.
These studies demonstrate that lipid modification simultaneously altered the quantity and quality of peptide antigen presentation. In vivo, injection of small doses of lipo-peptides results in induction of antigen-specific immunological tolerance that controls both T-cell- and B-cell-mediated immunity, and significant protection from autoimmune T1D development. Importantly, immunization of lipid modified peptide cocktail, but not the soluble peptide in the same dose, delays progression of acute T1D. This therapeutic effect has not previously been achieved with antigen-specific immunotherapy, irrespective of modification (mimotope) and mode of application. (23) Agonistic altered peptide ligands (APLs) with modification on MHC anchor residues have been extensively used to improve the presentation of peptides. (61) However, rational design of APLs has proven to be difficult, as the relationship between APL structures and therapeutic benefits is difficult to predict. (62) Substituting the peptide anchor residues can sometimes completely abrogate the recognition by T cells. (62) Thus, our approach provides a new method to improve peptide/MHC presentation without mutating the peptide sequences. Although no long-term reversal of T1D was observed once the treatment of lipo-peptides was stopped ( Fig. 5A ), it is possible that the therapeutic benefits in this study can be amplified by optimizing key variables such as dose, kinetics, or by combining other immunomodulating modalities such as tolerizing adjuvants.
In conclusion, the data presented here demonstrate a novel molecular approach to enhance CD4 + peptide antigen presentation. Lipid functionalization targets peptide antigen to DCs in the LN by binding and trafficking with endogenous albumin after subcutaneous injection. More importantly, lipid functionalization markedly enhances peptide presentation by a unique bivalent interaction with MHC II molecules on cell surface. This approach will be applicable not only for T1D but for other T cell mediated diseases wherein efficient antigen presentation is needed.
