Induction of antigen-specific tolerance by nanobody–antigen adducts that target class-II major histocompatibility complexes

The association of autoimmune diseases with particular allellic products of the class-II major histocompatibility complex (MHCII) region implicates the presentation of the offending self-antigens to T cells. Because antigen-presenting cells are tolerogenic when they encounter an antigen under non-inflammatory conditions, the manipulation of antigen presentation may induce antigen-specific tolerance. Here, we show that, in mouse models of experimental autoimmune encephalomyelitis, type 1 diabetes and rheumatoid arthritis, the systemic administration of a single dose of nanobodies that recognize MHCII molecules and conjugated to the relevant self-antigen under non-inflammatory conditions confers long-lasting protection against these diseases. Moreover, co-administration of a nanobody–antigen adduct and the glucocorticoid dexamethasone, conjugated to the nanobody via a cleavable linker, halted the progression of established experimental autoimmune encephalomyelitis in symptomatic mice and alleviated their symptoms. This approach may represent a means of treating autoimmune conditions. Nanobodies recognizing class-II major-histocompatibility-complex molecules and conjugated to relevant self-antigens confer protection against autoimmune diseases in mice.

A pproximately 10% of the human population suffer from an autoimmune condition, with symptoms that range from mild to life threatening 1 . Current treatments for autoimmune diseases include general immunosuppression, which blunts responses across the entire spectrum of antigens. Various pre-clinical models of autoimmunity involve the administration of a defined antigen under the appropriate stimulatory conditions to elicit pathology 2 . Engagement of antigen-presenting cells (APCs) under inflammatory conditions (for example, in the presence of adjuvants) elicits a strong response against foreign antigens 3,4 . In contrast, APCs that acquire antigen under non-inflammatory conditions fail to upregulate co-stimulatory signals and induce tolerance 5,6 . To target APCs under tolerogenic conditions, we developed and characterized alpaca-derived single-domain antibody fragments (nanobodies/VHHs) that recognize major histocompatibility complex class II (MHCII) molecules (VHH MHCII ) 7,8 . These nanobodies lack effector functions and target all MHCII-positive cells, which include APCs. Their small size ensures efficient tissue penetration and rapid clearance from the circulation of those nanobodies that fail to find their target. This makes nanobodies ideal vehicles for targeted delivery of payloads of interest, such as antigenic peptides or small-molecule drugs 9,10 . We have further established an engineering strategy for nanobodies that enables their site-specific modification at the carboxy (C) terminus with the aforementioned payloads 11 .
The distribution of a diverse set of APCs over different anatomical sites and their cellular dynamics complicate the identification of the relevant tolerogenic APC in vivo. The possible transfer of materials between various sets of APCs is an additional confounding factor. In fact, several distinct types of APCs or their products could all contribute and act in synergy to induce tolerance 12 . Here, we demonstrate the efficacy as tolerogens of VHHs that target the MHCII-positive cell population. This approach was effective in the prevention and treatment of experimental autoimmune encephalomyelitis (EAE), both in an accelerated model of type I diabetes in the mouse and in a T cell-mediated arthritis model. To extend this approach to disease interception, we delivered dexamethasone (DEX), an immunosuppressive small molecule, to MHCII-positive cells via VHH MHCII conjugated to DEX through a cleavable linker. We found that VHH MHCII -peptide adducts in combination with VHH MHCII -DEX are an effective treatment of animals symptomatic for EAE: it halts disease progression in animals with overt signs of disease.

A single dose of VHH MHCII -MOG 35-55 provides durable protection against induction of EAE.
We engineered an alpaca-derived single-domain antibody that recognizes a wide range of mouse MHCII molecules, including I-A b and I-A d (VHH MHCII ), with a sortase recognition motif-LPETGG-to allow its site-specific ligation (Fig. 1a) to antigenic peptides and small molecules modified with one or more suitably exposed glycine residues. Purified VHH adducts were characterized by liquid chromatography-mass spectrometry (LC-MS) ( Fig. 1b and Supplementary Fig. 1) to verify their identity.
Immunization of C57BL/6 mice with myelin oligodendrocyte glycoprotein peptide 35-55 (MOG  ) in the presence of complete Freund's adjuvant (CFA) and pertussis toxin (PTX) elicits EAE, a multiple sclerosis-like condition 13 . We hypothesized that previous administration of MOG  , delivered to MHCII + APCs under non-inflammatory conditions, might interfere with the induction of EAE. We administered three 20-μg doses of the VHH MHCII -MOG  adduct intravenously 7 d before induction of disease. This treatment suppressed induction of EAE. Mice that received the identical amount of VHH MHCII conjugated to an irrelevant peptide, ovalbumin peptide 323-339 (VHH MHCII -OVA 323-339 ) or MOG  peptide linked to a VHH of irrelevant specificity (an anti-green fluorescent protein/GFP VHH, that is, VHH GFP ) invariably developed EAE ( Fig. 1c and Supplementary Fig. 2). Even a single or two injection(s) of 20 μg VHH MHCII -MOG  still achieved full protection (Fig. 1d,e and Supplementary Fig. 2); a single dose was used in all subsequent experiments. Flow cytometry of CD4 + lymphocyte infiltrates recovered from the spinal cord of diseased mice at days 15-18 after immunization and of protected mice at day 30 after administration of the MOG 35-55 /CFA/PTX cocktail was consistent with the observed disease scores: diseased mice showed a pronounced influx of interleukin-17 (IL-17) and interferon-γ (IFN-γ)-producing CD4 + T cells, as well as some Foxp3 + CD4 + regulatory T cells ( Fig. 1f and Supplementary Fig. 2). Haematoxylin and eosin and Luxol fast blue staining of spinal cord sections from mice that received VHH MHCII -MOG  before attempts to induce EAE showed preservation of myelination and less immune cell infiltration (Fig. 1g,h).
To explore the durability of protection afforded by VHH MHCII -MOG  , we injected 20 μg VHH MHCII -MOG 35-55 1 or 2 months before administration of the MOG 35-55 /CFA/PTX cocktail. Even then, we observed delayed onset, if not complete suppression of EAE ( Fig. 1i and Supplementary Fig. 3). Despite the short circulatory half-life of free VHH MHCII -MOG  , which is estimated to be <0.5 h, VHH MHCII -MOG 35-55 confers prolonged protection. Five weeks after the VHH MHCII -MOG  injection, which established protection from a first exposure to MOG 35-55 /CFA/PTX, we re-challenged mice with a second administration of MOG 35-55 /incomplete Freund's adjuvant in the presence of PTX. Despite this second challenge, mice, once protected, showed no signs of developing EAE ( Fig. 1j and Supplementary Fig. 4). The tolerance evoked by a single dose of VHH MHCII -MOG  , even weeks after its administration, thus provides lasting protection to even a second challenge.
Splenic CD11c + dendritic cells are the APCs associated with induction of antigen-specific tolerance. To explore the possible mechanisms of VHH MHCII -mediated induction of tolerance, we generated VHH MHCII -Cy5 ( Supplementary Fig. 5) to follow the biodistribution of VHH MHCII -Cy5 injected intravenously into MHCII-GFP mice. These mice carry a targeted gene replacement that encodes an I-A b -GFP fusion. It replaces the endogenous I-A b locus and ensures that all MHCII + cells express GFP 14 . At 1.5 h after injection, VHH MHCII -Cy5 was captured by a splenic and circulatory MHCII-GFP + cell population (Fig. 2a). The fluorescent VHH MHCII adducts were captured by B cells and dendritic cell subsets, including splenic CD8a + dendritic cells, CD4 − conventional dendritic cells and CD4 + conventional dendritic cells, but not plasmacytoid dendritic cells (Supplementary Fig. 6).
Intravenous, but not subcutaneous or intraperitoneal, injection of VHH MHCII -MOG  protected against induction of EAE ( Supplementary Fig. 7). This hinted at a role of the spleen or bloodstream as a site where tolerance induction is initiated. One week after injection of 20 μg VHH MHCII -MOG 35-55 intravenously ( Fig. 2b and Supplementary Fig. 8), we harvested splenocytes and whole blood as a source of donor cells. Mice then received 20 million unfractionated splenocytes or peripheral blood mononuclear cells (PBMCs) from the VHH MHCII -MOG 35-55 -treated animals. One day after cell transfer, we administered MOG  in CFA + PTX to induce EAE. We saw a substantial reduction in the mean clinical EAE score in mice that received splenocytes from mice treated with VHH MHCII -MOG   (Fig. 2b and Supplementary Fig. 8). We eliminated macrophages and CD8 T cells in vivo by administering the corresponding depleting antibodies: anti-colony stimulating factor 1 receptor (anti-CSF1R) antibodies and anti-CD8a antibodies, respectively ( Fig. 2c and Supplementary Fig. 9) 15 . To deplete dendritic cells, we administered diphtheria toxin (DTX) in CD11c-DTR (DTX receptor) mice ( Fig. 2c and Supplementary Fig. 9) 16 . To test the possible involvement of B cells, we administered VHH MHCII -MOG  into μMt −/− mice, which lack mature B cells 17 . Only elimination of CD11c + dendritic cells reduced the measure of protection provided by VHH MHCII -MOG   (Fig. 2c and Supplementary Fig. 9). We created two VHH-MOG  adducts that presumably target different but overlapping subsets of myeloid cells: we used a VHH directed against CD11b (mostly present on macrophages) and a VHH that recognizes CD11c (mostly present on dendritic cells) (Supplementary Fig. 1) 7 . Only the VHH CD11c -MOG 35-55 combination provided an intermediate level of protection against induction of EAE ( Fig. 2d and Supplementary Fig. 10), consistent with the results from elimination of CD11c + cells. Batf3 −/− mice treated with VHH MHCII -MOG 35-55 remained resistant to induction of EAE 18 . In this setting, CD8a + dendritic cells therefore do not contribute to the set of tolerogenic APCs ( Supplementary Fig. 11).
To determine whether delivery by VHH MHCII of more than just the minimal epitope can likewise induce tolerance, we generated VHH MHCII -MOG 17-78 and treated mice 7 d before challenge (Fig. 2e). VHH MHCII -MOG 17-78 likewise protected against induction of EAE (Fig. 2f). This shows that our approach can tackle a condition where the offensive protein is known but not its minimal epitope(s).

Administration of VHH MHCII -MOG 35-55 elicits a burst of proliferation, followed by attrition, of MOG 35-55 -specific CD4 T cells.
To investigate the impact of VHH MHCII -MOG 35-55 adducts on T cells provides lasting protection against eAe. a, Schematic for nanobody C-terminal sortase labelling with GGG-carrying antigenic peptides. SrtA, Sortase A. b, LC-MS of purified VHH MHCII and VHH MHCII -antigen adducts.The number in the top right corner of each plot represents the molecular weight of the corresponding VHH or VHH-antigen adduct as indicated. c-e, Mean disease scores of mice that received VHH-peptide prophylactic treatment at three doses (c), two doses (d) and one dose (e), as indicated. Disease scores: 1 = limp tail; 2 = partial hind leg paralysis; 3 = complete hind leg paralysis; 4 = complete hind and partial front leg paralysis; and 5 = moribund. The data represent means ± s.e.m. of biological replicates (for VHH MHCII -OVA 323-339 , n = 7 (c) and n = 3 (d and e); for VHH GFP -MOG  , n = 8 (c) and n = 5 (d and e); for VHH MHCII -MOG 35-55 , n = 8 (c), n = 5 (d) and n = 9 (e)). ***P < 0.001 (two-way analysis of variance (ANOVA) with repeated measures). f, Flow cytometry of T H 1 and T H 17 CD4 + lymphocytes in the spinal cord, collected at the end point for mice that received one dose of VHH-antigen (bottom). The absolute number of infiltrating CD4 + T cells (top left) and the frequency of FoxP3 + CD4 + regulatory T cells (top right) are also indicated. The data represent means ± s.e.m. of biological replicates (n = 3-5). NS, not significant. ***P < 0.001 (unpaired t-test with Holm-Šídák adjustment). g,h, Representative haematoxylin and eosin (g) and Luxol fast blue staining (h) of spinal cord sections from mice having received a single dose of VHH-antigen adduct. Scale bars, 100 μm and 20 μm (inset). Two images per mouse were taken. The right-hand column in each panel shows magnified views of the images to the left. i, Mean disease scores of mice that received VHH-peptide prophylactic treatment at −60, −30 and −7 d before induction of EAE. The data represent means ± s.e.m. of biological replicates (n = 3-5, as indicated in the legend). ***P < 0.001 (two-way ANOVA with repeated measures). j, Mean clinical scores of VHH MHCII -MOG  recipients subjected to multiple challenges with MOG/CFA/PTX and MOG/incomplete Freund's adjuvant (IFA)/PTX. The data represent means ± s.e.m. of biological replicates (n = 5).
of defined antigen specificity, we used 2D2 T cell receptor (TCR) transgenic mice as a source of monoclonal CD4 + T cells that recognize the I-A b -MOG 35-55 complex 19 . We transferred congenically marked, CellTrace Violet-labelled 2D2 CD45.2 + CD4 + T cells into CD45.1 recipients, followed by intravenous injection of VHH MHCIIpeptide adducts 1 d later. We tracked the number of 2D2 cells in the    . The data represent means ± s.e.m. of biological replicates (n = 4-6, as indicated in the legend). c, Mean clinical scores of mice that received prophylactic treatment with VHH MHCII -OVA 323-339 or VHH MHCII -MOG  following depletion of the indicated cell subset before induction of EAE. The data represent means ± s.e.m. of biological replicates (n = 3-6, as indicated in the legend). d, Mean disease scores of mice that received the indicated VHH-antigen. The data represent means ± s.e.m. of biological replicates (n = 3-6, as indicated in the legend). e, LC-MS of purified VHH MHCII -MOG  . The number in the top right corner of the plot represents the molecular weight of VHH MHCII -MOG  . f, Mean disease scores of mice that received VHH-peptide prophylactic treatment. The data represent means ± s.e.m. of biological replicates (n = 3-5, as indicated in the legend). In b, c, d and f, ***P < 0.001 (two-way ANOVA with repeated measures). spleen, inguinal lymph nodes and blood for 10 d. In mice receiving VHH MHCII -MOG  , 2D2 CD4 + T cells underwent an initial burst of expansion, followed by contraction on day 5 post-injection, as judged from the absolute number of 2D2 cells recovered from the spleen, inguinal lymph nodes and blood, as well as whole-body imaging using non-invasive positron emission tomography for CD4 + cells ( Fig. 3a and Supplementary Fig. 12). Such disappearance occurred after several divisions: all of the recovered 2D2 CD4 + T cells were antigen-experienced and had divided, as was evident from CellTrace Violet dilution (Fig. 3b). Delivery of an amount of MOG  equimolar to that of the administered VHH MHCII -MOG  adduct led to division of no more than ~5% of the 2D2 T cells. VHH MHCII -mediated antigen delivery thus clearly enhances its presentation (Fig. 3b). At the protein level, these 2D2 T cells also showed higher levels of apoptotic and exhaustion markers, such as programmed cell death protein 1 (PD-1) and LAG3, but not Tim-3, Fas/CD95 or LAP (latency-associated peptide) ( Fig. 3e and Supplementary Fig. 14) 20 . At day 3 post-injection, 2D2 CD4 T cells in VHH MHCII -MOG  recipients failed to downregulate CD62L, while remaining CD44 + ( Supplementary Fig. 14). When we treated LAG3 −/− mice with a single dose of VHH MHCII -MOG  and then attempted to induce EAE, protection was lost, albeit with notable delay, whereas PD-1 −/− mice were still tolerized by VHH MHCII -MOG   (Fig. 3f). Deletion of LAG3 in 2D2 TCR transgenic mice has been shown to cause spontaneous EAE 21 . This indicates the importance of the LAG3 pathway in this tolerance induction strategy.

MOG-specific 2D2 CD4 T cells upregulate co-inhibitory receptors upon administration of VHH
Because both activated effector T cells and regulatory T cells (T reg cells) express LAG3, we then evaluated whether an increase in T reg cells contributes to VHH MHCII -MOG 35-55 -imposed tolerance 22 . To uncover a role for T reg cells in VHH MHCII -MOG 35-55 -mediated tolerance, we eliminated T reg cells in Foxp3-DTR mice by the administration of DTX ( Supplementary Fig. 15) 23 . Treated mice lost T reg cells and were no longer protected against EAE, demonstrating their contribution to VHH MHCII -MOG 35-55 -imposed tolerance ( Supplementary  Fig. 15). Administration of VHH MHCII -MOG 35-55 increases the number of FoxP3 + MOG 35-55 -specific T reg cells ( Supplementary Fig. 15).
Finally, we challenged mice that had received 2D2 T cells with MOG 35-55 /CFA at day 10. The 2D2 T cells in mice that received VHH MHCII -MOG  failed to respond, whereas 2D2 T cells in mice injected with VHH MHCII -OVA 323-339 proliferated robustly (Fig. 3g). This underscores the antigen specificity of tolerance induction by VHH MHCII -MOG  .
VHH MHCII -antigen adducts act in an antigen-specific manner in other models of autoimmunity. Next, we explored other examples of autoimmunity. For type 1 diabetes, we used the aggressive BDC2.5 T cell adoptive transfer model, which mimics the destruction of β cells by autoreactive T cells. TCR transgenic CD4 T cells that carry the BDC2.5 T cell receptor recognize pancreatic β cells and can be activated ex vivo with a ten-residue peptide, the mimotope p31. In non-obese diabetic/severe-combined immunodeficient (NOD/SCID) mice, such activated BDC2.5 T cells cause hyperglycaemia within 8 d after transfer 24 . We conjugated p31 to VHH MHCII ( Supplementary Fig. 1). NOD/SCID mice that received activated BDC2.5 splenocytes were treated 1 d later with either saline, 20 μg VHH MHCII -MOG  or 20 μg VHH MHCII -p31 ( Fig. 4a). Only mice treated with VHH MHCII -p31 remained normoglycaemic ( Fig. 4a and Supplementary Fig. 16). VHH MHCII -p31-treated mice had fewer BDC2.5 CD4 T cells in their pancreas and secondary lymphoid organs ( Supplementary Fig. 16). Islets in protected mice remained intact (Fig. 4b). We saw a mild protective effect even when we administered VHH MHCII -p31 to mice relatively late, on day 5 post-transfer of the activated BDC2.5 T cells ( Supplementary Fig. 16).
Arthritis can be induced in BALB/c recipients by intravenous transfer of ex vivo activated DO11.10 T cells that recognize OVA 323-339 , followed 1 d later by re-stimulation in vivo with a footpad injection of OVA/CFA emulsion and a challenge 10 d later by injection of heat-aggregated OVA (Fig. 4c) 25 . Mice were monitored for the development of arthritis by measuring paw thickness and by histological assessment at day 7 following a challenge with heat-aggregated OVA. Previous administration of VHH MHCII -OVA 323-339 reduced joint inflammation upon exposure to OVA, whereas VHH MHCII -MOG  had no effect ( Fig. 4c and Supplementary Fig. 17). Mice treated with VHH MHCII -OVA 323-339 showed fewer signs of cartilage destruction (Fig. 4d). Immune cells obtained from popliteal lymph nodes of mice treated with VHH MHCII -OVA 323-339, when stimulated ex vivo with OVA, failed to produce IFN-γ ( Supplementary Fig. 17). Perhaps not unexpectedly, serum from mice treated with VHH MHCII -OVA 323-339 also had lower levels of anti-OVA and anti-OVA 323-339 immunoglobulin G1 (IgG1) antibodies ( Supplementary Fig. 17). Combined, these results confirm the ability of VHH MHCII -antigen adducts to reduce the harm inflicted by activated, autoreactive CD4 Violet-labelled CD45.2 2D2 CD4 T cells 1 d before infusion of VHH-antigen. We followed the numbers of 2D2 CD4 T cells in the spleen, blood and inguinal lymph nodes by flow cytometry. The data represent means ± s.e.m. of biological replicates (n = 4 or 5 per group). ***P < 0.001 (unpaired t-test with Holm-Šídák adjustment). b, Violet trace dilution indicating proliferation of 2D2 T cells at day 3 (n = 3 per group). c, In a separate experiment, on day 3 post-infusion, spleens were collected and CD45.2 + CD4 + TCRa3.2 + TCRb11 + cells were sorted according to the number of divisions they underwent and then processed for transcriptomic analyses by RNA-seq. Volcano plots of RNA-seq data compare the 2D2 CD4 T cells in mice that received VHH MHCII -MOG 35-55 after three divisions with 2D2 CD4 T cells recovered from mice that received VHH MHCII -OVA 323-339 . The vertical dotted lines indicate a log 2 [fold change] of 2 and −2 (n = 2 per group). d, Heat map showing the expression of co-inhibitory receptors on 2D2 CD4 + T cells. TPM, transcript count per million. e, CellTrace Violet dilution reflecting proliferation of 2D2 T cells at day 3. VHH MHCII -MOG 35-55 administration led to a distinct pattern of phenotypic markers on 2D2 CD4 T cells. Representative flow images gated on: CD45.2 + CD4 + TCRβ11 + are shown. The mean fluorescence intensity (MFI) of each marker is plotted as the mean ± s.e.m. of biological replicates (n = 4 or 5 per group). ***P < 0.001 (unpaired t-test with Holm-Šídák adjustment). f, Mean disease scores of mice that received prophylactic treatment with VHH MHCII -OVA 323-339 or VHH MHCII -MOG  for the indicated genetic background. The data represent means ± s.e.m. of biological replicates (n = 3 or 4 per group, as indicated in the legend). ***P < 0.001 (two-way ANOVA with repeated measures). g, We challenged CD45.1 mice that received CD45.2 2D2 CD4 T cells and an infusion of VHH-antigen with MOG 35-55 emulsified in CFA on day 10. Spleens, blood and inguinal lymph nodes were collected 5 d later. 2D2 T cells in mice that had received VHH MHCII -MOG  failed to respond, unlike 2D2 T cells in mice injected with VHH MHCII -OVA 323-339 . The data represent means ± s.e.m. of biological replicates (n = 5 per group). ***P < 0.001 (unpaired t-test with Holm-Šídák adjustment).
T cells. The underlying mechanism(s) must be conserved across mouse MHC haplotypes.

VHH MHCII -antigen adducts also suppress CD8 T and B cell responses.
To determine whether CD8 T cell responses are affected by the administration of VHH MHCII -antigen adducts , we attached the OVA-derived CD8 T cell epitope SIINFEKL (this peptide, named the OTI peptide, is restricted by H-2K b ) to VHH MHCII (Supplementary Fig. 1) 26 . Mice received congenically marked OTI T cells, followed by an injection of VHH MHCII -OTI or VHH MHCII -ORF8 (with or without adjuvant) 1 d later (Fig. 4e). The open reading frame 8 (ORF8) epitope derived from murine cytomegalovirus is recognized by CD8 T cells in H-2 b mice and served as a control 27 . A re-challenge of the recipients with OVA/CFA at day 10 post-transfer failed to activate any remaining OTI T cells (Fig. 4e,f) legend). b, Representative haematoxylin and eosin staining of pancreas sections from mice that had received a single dose of VHH-antigen. Scale bars, 100 μm. Two images per mouse were taken. c, Mean paw thickness of BALB/c mice treated with VHH-antigen to assess the progression of rheumatoid arthritis. The data represent means ± s.e.m. of biological replicates (n = 6 per group). HAO, heat-aggregated ovalbumin. d, Representative toluidine blue staining of joint sections from mice that had received a single dose of VHH-antigen. Scale bars, 100 μm. Two images per mouse were taken. e, Mice (CD45.1 + ) received allotypically marked CD45.2 + CD8 + OTI T cells 1 d before injection of VHH MHCII -ORF8 604-612 or VHH MHCII -OVA 257-264 (OTI peptide). We challenged these mice with OTI peptide emulsified in CFA on day 10. Spleens, inguinal lymph nodes and blood were collected 5 d later and analysed by flow cytometry. The data represent means ± s.e.m. of biological replicates (n = 6 per group). f, Splenocytes were cultured for 3 d in complete RPMI supplemented with OT1 peptide. Supernatant was collected to measure the production of IFN-γ by ELISA. The data represent means ± s.e.m. of biological replicates (n = 3 per group). g,h, Antibodies against OB1 peptide (g) and OVA protein (h) were measured by ELISA in sera collected from C57BL/6J recipients that received three consecutive injections of saline, VHH MHCII -OB1 or equimolar amounts of free OVA. The data represent means ± s.e.m. of biological replicates (n = 3 or 4 per group). ***P < 0.001 (unpaired t-test with Holm-Šídák adjustment). Fig. 1) 28 . Three consecutive injections of VHH MHCII -OB1 into C57BL/6J recipients failed to elicit IgG antibody responses against either intact OVA protein or the OB1 peptide (Fig. 4g,h), whereas mice that received equimolar amounts of free OVA protein readily produced such antibodies.

Co-delivery of VHH MHCII -MOG 35-55 and VHH MHCII -DEX improves therapeutic efficacy.
We then explored the impact of VHH MHCII -MOG  administration to mice that showed active signs of EAE. Injection of VHH MHCII -MOG 35-55 into mice that had developed an EAE score of 1 (limp tail) halted the progression of EAE in nine out of 16 mice (Fig. 5a and Supplementary Fig. 18). For the remaining seven out of 16 mice, their overall condition rapidly deteriorated (shivering and reduced motor activity) after injection of VHH MHCII -MOG  , seemingly unrelated to EAE. A cytokine storm elicited by the targeted delivery of antigen into an already inflamed environment was responsible (Fig. 5c). The polyclonal nature of the evoked T cell response and the rather superficial clinical scoring system imply heterogeneity in the diseased cohort, which may explain why not all animals that received VHH MHCII -MOG  responded similarly. We wondered whether it might be possible to co-deliver an immunosuppressive drug to avert a cytokine storm. We delivered the immunosuppressive corticosteroid DEX-attached via a self-immolating hydrazone linker to VHH MHCII -to MHCII + cells (VHH MHCII -DEX; Fig. 5b and Supplementary Fig. 19) 29 .
We treated a cohort of symptomatic animals with different EAE disease scores of 1 (limp tail), 2 (partial hind leg paralysis) or 3 (complete hind leg paralysis) (Fig. 5d-f and Supplementary Fig. 20).
We administered a single 20-μg dose of VHH MHCII -OVA 323-339 or VHH MHCII -MOG  in combination with 20 μg VHH MHCII -DEX on the day they reached the indicated EAE score. Mice treated at stage 3 with VHH MHCII -OVA 323-339 rapidly became paralysed (score 4), which then required euthanasia (Fig. 5d-f). A single combined dose of VHH MHCII -MOG  and VHH MHCII -DEX abolished lethality in the subset of animals with signs of cytokine release syndrome. Mice treated at an EAE score of 1 all survived, with no signs of disease progression (Fig. 5c). Animals that had progressed to an EAE score of 2 or 3 improved in disease score, with reversal of symptoms. These mice no longer dragged their legs and were actively walking without any other obvious impediments (Fig. 5e,f). Improvements in disease scores were mirrored by a reduction in infiltrating CD4 T cells in the spinal cord ( Supplementary Fig. 20). The observed benefit required no more than the equivalent of 0.5 μg DEX in the form of the VHH MHCII -DEX adduct. The ability to specifically target anti-inflammatory drugs to the very same sets of APCs implicated in triggering cytokine release allowed us to use a dose of DEX some 200-fold lower than what would be required systemically (Fig. 5d-f and Supplementary Fig. 21).

discussion
Various modes of antigen delivery can induce antigen-specific tolerance in pre-clinical models of autoimmune disease, with some promise in early-stage clinical trials 30 . Drastic immune resetting by myeloablation, followed by autologous haematopoietic stem cell transplantation, has produced promising results in severely ill patients with myasthenia gravis and multiple sclerosis 31 . Other cell-based therapies include the transfusion of modified immune cells, including dendritic cells and engineered erythroid cells [32][33][34][35] . Tolerogenic nanoparticles have also been explored as a means of intervention in autoimmunity 36,37 . In addition to curbing inflammation, wholesale immunosuppression is the backstop in the treatment of autoimmunity-a therapy that can increase the risk of infectious disease. While antibiotic treatment can mitigate this drawback at least in part, the search for a more targeted approach to blunt undesirable immune reactions remains a priority. The induction of antigen-specific tolerance is a particularly high bar to clear if one considers the presence of pathology and pre-existing autoimmunity at diagnosis. Autoimmune destruction of target cells predates the onset of symptoms that bring the patient to medical attention. Therapy must therefore deal not only with pre-existing autoimmunity but also the possibility of epitope spreading beyond the initiating insult. Any type of prophylactic treatment will be of limited use unless susceptible populations can be identified unambiguously, and then only if the risk of eliciting unwanted side effects of the proposed treatment is acceptably small.
The recent and rather narrow focus on dendritic cells as a key component of cell-based interventions in immunity has overshadowed earlier work in which antigens were targeted to MHCII products, expressed on all APCs 5,38,39 . This was done through the creation of full-size anti-MHCII monoclonal antibodies conjugated to antigens to elicit an immune response. For this reason, we chose to deliver nanobody-autoantigen fusions under non-inflammatory conditions to MHCII + cells-a strategy that does not obviously differentiate among the various APC subsets, but is efficacious nonetheless. More importantly, our anti-mouse VHH MHCII does not distinguish between MHCII allotypes. We showed that it is not essential to deliver the minimal CD4 T cell epitope, but that larger fragments can also lead to tolerogenic antigen processing and presentation (Fig. 2e,f). Depending on the size of the autoantigen, VHH MHCIIantigen adducts can eliminate the need for precise epitope identification in a disease or protein replacement setting. This highlights advantages of the approach reported here, which was efficacious in the prevention of autoimmunity in three mouse models, compared with those that employ defined allotypes of MHCII molecules complexed with relevant antigenic epitopes 40 . Ideally, interventions ought to be antigen-specific and as simple as possible, both from a manufacturing and application perspective. The VHH MHCII adducts described here meet this criterion. We hypothesize that antigen delivery to MHCII + cells covers the relevant tolerogenic APCs and obviates the need for identifying disease-or organ-specific APCs. Sortase-catalysed modifications allow easy access to many structurally distinct VHH adducts without the need for separate expression vectors for each combination. Separating the production of the VHHs from that of the payloads to be attached enables the inclusion of small molecules and post-translationally modified sequences that are not accessible by standard recombinant expression methods. These adducts may then be used to explore interception in a wide range of immune conditions. The anti-DEC205 monoclonal antibody, genetically fused to MOG  , provided prophylactic protection in the EAE model 41 but as a full-sized immunoglobulin is expected to have a much longer circulatory half-life and worse tissue penetration than a similarly modified VHH.
In conclusion, a MOG 35-55 -modified VHH that recognizes MHCII products can protect mice against the induction of EAE. A single injection of 20 μg of the VHH MHCII -antigen adduct affords protection that lasts for at least 2 months. Administration of the same VHH MHCII -MOG  adduct in animals that already show symptoms of EAE (score 1, 2 or 3) halts progression and even allows partial reversal of the severity of the symptoms. Only a subset of animals symptomatic for EAE responded to treatment with VHH MHCII -MOG 35-55 without undesirable side effects, whereas the remainder showed rapid exacerbation attributable to a cytokine storm. An inflammatory environment must already exist in symptomatic animals, such that delivery of the VHH MHCII -MOG  adduct to APCs only adds fuel to the fire. Indeed, administration of nanobody-peptide adducts in the presence of anti-CD40 and polyinosinic:polycytidylic acid/poly (I:C) as adjuvants strongly potentiates antibody responses against them 7 . To overcome this acute response, we co-delivered a VHH MHCII -DEX adduct, which rescued survival. In a setting where there is a chronic inflammatory response, administration of the type of nanobody adducts developed here would be possible only if appropriate supporting measures were available, as in the case of the VHH MHCII -DEX adduct.
We have yet to determine the minimally effective dose and timing of co-administration of the VHH MHCII -DEX adduct-a parameter that might inform possible translation to a clinical application.
Epitope spreading can occur in many autoimmune diseases 42 . There are no practical hurdles to the inclusion of multiple antigens, prepared as separate fusions to VHH MHCII . For example, in the case of multiple sclerosis, the set of dominant self-antigens (that is, myelin basic protein, myelin oligodendrocyte glycoprotein and myelin proteolipid protein) is limited. We hypothesize that in remitting/relapsing multiple sclerosis, a subset of these antigens delivered to APCs under non-inflammatory conditions may suffice to reduce disease severity or halt progression, when treated with this combination approach. Similarly, in the NOD mouse model of type 1 diabetes, one might consider the administration of a cocktail of nanobodyantigen adducts: VHH MHCII conjugated to (pro)insulin, glutamic acid decarboxylase, islet antigen 2 and Islet amyloid polypeptide 43 .
The pharmacokinetic properties of nanobodies make them particularly attractive for the construction of antibody-drug conjugates 44,45 . Full-sized immunoglobulin-based antibody-drug conjugates continue to circulate for periods of up to weeks and continue to release payloads directly into the bloodstream upon hydrolysis of the linkers via which the drugs are attached. This can result in unwanted systemic drug exposure. In contrast, the VHH MHCII -DEX adduct has the desired properties of excellent targeting, as verified by non-invasive imaging, a short circulatory half-life and ease of modification 46 . The cellular targets recognized by VHH MHCII include all MHCII + cells. Even if the types of APC responsible for induction of tolerance and for provoking a cytokine storm are several and distinct, the MHCII-based targeting approach would cover them. Nanobody-drug adducts have yet to find the broad range of applications of their full-sized counterparts, but our data show that this is an opportunity not to be discounted and point to the immense potential of our method. The amino acid sequence of the sortaggable VHH GFP was Q VQ LQESGGALVQ PGGSLRLSCAASGFPVNRYSMRWYRQAPGKEREWVAGMSSAGDRSSYEDSVK GRFTISRDDARNTVYLQMNSLKPEDTAVYYCNVNVGFEYWGQGTQVTVSSL PETGGHHHHHH.

Expression of VHHs and endotoxin removal. WK6
Chemical synthesis of the GGG antigens GGG-Cy5 and GGG-DEX. The peptides were synthesized on 2-chlorotrityl resin (Chem-Impex) following the standard solid-phase peptide synthesis protocol or ordered from GenScript. For GGG-Cy5, GGGC (7.0 mg; 24 µmol) was dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich; 400 µl) and added to Cyanine5 maleimide (Lumiprobe; 5.0 mg; 7.8 µmol). The resulting mixture was gently agitated at room temperature until LC-MS analysis showed no remaining starting material. The ligated product was then purified by reversed-phase high-performance liquid chromatography

C-terminal sortagging of VHH-LPETGG or GFP-LPETGG with GGG-carrying moieties.
Reactions were carried out in 1 ml containing Tris-HCl (50 mM; pH 7.5), CaCl 2 (10 mM), NaCl (150 mM), 1 mg LPETGG-containing proteins, GGG-containing probe (100 µM) and 5M-Sortase A (5 µM). After incubation at 4 °C with agitation for 1.5 h, unreacted VHH and 5M-Sortase A were removed by adsorption onto Ni-NTA agarose beads. The unbound fraction was concentrated and excess nucleophile was removed using an Amicon 3,000 KDa MWCO filtration unit (Millipore). The average yield of sortase-mediated ligation was at least 60% of the input LPETGG-containing proteins. Reaction products were analysed by LC-MS to assess the purity and stored frozen at −80 °C. Flow cytometry analyses. Cells were harvested from the spleen, lymph nodes or other organs and were dispersed into RPMI 1640 through a 40-µm cell strainer using the back of a 1-ml syringe plunger. Cell mixtures were subjected to hypotonic lysis (NH 4 Cl) to remove red blood cells, washed twice in FACS buffer (2 mM EDTA and 1% foetal bovine serum (FBS) in PBS) and resuspended in FACS buffer containing the corresponding fluorescent dye-conjugated antibodies. All staining was carried out at 1:100 dilution and with an Fc block for 30 min at 4 °C in the dark. Samples were washed twice with FACS buffer before further analysis. All flow data were acquired on a FACS Fortessa flow cytometer (BD Biosciences) and analysed using FlowJo software (Tree Star) ( Table 1).

Mice
EAE model in C57BL/6J mice. Female C57BL/6 mice (10-12 weeks of age) or other mouse lines with the C57BL/6J genetic background were immunized with Hooke kits (an emulsion of MOG  in CFA and PTX in PBS) according to the manufacturer's instructions (Hooke Laboratories). Mice were scored daily, starting on day 7 post-immunization, by an investigator blinded to the experimental treatment of individual mice. Mice were randomly assigned to different experimental treatments and cohoused together to eliminate inter-cage variability. All treatments were carried out on at least three mice and in at least two independent experiments, as indicated in the figure captions. All of the animals were included in the analyses. Clinical scores were defined as follows: 1 = limp tail; 2 = partial hind leg paralysis; 3 = complete hind leg paralysis; 4 = complete hind and partial front leg paralysis; and 5 = moribund. Easy access to wet food and water was provided for the experimental mice throughout disease progression. Unless indicated otherwise, for prophylactic treatment, 20 μg sortagged VHHantigens were administered intravenously 7 d before the induction of EAE. For therapeutic treatment, 20 μg VHH MHCII -OVA 323-339 , 20 μg VHH MHCII -MOG  or 20 μg VHH MHCII -MOG 35-55 mixed with 20 μg VHH MHCII -DEX were administered on the day of EAE when the mice exhibited symptoms defined as a clinical score of 1, 2 or 3, as indicated. At day 30 post-EAE induction or when mice reached a clinical score of 4, they were sacrificed by asphyxiation, then perfused with 5 mM EDTA in PBS. Spinal cords were isolated and fixed in 10% (wt/vol) formalin solution (Sigma-Aldrich), embedded in paraffin, sectioned at 20 μm and stained with haematoxylin and eosin or Luxol fast blue (Harvard Medical School Rodent Histology Core Facility). Stained sections were imaged at 4 and 10× magnification. Isolation of the immune cells infiltrating the spinal cord was carried out by homogenizing the spinal cord, followed by 38% Percoll (Sigma-Aldrich) gradient separation (100% Percoll is 1.123 g ml −1 ). Isolated cells were plated in 48-well plates and treated with 50 ng ml −1 PMA (Sigma-Aldrich) and 500 ng ml −1 ionomycin (Sigma-Aldrich) for 2 h at 37 °C in complete RPMI media, followed by the addition of 10 μg ml −1 Monensin (Sigma-Aldrich), and incubated for two more hours. Cells were then surface stained, fixed and permeabilized using the Foxp3/ Transcription Factor Staining Buffer Set (Thermo Fisher Scientific; 00-5523-00) according to the manufacturer's protocol. Intracellular and Foxp3 staining were performed according to the manufacturer's protocols and cell samples were then used for flow cytometry.
Cellular subset depletion. CD8 T cells were depleted by administering 400 μg anti-CD8α depleting antibody (clone 2.43; BioXCell) intraperitoneally twice weekly beginning 2 weeks before prophylactic treatment with VHH-antigen and throughout the EAE observation window. Macrophage subsets were ablated by injecting 300 μg anti-CSF1R (clone AFS98; BioXCell) every other day from 2 weeks before prophylactic treatment up to the end of the experimental set up. To deplete dendritic cells, we administered 100 ng DTX (Sigma-Aldrich) intraperitoneally into CD11c-DTR mice 2 d before VHH-antigen administration. To deplete T reg cells, FoxP3-DTR mice were injected with three doses of 1 μg DTX (Sigma-Aldrich) intraperitoneally at days −9, −8 and −1 before prophylactic treatment with VHH-antigen and weekly afterwards until the end of the observation window. Cellular depletions were confirmed by flow cytometry of PBMCs or splenocytes.
2D2 CD4 T cell adoptive transfer and challenge. Splenic and inguinal lymph node-derived CD4 T cells from 2D2 mice were enriched by negative selection using magnetic beads (Miltenyi Biotec; 130-104-453) and labelled with CellTrace Violet (Thermo Fisher Scientific, C34571) per the manufacturer's protocol. Some 500,000 of these 2D2 CD4 + T cells were transferred into CD45.1 + mice. Transfusion of 20 μg VHH MHCII -OVA 323-339 , 20 μg VHH MHCII -MOG  , equimolar of MOG 35-55 peptides or 100 μg MOG 35-55 peptides mixed with 25 μg anti-CD40 (SouthernBiotech) and 50 μg PolyI:C (Sigma-Aldrich) as adjuvant was carried out the day after adoptive transfer. At days 3, 5 and 10, mice were sacrificed and spleens, inguinal lymph nodes and blood were collected and analysed by flow cytometry. Some of these 2D2 T cell adoptively transferred mice were also challenged on day 3 or 10 with 100 μg MOG  in CFA subcutaneously. Mice were sacrificed 7 or 5 d later, as indicated in the respective experimental set up. Spleens, inguinal lymph nodes and blood were harvested and analysed by flow cytometry.
2D2 CD4 T cell RNA-seq. Cells were sorted and lysed in RLT lysis buffer (Qiagen) supplemented with β-mercaptoethanol. RNA was then isolated using the RNeasy Micro kit (Qiagen) according to the manufacturer's protocol. Next, 20 ng RNA was used as input to a modified SMART-seq2 protocol. The resulting library was confirmed using a High Sensitivity DNA Chip run on a Bioanalyzer 2100 system (Agilent), followed by library preparation using the Nextera XT kit (Illumina) and custom index primers according to the manufacturer's protocol. Final libraries were quantified using a Qubit dsDNA HS Assay kit (Invitrogen) and a High Sensitivity DNA Chip run on a Bioanalyzer 2100 system (Agilent). All libraries were sequenced using NextSeq High Output Cartridge kits and a NextSeq 500 sequencer (Illumina). Sequenced libraries were demultiplexed using the bcl2fastq program and the resulting fastq data were trimmed and cropped with Trimmomatic. Alignment to the mouse mm10 reference genome and gene expression counts were carried out using Kallisto. Principal component analyses were carried out in R. To test for differential gene Repeated transfusions of VHH MHCII -OB1. OB1 is a 17-mer B cell epitope derived from OVA. C57BL6/J recipient mice were intravenously injected with 20 μg VHH MHCII -OB1 or an equimolar amount of OVA proteins or PBS at day 0. Subsequent boosts were carried out on days 7 and 14. Serum samples were collected pre-immunization and 7 d after the last boost. For OVA-specific and OB1 peptide-specific ELISAs, 96-well plates were coated with 10 μg ml −1 OVA or GFP-OB1 proteins in PBS overnight at 4 °C and incubated in blocking buffer (0.05% Tween 20 + 2% BSA in PBS) before the addition of serum samples. Incubation with tested serum was for 3 h at room temperature. Plates were washed four times with PBS, incubated with goat anti-mouse IgG-HRP (SouthernBiotech) at 1:10,000 in blocking buffer for 1 h and developed with 3,3′,5,5′-tetramethylbenzidine liquid substrate reagent (Sigma-Aldrich). The reaction was stopped with 1 N HCl and the absorbance was read at 450 nm.
Statistical analysis. All of the data represent at least two independent experiments. All statistical analyses were performed using Prism 6. The statistical methods used are indicated in the corresponding figure captions. Statistically significant differences are indicated by asterisks (***P < 0.001).
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

data availability
The main data supporting the results of this study, including the nanobody sequences used, are available within the paper and its Supplementary Information. The transcriptomic datasets generated during the current study are available from the corresponding authors upon reasonable request.

nature research | reporting summary
April 2020 Corresponding author(s): Novalia Pishesha and Hidde L. Ploegh Last updated by author(s): Apr 23, 2021 Repor�ng Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in repor�ng. For further informa�on on Nature Research policies, see our Editorial Policies and the Editorial Policy Checklist.

Sta�s�cs
For all sta�s�cal analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods sec�on.

n/a Confirmed
The exact sample size (n) for each experimental group/condi�on, given as a discrete number and unit of measurement A statement on whether measurements were taken from dis�nct samples or whether the same sample was measured repeatedly The sta�s�cal test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A descrip�on of all covariates tested A descrip�on of any assump�ons or correc�ons, such as tests of normality and adjustment for mul�ple comparisons A full descrip�on of the sta�s�cal parameters including central tendency (e.g. means) or other basic es�mates (e.g. regression coefficient) AND varia�on (e.g. standard devia�on) or associated es�mates of uncertainty (e.g. confidence intervals) For null hypothesis tes�ng, the test sta�s�c (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted

So�ware and code
Policy informa�on about availability of computer code Data collec�on No so�ware was used.

Data analysis
No so�ware was used.
For manuscripts u�lizing custom algorithms or so�ware that are central to the research but not yet described in published literature, so�ware must be made available to editors and reviewers. We strongly encourage code deposi�on in a community repository (e.g. GitHub). See the Nature Research guidelines for submi�ng code & so�ware for further informa�on.

Data
Policy informa�on about availability of data All manuscripts must include a data availability statement. This statement should provide the following informa�on, where applicable: -Accession codes, unique iden�fiers, or web links for publicly available datasets -A list of figures that have associated raw data -A descrip�on of any restric�ons on data availability The main data suppor�ng the results, including the nanobody sequences used in this study, are available within the paper and its Supplementary Informa�on. The transcriptomic datasets generated during the current study are available from the corresponding authors on reasonable request.