Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 https://doi.org/10.1186/s12974-019-1541-1 RESEARCH Open Access Clinical and immunological control of experimental autoimmune encephalomyelitis by tolerogenic dendritic cells loaded with MOG-encoding mRNA Judith Derdelinckx1,2* , María José Mansilla3,4, Maxime De Laere1, Wai-Ping Lee1,5, Juan Navarro-Barriuso3,4, Inez Wens1, Irene Nkansah1, Jasmijn Daans1, Hans De Reu1, Aneta Jolanta Keliris6, Johan Van Audekerke6, Verdi Vanreusel6, Zoë Pieters7,8, Annemie Van der Linden6, Marleen Verhoye6, Geert Molenberghs7,9, Niel Hens7,8, Herman Goossens10, Barbara Willekens1,2, Patrick Cras2,11, Peter Ponsaerts1, Zwi N. Berneman1,5, Eva María Martínez-Cáceres3,4 and Nathalie Cools1,5 Abstract Background: Although effective in reducing relapse rate and delaying progression, current therapies for multiple sclerosis (MS) do not completely halt disease progression. T cell autoimmunity to myelin antigens is considered one of the main mechanisms driving MS. It is characterized by autoreactivity to disease-initiating myelin antigen epitope(s), followed by a cascade of epitope spreading, which are both strongly patient-dependent. Targeting a variety of MS- associated antigens by myelin antigen-presenting tolerogenic dendritic cells (tolDC) is a promising treatment strategy to re-establish tolerance in MS. Electroporation with mRNA encoding myelin proteins is an innovative technique to load tolDC with the full spectrum of naturally processed myelin-derived epitopes. Methods: In this study, we generated murine tolDC presenting myelin oligodendrocyte glycoprotein (MOG) using mRNA electroporation and we assessed the efficacy of MOG mRNA-electroporated tolDC to dampen pathogenic T cell responses in experimental autoimmune encephalomyelitis (EAE). For this, MOG35–55- immunized C57BL/6 mice were injected intravenously at days 13, 17, and 21 post-disease induction with 1α, 25-dihydroxyvitamin D3-treated tolDC electroporated with MOG-encoding mRNA. Mice were scored daily for signs of paralysis. At day 25, myelin reactivity was evaluated following restimulation of splenocytes with myelin-derived epitopes. Ex vivo magnetic resonance imaging (MRI) was performed to assess spinal cord inflammatory lesion load. (Continued on next page) * Correspondence: judith.derdelinckx@uza.be; Judith.derdelinckx@uantwerp.be 1 Laboratory of Experimental Hematology, Faculty of Medicine and Health Sciences, Vaccine and Infectious Disease Institute (VaxInfectio), University of Antwerp, Antwerp University Hospital (UZA), Wilrijkstraat 10, 2650 Edegem, Belgium 2 Division of Neurology, Antwerp University Hospital, Edegem, Belgium Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 2 of 20 (Continued from previous page) Results: Treatment of MOG35–55-immunized C57BL/6 mice with MOG mRNA-electroporated or MOG35–55- pulsed tolDC led to a stabilization of the EAE clinical score from the first administration onwards, whereas it worsened in mice treated with non-antigen-loaded tolDC or with vehicle only. In addition, MOG35–55-specific pro-inflammatory pathogenic T cell responses and myelin antigen epitope spreading were inhibited in the peripheral immune system of tolDC-treated mice. Finally, magnetic resonance imaging analysis of hyperintense spots along the spinal cord was in line with the clinical score. Conclusions: Electroporation with mRNA is an efficient and versatile tool to generate myelin-presenting tolDC that are capable to stabilize the clinical score in EAE. These results pave the way for further research into mRNA-electroporated tolDC treatment as a patient-tailored therapy for MS. Keywords: Messenger RNA electroporation, Tolerogenic dendritic cells, Antigen-specific treatment, Experimental autoimmune encephalomyelitis, Multiple sclerosis, Tolerance induction, Background knowledge of the disease-related antigens. In the set- Multiple sclerosis (MS) is a chronic demyelinating dis- ting of MS, there are indications that the adaptive order of the central nervous system (CNS), clinically autoimmune reaction is directed toward myelin pro- reflected by episodes of neurological dysfunction. Exten- teins, including MOG, myelin basic protein (MBP), sive research over the past decades has revealed a key and proteolipid protein (PLP) [2]. However, although role for the immune system, with a complex attribution immune-dominant myelin epitopes derived from of both the innate and adaptive arms of immunity [1]. various myelin proteins have been described in MS This knowledge drove the development of a wide arma- patients [10], the specific pattern of myelin reactivity mentarium of disease-modifying therapies for MS, which is diverse and strongly patient-dependent [11, 12]. are able to reduce autoimmune inflammation-driven Moreover, there is strong evidence that the myelin- clinical and radiological disease activity with varying effi- specific T cell reactivity in MS patients is dynamic cacy. However, their immunomodulatory mode of action over time, a phenomenon called epitope spreading does not specifically tackle the dysfunctional immune re- [13–15]. This is believed to play an important role in sponse toward myelin, which is presumed to steer MS the progression of both EAE and MS [14–16]. How- pathogenesis [2]. Moreover, this broader immune modu- ever, in contrast to animal models of MS, in which lation can induce off-target effects, leading to adverse consistent epitope spreading cascades could be events such as opportunistic infections, secondary auto- detected [17], the nature of epitope spreading in MS immunity, and malignancies [3]. Thus, the need for patients seems to be more complex [13] and further more targeted, disease antigen-specific treatment strat- complicates the choice of target antigens for tolDC egies remains. loading. In addition, one needs to carefully distinguish In this context, the use of myelin-specific tolerogenic between naturally processed and cryptic epitopes. In- dendritic cells (tolDC) is a promising strategy to re-estab- deed, even though reactivity toward certain so-called lish tolerance in an antigen-specific manner. Indeed, a cryptic epitopes can be demonstrated ex vivo, these beneficial effect of intravenous administration of bone epitopes are not naturally processed and are thus not marrow-derived, 1α,25-dihydroxyvitamin D3 (vitD3)- directly involved in disease pathogenesis [18–22]. treated tolDC loaded with myelin oligodendrocyte glyco- Therefore, it can be hypothesized that induction of protein (MOG)40–55 peptide was demonstrated in a mouse tolerance toward cryptic epitopes may have no benefi- model of experimental autoimmune encephalomyelitis cial effect on disease course, as was demonstrated by (EAE) [4, 5]. Interestingly, there was no clear beneficial ef- Anderton et al. who failed to prevent MBP-directed fect on the clinical score of EAE mice when they were EAE by inducing tolerance toward the cryptic epitope treated with tolDC that were not loaded with myelin-de- MBP89–101 [21]. Lastly, many synthetic peptides are rived peptides [4–6]. Similar findings have been restricted to specific human leukocyte antigen (HLA) demonstrated in other animal models of autoimmune subtypes [23, 24] and therefore cannot be used irre- diseases, including collagen-induced arthritis [7, 8] and spective of the HLA status of the donor, which limits autoimmune thyroiditis [9]. Altogether, these findings their use for antigen loading of tolDC. suggest that antigen loading is indispensable for optimal An elegant strategy to tackle the above-mentioned functioning of tolDC-based therapies. obstacles is to load DC with antigens using mRNA The choice of antigen for tolDC loading is challen- electroporation. This non-viral, non-DNA-based gene ging due to different factors. First, it requires prior delivery method introduces mRNA encoding full- Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 3 of 20 length proteins intracellularly by application of a low- culture medium. Half of the culture medium was voltage electrical field. Therefore, electroporation replenished on days 2, 4, and 6 with complete facilitates the presentation of multiple naturally proc- medium. On day 7, cells were either left untreated for essed epitopes by all HLA molecules of a particular the generation of immature DC (iDC), or were patient [25–28], rather than a limited number of spe- stimulated for 24 h with 0.1 μg/mL lipopolysaccharide cific HLA-restricted epitopes. This strategy will allow (LPS) (Invivogen, Toulouse, France) for the generation broadening the repertoire of responding lymphocytes of mature DC (mDC) and tolDC. We refer to without the need for prior selection of specific epi- Additional file 1: Figure S1 for a graphical overview topes to be targeted. of the cell culture protocol. In the present study, we addressed the feasibility to generate murine MOG-presenting tolDC by mRNA Plasmids and mRNA electroporation. Furthermore, we aimed to provide proof- The complementary DNA sequences of murine of-concept for the use of myelin mRNA-electroporated myelin oligodendrocyte glycoprotein (MOG) were tolDC as a potential treatment for MS by assessment of modified for optimal codon use in murine cells and the efficacy of MOG mRNA-electroporated tolDC to subcloned into a pST1-plasmid vector under the dampen pathogenic T cell responses in an animal model control of a T7 promotor and with the addition of a of MS, i.e., EAE. Ultimately, the use of mRNA electropor- poly(A) tail (GeneArt, Thermo Fisher Scientific, Life ation to load tolDC with myelin antigens may offer the Technologies, Merelbeke, Belgium). For lysosomal potential to tackle the dynamic and complex multi-epi- targeting of the translated protein, the MOG se- tope-targeted loss of tolerance in MS. quence was preceded by a lysosomal-targeting signal sequence (Sig) [29] and followed by the transmem- Materials and methods brane and extracellular domain of the murine DC- Ethics lysosomal-associated membrane protein (LAMP)-3. The procedures described in this study were approved Two plasmid vectors were generated: one encoding by the Ethics Committee for Animals of the University full-length murine MOG (amino acids 1-246) (Sig- of Antwerp (study number 2014-93 and 2017-08) or by MOG-LAMP) and one encoding extracellular murine the Ethics Committee on Animal Experimentation of the MOG (amino acids 29–156) (Sig-extracellular MOG- Germans Trias i Pujol Research Institute and by the LAMP) (see Additional file 2: Figure S2 for a Generalitat de Catalunya (protocol number 9469). All graphical overview of the protein structure of MOG experiments were performed in strict accordance with and Additional file 3: Figure S3 for the codon-opti- EU and governmental regulations. mized plasmid sequences). Additionally, DNA plas- mids encoding enhanced green fluorescent protein Cell culture (eGFP, pGEM4Z/EGFP/A64 vector [30] kindly pro- The myeloid leukemia cell line K562 (American Type vided by Dr. Eli Gilboa, Duke University Medical Culture Collection, Rockville, MD, USA) was cultured Center, Durham, NC, USA) or codon-optimized hu- in culture medium consisting of Iscove’s modified man C-C motif chemokine receptor 5 (CCR5) were Dulbecco’s medium (IMDM) supplemented with 10% used. All plasmids were propagated in Escherichia fetal bovine serum (FBS; Life Technologies). coli supercompetent cells (Stratagene, La Jolla, CA, Murine dendritic cells (DC) were generated from USA) and plasmid DNA was purified using a bone marrow precursor cells as described previously NucleoBond® Xtra Midi EF kit (Macherey-Nagel, by Mansilla et al. [4, 5]. Briefly, bone marrow cells Düren, Germany). Next, plasmid DNA was linearized from C57BL/6 mice were cultured following 8 days in by SapI digestion for the MOG and CCR5 plasmids complete medium consisting of Roswell Park Memor- and by SpeI digestion for the eGFP plasmid. Subse- ial Institute (RPMI) 1640 medium (Life Technologies, quently, linearized plasmid DNA was used as a DNA Thermo Fisher Scientific, Merelbeke, Belgium) supple- template for in vitro transcription with a T7 in vitro mented with 4 mM L-glutamine (Life Technologies), transcription kit (mMessage mMachine T7 kit, 10 μg/mL gentamicin (Life Technologies), 1 μg/mL Ambion, Life Technologies), according to the manu- amphotericin B (Life Technologies), 20% FBS (Life facturer’s protocol. All mRNA constructs were stored Technologies), 1 mM sodium pyruvate (Life Tech- at − 20 °C at a concentration of 1 μg/μL. nologies), and 1000 IU/mL murine granulocyte macrophage colony-stimulating factor (GM-CSF) Antigen loading (Peprotech, London, UK) to generate DC. For the On day 7 of the DC culture, cells were electroporated generation of tolDC, 1 nM vitamin D3 (vitD3, Calci- with mRNA encoding full-length or extracellular murine jex, Abbott Laboratories, Turkey) was added to the MOG, as described previously [31]. For this, cells were Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 4 of 20 resuspended in Opti-MEM medium (Life Technologies) Flow cytometry at a concentration of 5–50 × 106 cells/mL. Electro- Expression of MOG by the K562 cell line was assessed poration was performed with a Gene Pulser Xcell™ by flow cytometric analysis 2 h, 4 h, 24 h, 96 h following Electroporation system (Bio-Rad, California, US) using a electroporation using intracellular staining with the anti- time constant pulse of 300 V for 7 ms and 1 μg mRNA MOG Z2 primary antibody [32], kindly provided by prof. per 106 cells with a minimum of 5 μg of mRNA per elec- C. Linington (University of Glasgow, UK), and a rabbit troporation. Non-electroporated tolDC and tolDC anti-mouse Fab′-FITC secondary antibody (Dako, Agi- electroporated in the absence of mRNA (i.e., mock-elec- lent Technologies, Diegem, Belgium). Cells were ana- troporated) were used as control cells. For evaluation of lyzed on a BD FACScan flow cytometer (Becton the transfection efficacy following electroporation with Dickinson). Sig-MOG-LAMP and Sig-extracellular MOG-LAMP Immunophenotyping of DC was performed using the mRNA (mRNA length 1380 and 1023 nucleotides, re- following fluorochrome-labeled anti-mouse monoclonal spectively), K562 myeloid cells, mDC, and tolDC were antibodies: anti-CD11c-PE-Cy7 (BD), anti-CD86-PE electroporated with equal numbers of mRNA copies (i.e., (eBioscience, Thermo Fisher Scientific), anti-CD40-APC 5 μg of Sig-extracellular MOG-LAMP mRNA and (Biolegend, San Diego, United States), anti-MHC-II I/A I/ 6.75 μg Sig-MOG-LAMP mRNA per electroporation, E-Pacific Blue (Biolegend), anti-CD169-PE (Biolegend), representing 9.16 × 1012 RNA copies for both con- and anti-F4/80-APC (eBioscience). Phenotypic analysis of structs). Following electroporation, cells were left splenocytes obtained from mice in the in vivo model was untreated for the generation of iDC, or were stimulated performed upon blocking with TruStain FcX™ (anti-mouse with 0.1 μg/mL LPS for the generation of mDC and CD16/32) antibody (BioLegend) using the following tolDC. Twenty-four hours later, transfected cells were fluorochrome-labeled anti-mouse monoclonal antibodies: used in further experiments or cryopreserved. anti-CD3-FITC (BioLegend), anti-CD4-V450 (BD), anti- For the generation of MOG35–55 peptide-pulsed CD25-PE (BioLegend), anti-CD19-FITC (BioLegend), tolDC, cells were cultured for an additional 18 h in anti-CD5-PE (BioLegend), anti-CD1d-Alexa Fluor 647 the presence of 10 μM murine MOG35–55 (MEVG- (BioLegend), anti-NK1.1-APC-Cy7 (BioLegend), anti- WYRSPFSRVVHLYRNGK) (Immunostep, Salamanca, NKp46-PE (BioLegend), and anti-CD27-Alexa Fluor 647 Spain) and subsequently cryopreserved. (BioLegend). LIVE/DEAD™ Fixable Aqua Dead Cell Stain (Life Technologies, Thermo Fisher Scientific) was added to assess cell viability. Isotype-matched control monoclo- Cryopreservation conditions nal antibodies or Fluorescence Minus One (FMO) DC or lymphocytes were cryopreserved in freezing controls were used to determine non-specific background medium containing FBS and 10% dimethyl sulfoxide staining in the DC or the splenocytes phenotyping experi- (DMSO, Sigma-Aldrich, Bornem, Belgium). Cells were ments, respectively. Cells were analyzed on a CyFlow® ML slowly frozen to − 80 °C at a cooling rate of − 1 °C/min flow cytometer (Sysmex Partec, Norderstedt, Germany). by using a Mr. Frosty freezing container (Nalgene, For analytical flow cytometry, at least 104 events were Rochester, USA). After 24–72 h, cells were transferred measured. All results were analyzed using FlowJo soft- to liquid nitrogen for long-term storage. ware (Tree Star, Ashland, USA). qRT-PCR EAE induction and in vivo administration of Total RNA was extracted from the transfected cells 2 h electroporated tolDC following electroporation using an RNeasy Micro Kit Female C57BL/6JOlaHsd mice (8–10 weeks old) were (Qiagen, Düsseldorf, Germany), according to the manu- purchased from Envigo (Horst, The Netherlands). facturer’s instructions. PCR was performed using SYBR Mice were housed in the animal house of the Green technology (SsoAdvanced™ Universal SYBR® Germans Trias i Pujol Research Institute (Badalona, Green Supermix, Bio-Rad). Primer pairs specific for the Spain) under standard light- and climate-controlled codon-optimized MOG construct were used (forward conditions, with standard chow and water provided 5′-TGAGAATCCAGAACGTGCG-3′, reverse 5′-TCCT ad libitum. CGACCTTCAGTTCCA-3′, Bio-Rad). Relative mRNA Mice were immunized subcutaneously with 100 μg levels were normalized to levels of the reference gene MOG35–55 peptide (Immunostep) emulsified in Freund’s glyceraldehyde-3-phosphate dehydrogenase (GAPDH). complete adjuvant (1:1) containing 4 mg/mL Mycobac- Experiments were performed using a CFX96 thermal terium tuberculosis (strain H37RA, Difco, Detroit, MI, cycler (Bio-Rad) and data analysis was performed using USA). Mice received 250 ng pertussis toxin (Sigma Bio-Rad CFX Manager version 3.1 and qbase+ (Bioga- Chemical, St. Louis, MO, USA) intravenously at days 0 zelle, Ghent, Belgium). and 2. All animals were examined daily for well-being Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 5 of 20 and clinical signs by two blinded observers according to gel containing 1.5% agar (Sigma-Aldrich, Belgium), 5% the following criteria: 0, asymptomatic; 0.5, loss of tone sucrose (Sigma-Aldrich, Belgium), and 1% Dotarem. MR in the distal half of the tail; 1, loss of tone in the entire images were acquired in the axial plane (approx. perpen- tail; 1.5, hind limb weakness; 2, hind limb paralysis; 2.5, dicular on spinal cord) on a 7T MRI scanner (Bruker, hind limb paraplegia; 3, forelimb weakness; 4, quadripar- Ettlingen, Germany), operating on the Paravision 6 soft- esis; 4.5, severe quadriparesis; 5, quadriplegia; and 6, ware platform (Bruker, Ettlingen, Germany), using a death. Endpoint criteria were established to minimize quadrature transmit-only resonator 89/72 (Bruker, suffering and ensure animal well-being. To analyze the Ettlingen, Germany) combined with a receive-only four- efficacy of the different treatment options, the number channel array coil. For acquisition, a multi-slice single of responder mice was calculated, with a responder as spin echo sequence was used with the following parame- previously defined by Mansilla et al. [4], i.e., a mouse ters: TR 1700 ms, TE 11 ms, 40 slices, TA 12 h, FOV showing < 1-point increase in the mean clinical score 10 × 10 mm2, acquisition matrix 256 × 256, slice thick- compared to their respective initial clinical score at first ness 0.6 mm, no slice gap, number of slices 40, voxel treatment administration (day 13 pi). resolution 40 × 40 × 600 μm3. Images were analyzed To predetermine the sample size for the in vivo ex- using MRIcron software [35]. Three blinded observers periment, an a priori ANOVA power analysis was per- assessed the number of hyperintense white matter spots formed using G*Power [33], following data from independently. Mansilla et al. [5], with effect size = 0.51, α = 0.05, power = 85%, and number of groups = 4. This showed a MOG-specific proliferation assay required sample size of 13 mice per treatment group. For evaluation of the MOG35–55-specific splenocyte Mice were injected intravenously at the first day the reactivity, 2 × 105 fresh splenocytes were restimulated mean clinical score exceeded 1.0, reflecting a clinically with 5 μM MOG35–55 peptide (Immunostep) in 96- relevant disease onset in the majority of mice, (day 13 well plates for 3 days. Unstimulated splenocytes were pi) with (i) 1 × 106 Sig-extracellular MOG-LAMP used as a negative control. Splenocytes stimulated mRNA-electroporated tolDC, (ii) 1 × 106 MOG35–55- with 50 ng/mL phorbol myristate acetate (PMA) peptide-pulsed tolDC, (iii) 1 × 106 non-antigen-loaded (Sigma-Aldrich) and 500 ng/mL ionomycin (Sigma-Al- and non-electroporated tolDC, or (iv) PBS (vehicle drich) served as a positive control. After 48 h, 50 μL control). Injections were repeated at day 17 and 21 of cell culture supernatant was collected and stored at pi. Mice were sacrificed at different time points dur- − 20 °C for later cytokine secretion analysis. Next, ing the experiment: (i) before EAE induction (n = 4), 1 μCi of [3H]-thymidin (PerkinElmer, Waltham, MA, i.e., healthy mice; (ii) following EAE induction, but USA) was added per well for an additional 18 h of without treatment, at day 14 pi, i.e., immunized mice culture. The stimulation index (SI) was calculated as (n = 3); (iii) between the second and third vaccination, the mean counts per minute (cpm) of antigen-stimu- at day 20 pi, i.e., intermediate time point (n = 3 per lated conditions divided by the mean cpm of the treatment group); and (iv) at the end of the experi- non-stimulated conditions. ment, at day 25 pi, i.e., end point (n = 11 for the PBS- treated group and n = 10 for the other treatment Cytokine release assays groups) (Additional file 4: Figure S4). Following eu- The secretion of interleukin (IL)-12p70 by DC was thanasia, spleens and inguinal and axillary lymph determined using a murine IL-12p70 (eBioscience) nodes were obtained. Spinal cords were obtained from ELISA kit. three mice per treatment group, representative of The concentration of IL-4, IL-17A, GM-CSF, IL-10, treatment group’s mean and range, and fixed in 4% vascular endothelial growth factor (VEGF)-A, IL-13, and paraformaldehyde (PFA). tumor necrosis factor (TNF)-α in the supernatant fol- lowing restimulation of fresh splenocytes with MOG35– Magnetic resonance imaging 55 was quantified using a Meso Scale Discovery (MSD) PFA-fixed spinal cords from three mice per treatment mouse U-plex assay kit (Meso Scale Discovery, Rockville, group were prepared for ex vivo magnetic resonance im- MD, USA). The concentration of interferon (IFN)-γ was aging (MRI) acquisition as described by Noristani et al. determined using an IFN-γ (Peprotech) ELISA kit. All [34]. Briefly, spinal cords were soaked for 48 h in a assays were performed following the manufacturer’s 0.1 M PBS (10× PBS; pH 7.4) solution supplemented instructions. with a gadolinium-based MRI contrast agent (1% To assess the in vitro allogeneic T cell-stimulatory Dotarem, 0.5 mmol gadoteric acid/mL; Guerbet, France) capacity of DC, splenocytes obtained from BALB/C mice in order to enhance the MRI contrast. Next for scan- were stimulated with DC at a 10:1 ratio. Unstimulated ning, spinal cords were fixed in a 1 mL syringe using a splenocytes served as negative control. As positive Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 6 of 20 control, splenocytes were stimulated with 50 ng/mL or non-parametric analyses, as appropriate, following PMA (Sigma-Aldrich) and 500 ng/mL ionomycin Kolmogorov-Smirnoff analysis for normal distribution. (Sigma-Aldrich). Following 5 days of coculture, the level To examine the difference in clinical score between the of IFN-γ secretion in the supernatant was determined by groups, a nonlinear model was designed, taking into ac- means of a murine IFN-γ ELISA kit (Peprotech) follow- count the longitudinal nature of the data and the miss- ing the manufacturer’s instructions. ing data (introduced by sacrifice of mice during the To assess the in vitro MOG-specific T cell stimulatory course of the animal model) while providing a natural capacity following mRNA electroporation, 2.5 × 105 interpretation of the results. The model is described by thawed MOG35–55 reactive splenocytes were stimulated the following equation: with 2.5 × 104 of different mDC or tolDC conditions. β11 Puls þ β12 Unpuls þ β13 mRNA þ β14 PBS þ ui Co-cultures were performed overnight (18 h) at 37 °C in yij ¼   þ εij t ij − β21 Puls þ β22 Unpuls þ β23 mRNA þ β24 PBS a 96-well PVDF plate (Millipore, Bedford, MA, USA) 1 þ exp − β31 Puls þ β32 Unpuls þ β33 mRNA þ β34 PBS coated with anti-IFN-γ antibody. Stimulation of spleno- cytes was quantified by murine IFN-γ ELISPOT where yij is the clinical score for mouse i at time point j (Mabtech, NackaStrand, Sweden), according to manufac- and tij corresponds to the time at which mouse i at time turer’s instructions. Frequencies of IFN-γ-secreting cells point j is scored. The error term εij corresponds to the were calculated based on the number of spots counted error in measurement for mouse i at time j and is inde- using an automated AID ELISPOT Reader system (AID pendent and identically distributed (i.i.d.) N(0, σ2). The GmbH, Strassberg, Germany) and analyzed using AID random effects ui correspond to the additional change in ELISPOT software version 5.0. clinical score for an individual mouse; hence, the ran- Screening for epitope spreading was performed with dom effects measure the deviation of each mouse from IFN-γ ELISPOT, as described above, on thawed spleno- the average clinical score. ui follows a Nð0; σ 2u Þ . Thus, cytes. A total of 2 × 105 cells were stimulated overnight the combination of Beta1p + ui allow for a mouse-spe- with 5 μM of myelin peptide at 37 °C. The following mur- cific asymptotic clinical score. The error term εij is inde- ine myelin peptides were used: MBP4–14, MBP84–97, pendent and identically distributed (i.i.d.), N(0, σ2), and PLP139–151, PLP178–191, MOG92–106 (all from Anaspec, CA, the random effects ui follow a Nð0; σ 2u Þ. Figure 1 visual- USA, purchased through Eurogentec, Seraing, Belgium), izes the interpretation of the model parameters. β1p cor- PLP56–70 (Pepscan, Lelystad, The Netherlands), and responds to the asymptotic clinical score, β2p MOG35–55 (Sigma-Aldrich) (see Additional file 7: Table corresponds to the time at which half of the asymptotic S1). Unstimulated splenocytes served as a negative con- clinical score is achieved, and β3p corresponds to the trol. The following responder criteria were used: per 106 steepness of the linear part of the curve. The subscript p splenocytes, the mean antigen-specific spot counts (i.e., allows that all the features can take on different values the spot count in the peptide-stimulated condition) must depending on the treatment group. Here, p = 1–4 since be greater than or equal to 15 spots per well and at least four treatment groups are compared (Puls, Unpuls, 1.5 times as high as the background reactivity (i.e., the mRNA, PBS). At day 20, three mice from each group spot count in the negative control). were sacrificed. The model is able to handle missing data under the assumption of missing at random (MAR), i.e., Histological analysis if, conditional on the observed data, missing values are PFA-fixed spinal cord tissue was embedded longitudin- not dependent on unobserved data [36]. Under the as- ally in PELCO® cryo-embedding compound (Ted Pella, sumption of MAR, valid inference is obtained using a Redding, CA, USA) and 10-μm-thick cryosections were direct likelihood analysis, meaning the data is analyzed made in a Leica CM1950 cryotome (Leica, Nussloch, as is and no imputations are performed [36]. Germany). Staining was performed on spinal cord slides Of interest for the analysis is the parameter β1p. β1p using a classical hematoxylin-eosin (H&E) and a Luxol corresponds to the asymptotic clinical score; however, Fast Blue staining. Images were taken with an inverted we will refer to β1p as the maximum clinical score which Nikon Ti wide-field microscope equipped with auto- in turn is related to disease severity. This approximation mated stage and perfect focus system. is justified since the function arrives closely to the asymptotic value in a short time interval. A difference in Statistical analysis β1p value between the treatment groups represents a dif- Unless mentioned otherwise, results are shown as ference in severity of disease. Holm’s procedure was mean ± standard deviation (SD). Statistical analyses and used to correct p-values for multiple testing. graphical representations were performed with Graph- For all analyses, the p values were corrected using the Pad Prism 5.0 software (GraphPad Software) and SAS appropriate procedure. The adjusted p value < 0.05 was 9.4. Statistical analysis was performed using parametric considered as statistically significant. Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 7 of 20 In order to determine whether mRNA-electroporated DC could efficiently process the protein and efficiently present MOG peptide fragments, MOG mRNA-electropo- rated DC were co-cultured with splenocytes from MOG35–55-induced EAE mice. Non-electroporated mDC pulsed with MOG35–55 peptide served as a control for MOG presentation. As shown by the number of interferon (IFN)-γ-producing T cells quantified by means of ELI- SPOT analysis, extracellular MOG mRNA-electroporated mDC are able to present MOG antigen to MOG-reactive splenocytes to the same extent as non-electroporated mDC pulsed with MOG35–55 peptide (Fig. 2d). In contrast, no difference in the magnitude of the IFN-γ response could be detected following stimulation with full-length MOG mRNA-electroporated mDC in comparison to sple- Fig. 1 Interpretation of the model parameters. β1p corresponds to nocytes stimulated with non-electroporated mDC, dem- the maximum clinical score, β2p corresponds to the time at which onstrating no evidence for MOG presentation by full- half of the maximum clinical score is achieved, and β3p corresponds length MOG mRNA-electroporated mDC (Fig. 2d). to the steepness of the linear part of the curve. These features are allowed to vary depending on the treatment group p = 1, …,4 (Puls, In summary, our findings demonstrate that electropor- Unpuls, mRNA, PBS) ation with extracellular MOG-encoding mRNA is a valid method to induce MOG expression. Moreover, mDC elec- troporated with extracellular MOG mRNA are capable to effectively present antigenic MOG peptides to T cells. Results MOG mRNA electroporation of murine dendritic cells is Following MOG mRNA electroporation, vitamin D3-treated an effective tool to induce long-term myelin expression tolDC retain their maturation-resistant phenotype and presentation As previously described [4, 5], treatment of murine DC To evaluate the feasibility of electroporation with with vitamin D3 (vitD3) results in the generation of tolDC, mRNA encoding full-length and extracellular MOG, characterized by a maturation-resistant phenotype. Here, MOG mRNA levels were analyzed 2 h following we demonstrate that treatment with vitD3 results in a sig- mRNA electroporation in K562 myeloid cells, mature nificantly higher proportion of cells positive for the DC dendritic cells (mDC), and tolDC. MOG mRNA was marker CD11c (Table 1). Moreover, upregulation of the detected using RT-qPCR in cells electroporated with expression of costimulatory markers CD40 and CD86 and full-length MOG as well as in cells electroporated of major histocompatibility complex (MHC) class II mole- with extracellular MOG, while no mRNA was cules following stimulation with pro-inflammatory lipo- detected in non-electroporated cells (Fig. 2a). No dif- polysaccharide (LPS) was reduced in vitD3-treated tolDC ference in intracellular mRNA levels could be ob- as compared to mDC, reflected by a statistically significant served between cells that were electroporated with difference in the fold change of the mean fluorescent in- the full-length and extracellular MOG. As this con- tensity between mDC and tolDC as compared to imma- firmed that electroporation is a successful strategy to ture DC (iDC) (Table 1). In addition, vitD3-treated introduce mRNA into DC, we next investigated tolDC secreted lower amounts of interleukin (IL)- whether electroporated MOG mRNA was also trans- 12p70 following stimulation with pro-inflammatory lated into MOG protein using flow cytometric ana- LPS as compared to mDC (Table 1). Importantly, the lysis. MOG protein could be readily detected in reduced upregulation of the expression of costimula- K562 cells electroporated with mRNA encoding tory markers as well as the reduced secretion of IL- extracellular MOG, while no expression of MOG 12p70 by tolDC following challenge with an pro-in- protein could be detected in K562 cells electropo- flammatory product was not affected by mock or rated with full-length MOG mRNA (Fig. 2b). Expres- mRNA electroporation, except for the elevated ex- sion of MOG protein following electroporation with pression of CD86 following mRNA electroporation of mRNA encoding extracellular MOG was detected as tolDC. Similarly, the capacity of tolDC to induce T early as 2 h following transfection, with an median cell hyporesponsiveness in an allogeneic mixed of 84.8% (7.5% interquartile range) MOG-expressing lymphocyte reaction was retained after mock and cells, and lasted up to 96 h following electroporation mRNA electroporation. Indeed, whereas mDC were (Fig. 2c). strongly capable of inducing IFN-γ secretion by Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 8 of 20 Fig. 2 Induction of MOG expression and presentation by mRNA electroporation. a RT-qPCR analysis demonstrates intracellular presence of Sig- MOG-LAMP mRNA (Full-length MOG) and Sig-extracellular MOG-LAMP mRNA (Extracell. MOG), 2 h following electroporation. Results are shown as median log-transformed calibrated normalized relative quantity (CNRQ) values with minimum and maximum value (n = 3). Mann-Whitney U test was used for statistical analysis. b Flow cytometric analysis of MOG expression in K562 cells 24 h following electroporation with full-length MOG mRNA (red line), extracellular MOG mRNA (filled gray histogram), or irrelevant mRNA (CCR5 mRNA, green line). Controls are non-electroporated and mock-electroporated K562 cells (black line and yellow line, respectively). Representative histograms from one experiment are shown (n = 3). c Kinetics of MOG expression after electroporation of K562 with extracellular MOG mRNA (filled histogram) as compared to non-electroporated K562 cells (dotted line). Flow cytometric results from one representative experiment are shown (n = 3). d MOG presentation following mRNA electroporation. MOG35–55-reactive splenocytes were co-cultured with different conditions of mDC in a 10:1 ratio. Results are shown as mean and SD and groups are compared via one-way ANOVA with Bonferroni post-hoc test; *, statistically significant compared to mDC; +, statistically significant comparisons between groups denoted by solid line; **/++, p < 0.01; ***/+++, p < 0.001. non-EP non-electroporated, MOG myelin oligodendrocyte glycoprotein, CNRQ Calibrated Normalized Relative Quantity, B.D.L., below detection limit, N.S. not significant, SFC spot-forming cells Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 9 of 20 Table 1 Characterization of the tolerogenic phenotype of mRNA-electroporated tolDC iDC mDC tolDC tolDC mock EP tolDC mRNA EP A. Immune phenotype Gated on CD11c % positive cells 45.2 42.1 73.5 68.7 68.0 viable cells (42.9–53.5) (27.3–56.2) (71.9–76.8)AA,BB (66.0–80.1)A,B (65.2–73.8)A CD169 % positive cells 12.0 22.5 57.5 36.4 42.5 (6.0–23.6) (18.4–28.5) (23.4–70.3)A (20.5–65.7) (19.6–69.5) F4/80 % positive cells 24.1 17.3 76.3 70.4 75.5 (16.4–31.4) (11.8–27.6) (30.7–82.2)B (39.7–82.5)B (39.7–82.5)A,B Gated on viable CD86 MFI 13.9 55.1 21.4 37.0 60.8 CD11c+ cells (9.0–24.9)B (37.3–59.4)A (16.8–26.5) (29.2–52.0) (46.6–71.2)AAA,CC Fold change in MFI – 3.8 1.9 2.3 3.3 compared to iDC (1.7–4.8) (0.8–2.6)D (1.6–4.5) (2.3–7.1) MHC-II MFI 17.3 19.85 4.8 4.7 3.6 (9.3–28.8) (10.6–27.0) (2.1–5.6)A,B (2.3–5.3)A,B (2.5–5.6)A,B Fold change in MFI – 1.2 0.3 0.3 0.3 compared to iDC (0.9–1.5) (0.1–0.5)DD,BB (0.1–0.4)B (0.1–0.4)B CD40 MFI 2.5 12.8 7.5 6.0 7.3 (1.6–2.8)BBB (11.0–21.3)AAA (5.7–10.1)A (4.7–9.4) (5.2–9.1)A Fold change in MFI – 6.0 2.8 2.9 3.2 compared to iDC (4.4–7.8) (2.3–5.0)D,B (2.2–3.6)BB (2.8–3.9)B B. Cytokine secretion profile IL-12p70 (pg/mL) 2.3 356.3 73.7 74.3 99.0 (1.5–5.8)BBB (193.2–409.6)AAA (61.9–132.9)BBB (43.2–98.4)BBB (48.1–122.3)BBB A. Flow cytometric analysis of the expression of hallmark DC/macrophage and costimulatory markers by non-electroporated iDC, mDC, tolDC, and mock- or mRNA-electroporated tolDC (n = 8). mRNA-electroporated tolDC were electroporated with eGFP mRNA or full-length MOG mRNA. Results are shown as median (1st quartile-3rd quartile) of percentage of positive cells (compared to isotype control) or of mean fluorescent intensity (MFI). A, statistically significant when compared to iDC; B, statistically significant when compared to mDC; C, statistically significant when compared to tolDC, all using Kruskal-Wallis test with Dunn’s post-hoc test; D , statistically significant when compared to mDC using Mann-Whitney U test; A/B/C/Dp < 0.05; AA/BB/CC/DDp < 0.01; AAA/BBB/CCC/DDDp < 0.001. B. IL-12p70 secretion in the culture supernatant of iDC, mDC, and tolDC (n = 6). For this, DC were electroporated on day 6 of the cell culture protocol and stimulated on day 7 with a maturation cocktail consisting of 1 μg/mL LPS and 1000 IU/mL IFN-γ. Cell culture supernatant was harvested 24 h after addition of the maturation stimulus. Results are shown as mean ± standard deviation; A, statistically significant when compared to iDC; B, statistically significant when compared to mDC, using one- way ANOVA with Bonferonni’s post-hoc test; AAA/BBBp < 0.001 allogeneic splenocytes in co-culture, IFN-γ production electroporated with extracellular MOG mRNA. As was reduced when allogeneic splenocytes were stimu- depicted in Fig. 3b, extracellular MOG mRNA-electropo- lated with tolDC, irrespective of their electroporation rated tolDC displayed a reduced capacity to stimulate status (Fig. 3a). Hence, the statistically significant in- MOG-reactive splenocytes, as evidenced by a lower crease of CD86 upregulation on tolDC following elec- number of IFN-γ-producing cells in comparison with troporation does not reflect a physiologically relevant splenocytes stimulated with MOG mRNA-electroporated phenomenon, as indicated by the absence of a signifi- mDC. Similarly, MOG-specific splenocyte hyporespon- cant correlation between CD86 MFI and IFN-γ secre- siveness was observed following stimulation with tolDC tion by tolDC (Spearman r = − 0.298, p = 0.347). Based pulsed with MOG35–55, albeit, significantly less potently on these findings, we conclude that mRNA electro- than with MOG mRNA-electroporated tolDC (Fig. 3b; poration does not affect the tolerogenic phenotype of 59.9 ± 12% vs 86.6 ± 5% reduction in IFN-γ spot-forming vitD3-treated tolDC. cells, respectively, p < 0.05). In conclusion, extracellular MOG mRNA-electroporated tolDC induce hyporespon- siveness of MOG-reactive splenocytes in vitro. Extracellular MOG mRNA-electroporated tolDC effectively suppress MOG-specific splenocytes In order to evaluate the capacity of extracellular MOG In vivo administration of extracellular MOG mRNA- mRNA-electroporated tolDC to modulate MOG-specific electroporated tolDC reduces EAE development T cell responses, splenocytes from MOG35–55 EAE mice Next, we evaluated the therapeutic clinical effect of were stimulated overnight with (i) non-antigen-loaded, administering extracellular MOG mRNA-electroporated non-electroporated mDC or tolDC, (ii) mDC or tolDC tolDC on EAE. For this, mice were injected intraven- pulsed with MOG35–55, or (iii) mDC or tolDC ously as soon as the first clinical signs of disease Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 10 of 20 Fig. 3 Modulatory capacity of MOG mRNA-electroporated tolDC. a IFN-γ secretion by allogeneic splenocytes following stimulation with DC (n = 5). Results are shown as mean ± standard deviation. *p < 0.05; **p < 0.01; ***p < 0.001 using two-way ANOVA with Bonferroni post-hoc test. Importantly, no differences in IFN-γ secretion could be detected between tolDC, mock-electroporated tolDC, and mRNA-electroporated tolDC mRNA. N.S. not significant. b Modulation of MOG-specific T cell responses by tolDC following electroporation with MOG mRNA. MOG35–55-reactive splenocytes were cocultured with different conditions of mDC and tolDC in a 10:1 ratio. Following overnight incubation, the number of IFN-γ- secreting T cells was quantified using ELISPOT. Results are shown as mean + standard deviation and groups are compared via a one-way ANOVA with Bonferroni post-hoc test. A sole asterisk (*) indicates the results is statistically significant compared to mDC; +, statistically significant comparisons between groups denoted by solid line; **/++p < 0.01; ***/+++p < 0.001. SFC spot-forming cells, N.S. not significant, extracell. extracellular appeared (day 13 post induction [pi]) with (i) 1 × 106 between the groups, a significant difference was ob- extracellular MOG mRNA-electroporated tolDC, (ii) 1 × served for mice treated with MOG mRNA-electropo- 106 MOG35–55-pulsed tolDC, (iii) 1 × 106 non-antigen- rated (p = 0.04) and MOG35–55-pulsed tolDC (p = 0.01) loaded, non-electroporated tolDC, or (iv) phosphate- compared to mice treated with PBS. No significant dif- buffered saline (PBS) (vehicle control). Mice were scored ference in severity of disease was observed between mice daily from day 7 pi onwards for signs of paralysis treated with non-antigen-loaded tolDC (p = 0.28) and PBS (Fig. 4a). Upon examining the severity in disease (Fig. 4b). No statistically significant difference in the Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 11 of 20 Fig. 4 In vivo administration of extracellular MOG mRNA-electroporated tolDC abrogates EAE development. a Clinical follow-up of mice treated with 1 × 106 viable non-antigen-loaded and non-electroporated tolDC (n = 13), MOG35–55-pulsed tolDC (n = 13), MOG mRNA-electroporated tolDC (n = 13) or PBS (n = 14). Arrows represent days of treatment (day 13 pi, day 17 pi, and day 21 pi). Results are represented as mean ± standard error of the mean. b Marginal average evolutions of the clinical score, as generated by the non-linear model described in the “Materials and methods” section; *p < 0.05. N.S. not significant, PBS phosphate-buffered saline clinical course between the MOG mRNA-electroporated cases, the clinical score stabilized from the first adminis- and the MOG35–55-pulsed tolDC-treated groups could be tration onwards. observed (p = 0.53). We also determined the number of responder mice per group. As shown in Additional file 8: Table S2, a similar response rate (10/13, 76.9%) was ob- TolDC treatment in EAE mice inhibits MOG-specific tained following treatment with both MOG mRNA-elec- proliferation by splenocytes and lymph node-derived troporated tolDC and MOG35–55-pulsed tolDC with more lymphocytes animals responding to the treatment as compared to the MOG35–55-specific proliferation of fresh splenocytes and vehicle control group (4/14, 28.6%) or the non-antigen- lymph node-derived lymphocytes was evaluated during loaded tolDC group (7/13, 53.8%). A graphical presenta- the course of EAE. As shown in Fig. 5, no MOG35–55 tion of the responder mice per treatment group is antigen-specific proliferative reactivity could be demon- depicted in Additional file 5: Figure S5. strated in healthy mice, whereas this response peaked In summary, these findings demonstrate that treatment during the first inflammatory phase of EAE (day 14 pi), with tolDC has a beneficial effect on the clinical course of following in vitro rechallenge of splenocytes (Fig. 5a) as EAE only when cells are loaded with antigen, either by well as of lymph node-derived lymphocytes (Fig. 5b) using peptide-pulsing or mRNA electroporation. In both with MOG35–55. Treatment with tolDC inhibited Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 12 of 20 Fig. 5 Myelin reactivity of splenocytes and lymph node cells obtained from tolDC- or PBS-treated EAE mice. MOG35–55-specific proliferation of splenocytes (a) and lymph node cells (b) at different time points during the in vivo EAE model. Cells were rechallenged with MOG35–55 or were left unstimulated as a negative control. The stimulation index was calculated as the mean counts per minute (cpm) of MOG35–55-stimulated conditions divided by the mean cpm of the non-stimulated conditions. The mean per treatment group is depicted as a vertical black line. pi post induction MOG35–55-specific proliferation both at day 20 and day spreading to MOG92–106 occurs in PBS-treated mice 25 pi. at day 20 pi, but not in tolDC-treated mice, irrespective of antigen loading of the cells. Some PBS- Epitope spreading can be detected in untreated and PBS- treated mice also displayed intermolecular epitope treated EAE mice, but not in tolDC-treated mice spreading at this time point (one out of three mice Next, we evaluated the presence of intra- and inter- for MBP84–97 and PLP139–151, two out of three mice molecular epitope spreading during the course of for PLP178–191). Markedly, hardly any MOG35–55 re- EAE. Cryopreserved splenocytes were restimulated activity or epitope spreading could be detected in with different myelin-derived peptides. The number of tolDC-treated mice. IFN-γ-secreting cells was evaluated by means of ELI- In conclusion, these data demonstrate that both SPOT as a measure for antigen-specific stimulation intra- and intermolecular epitope spreading is present (Table 2). Only a very low number of myelin-specific in the chronic MOG35–55 EAE model in a dynamic responder mice was observed among the healthy and time-dependent manner. Neither myelin reactivity C57BL/6 mice, whereas immunized mice demon- nor epitope spreading could be detected in spleno- strated a clear and focused MOG35–55-specific re- cytes of mice treated with MOG mRNA-electropo- sponse at day 14 pi, corresponding with the rapid rated tolDC, and only sporadically in mice treated development of clinical symptoms. Interestingly, with with MOG35–55-pulsed tolDC or non-antigen-loaded further progression of the disease, intramolecular tolDC. Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 13 of 20 Table 2 Myelin-specific IFN-γ ELISPOT responses at different time points in the EAE disease course Number of responder mice is shown per myelin peptide, with between brackets the mean ratio of antigen-specific spot count over background spot count ± standard deviation. The following responder criteria were used: per 106 splenocytes, the mean antigen-specific spot count must be greater than or equal to 15 spots per well and at least 1.5 times as high as the background reactivity (i.e., the spot count in the negative control). pi post induction TolDC treatment effectively suppresses the strong lesser extent. The MOG35–55-induced pro-inflammatory EAE-mediated pro-inflammatory cytokine secretion cytokine secretion release reached its zenith at the end of In order to assess the dynamics of the cytokine secretion the experiment in the PBS-treated mice. At this time point, profile in EAE, we determined the levels of eight different secretion of VEGF-A and IL-13 could also be observed. cytokines secreted by splenocytes collected at different Interestingly, treatment with tolDC strongly suppressed time points during the disease and following rechallenge this MOG35–55-specific pro-inflammatory response. In with MOG35–55 (Table 3). At the moment of EAE disease conclusion, MOG35–55-directed splenocyte secretion pro- initiation, i.e., at day 14 pi, restimulation of splenocytes file in EAE is characterized by a strong Th17, Th1, and with MOG35–55 elicited strong T helper 17 (Th17) (IL- GM-CSF response, which is effectively suppressed by treat- 17A) and Th1 (IFN-γ, TNF-α) responses, known to be the ment with tolDC. major pro-inflammatory cytokines driving CNS inflamma- tion in EAE. Additionally, a high secretion of the pro-in- TolDC treatment attenuates spinal cord inflammatory flammatory cytokine GM-CSF was detected, along with lesion load secretion of Th2 cytokines (IL-4 and IL-10; the latter Magnetic resonance imaging of the spinal cord was known to have anti-inflammatory functions as well) to a performed in order to evaluate the effect of tolDC Table 3 Cytokine secretion profile of splenocytes following restimulation with MOG35–55 The cytokine secretion profile upon restimulation with MOG35–55, with particular focus on Th1, Th2, and Th17 cytokines, was analyzed during the in vivo model. For this, mice from each treatment group were sacrificed at different time points, allowing to analyze the kinetics of cytokine secretion along the course of EAE. The fold increase in cytokine secretion following MOG35–55 restimulation was calculated by comparison to the negative control (unstimulated splenocytes). Results are expressed as the mean fold increase from three replicates ± standard deviation. In case the cytokine concentration was below the detection limit in the negative control, no fold increase could be calculated. Mean fold increases of > 2 log are indicated in bold. EP electroporation, p.i. post induction, B.D.L. protein concentration below detection limit in the MOG35–55-restimulated condition Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 14 of 20 treatment on inflammatory lesion load within the Discussion central nervous system (CNS). For this, MR images Although the exact cause of the breach in tolerance to- from three mice per treatment group were analyzed ward myelin in MS is not yet fully understood, a regula- for the presence of white matter hyperintense areas tory role for DC has been suggested [37]. Indeed, driven reflecting inflammatory lesions. The median number by the important function of DC in central and periph- of hyperintense spots was 4.0 following treatment of eral tolerance [38–42], it has been demonstrated that EAE mice with MOG mRNA-electroporated tolDC, 5.3 DC in MS patients are in a pro-inflammatory state and with MOG35–55-pulsed tolDC, 7.0 with non-antigen- have an altered, migratory phenotype as compared to loaded tolDC, and 7.7 with PBS (Fig. 6). Additionally, the DC from healthy individuals [37, 43]. In doing so, they presence of demyelinated lesions on cryosections from may disturb the balance between immunity and toler- spinal cord regions with a high lesion load on MRI scan ance. These findings underline the important role of DC was confirmed using hematoxylin-eosin staining (for cell in the pathogenesis of MS and substantiate the choice infiltration) and Luxol Fast Blue staining (for demyelin- for modulated DC as promising players for an antigen- ation) (Fig. 7). specific cell-based approach for the treatment of MS. Fig. 6 Evaluation of inflammatory lesion load within the spinal cord of tolDC-treated and PBS-treated mice using ex vivo MRI imaging. PFA-fixed spinal cords from three representative mice per treatment group were analyzed using MRI for the presence of white matter hyperintense lesions. a Representative MRI of spinal cord with hyperintense white matter spots marked with a red arrow. Two representative axial slices per treatment group are shown. b The total number of hyperintense white matter spots along the entire spinal cord was quantified as a measure of lesion load in three mice per treatment group. Results are presented as individual scores for hyperintense spots with median Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 15 of 20 Fig. 7 Hematoxylin-eosin (left panel) and Luxol Fast Blue (right panel) staining of spinal cord sections from mice either treated with PBS, with non-antigen-loaded tolDC, with MOG35–55-pulsed tolDC or with MOG mRNA-electroporated tolDC. For this, spinal cords were cryosected longitudinally at 20%, 40%, and 80% of depth. Arrows indicate the location of spinal cord demyelinization lesions. In the right upper corner of each panel group, the number of sections on which lesions could be identified is mentioned Irrespective of the cell type used for therapeutic vaccin- EAE models, including PLP139–151-induced EAE [45], it is ation of autoimmune diseases, whether or not to target the less investigated in chronic MOG35–55 EAE. In this study, cell-based approach toward disease-associated antigens re- we demonstrated the presence of a targeted MOG35–55 im- mains to be elucidated. Previously, others demonstrated the mune response associated with the first clinical signs of dis- need for auto-antigen loading of tolDC for the treatment of ease, followed by both intra- and intermolecular epitope EAE [4, 5, 44], collagen-induced arthritis [7, 8], and auto- spreading in PBS-treated EAE mice in later stages of dis- immune thyroiditis [9]. In this study, we evaluated whether ease. Similarly, Van Zwam et al. detected intermolecular mRNA electroporation is an efficient method to induce epitope spreading toward PLP56–70 and myelin-associated myelin expression and presentation by DC. We demon- glycoprotein (MAG)97–112 in splenocytes obtained from strated that MOG protein expression could be detected up MOG35–55-induced C57BL/6 EAE mice [46]. Interestingly, to 4 days following mRNA electroporation. Moreover, in our hands, initiation of the epitope spreading cascade MOG mRNA-electroporated DC was able to stimulate could be prevented by early treatment with tolDC. Indeed, MOG35–55-reactive splenocytes, indicating antigen-present- administration of tolDC when a targeted MOG35–55 re- ing capacity of MOG mRNA-electroporated DC. The use sponse was present abrogated epitope spreading, reduced of mRNA electroporation as an antigen-loading strategy MOG35–55-specific splenocyte proliferation, and inhibited promises to overcome several key issues associated with the the pro-inflammatory response toward MOG35–55. Strik- selection of target antigens, including the need for prior ingly, also non-antigen-loaded tolDC were able to induce knowledge of culprit antigens and the dynamic nature of these immunological alterations. However, in the absence antigen reactivity in autoimmunity, i.e., epitope spreading. of antigen, they failed to result in a clinical benefit. This Whereas epitope spreading is well-described in relapsing discrepancy between clinical results and immunological Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 16 of 20 findings, as demonstrated by us and others [4, 5, 8], might the demonstration of MOG presentation in Sig-extracel- point toward other antigen-specific mechanisms of toler- lular MOG-LAMP mRNA-electroporated mDC and ance induction responsible for clinical benefit. For instance, tolDC indirectly implies the presence of MOG expres- it has been demonstrated that amelioration of EAE follow- sion. This is in line with previous studies by other ing administration of MOG-loaded tolDC was related to authors [51], who were equally unable to directly dem- expansion of splenic regulatory T cells (Treg) [5, 6], in onstrate MOG expression in DC following transfection contrast to treatment with unpulsed tolDC [5]. In our with nucleic acids encoding MOG, even though MOG hands, however, no differences in number of Treg, regula- presentation was present. tory B cells (Breg), or natural killer cells (NK) in the spleen Additionally, we explored the effect of the different population between the different treatment groups could tolDC treatments on CNS inflammation. For this, the in- be detected (Additional file 6: Figure S6). Likewise, Stoop et flammatory lesion load was evaluated using MRI in an al. demonstrated that the inhibition of arthritis severity and explorative experiment with spinal cord tissues from progression following treatment with type II collagen- three mice per treatment group. This technique previ- pulsed tolDC was not accompanied by an expansion of ously allowed to visualize inflammation-related spinal splenic Foxp3-positive Tregs, but by an increase in the cord edema by increase of the T2 signal in EAE [52] and number of IL-10-producing T cells and a decrease in IL- spinal cord injury [34]. In this study, our preliminary 17-producing T cells [8]. The variety of methods and proto- data of the MRI analysis of spinal cord tissue revealed a cols available to detect antigen-specific T cell responses has trend to reduction in the number of hyperintense spots complicated the interpretation of the results obtained from in the spinal cord white matter in a small subgroup of immunomonitoring approaches. Despite the overwhelming tolDC-treated mice. Moreover, we confirmed the pres- use of various immune assays, reported results are often ence of demyelinated lesions by histological analysis of met with skepticism caused mainly by two reasons: (i) high spinal cord regions with a high lesion load on MRI variability of results from the same laboratories and/or scans. This is in line with previous research showing a among different laboratories, and (ii) the lack of strong correlation between MRI and histological lesions prioritization to report standardization, validation, and [34]. Similar findings were described by Duraes et al., training strategies as well as assay criteria by the laborator- who demonstrated that clinical improvement of EAE fol- ies conducting immune testing. Hence, to further elucidate lowing transfer of MOG35–55-loaded plasmacytoid DC the underlying mechanisms that mediate tolDC-induced (pDC) was associated with a reduction of histological tolerance in vivo by MOG antigen-loaded tolDC or to dem- spinal cord inflammatory foci [53]. This protective effect onstrate proof-of-concept of administering antigen-loaded was mediated by recruitment and modulation of en- tolDC, harmonization strategies for the immune assay pro- dogenous pDC and by inhibition of encephalitogenic T tocols should be aimed for allowing meta-analysis for future cells within the spinal cord following transfer and CNS experiments [47, 48]. migration of MOG antigen-loaded pDC, but was com- Following electroporation with Sig-MOG-LAMP pletely abolished upon transfer of unloaded pDC [53]. mRNA and Sig-extracellular MOG-LAMP mRNA, ex- Therefore, both the migratory capacity to the inflamed pression of MOG protein could only be demonstrated CNS and disease antigen presentation appear to be cru- with the latter mRNA construct and only in the K562 cial for the clinical effect of tolerogenic DC-based cell line, not in murine dendritic cells, even though suc- therapies. Although we did not perform tracking experi- cessful transfection was demonstrated using RT-qPCR. ments to study the migration of tolDC following admin- Possibly, this is due to differences in tertiary structure of istration, migration to the CNS can be expected based the translated protein in K562 cells compared to DC, on findings from Mansilla et al., who detected a transi- since the Z2 anti-MOG antibody recognizes a conform- tory signal from murine NIR815 dye-labeled, vitD3- ational epitope [49] and therefore differences in protein treated MOG40–55-loaded tolDC in the brain 24 h fol- folding could interfere with proper binding of the anti- lowing administration [5]. Overall, our preliminary ob- body. Additionally, the Sig-LAMP mRNA sequence, servations together with the findings by others raise the flanking the MOG sequence, guides the MOG protein hypothesis that tolDC treatment mediates suppression of into the lysosomal pathway, hereby stimulating MHC CNS inflammation in EAE. Since the need for antigen class II presentation of MOG antigenic peptides [29]. loading of tolDC for clinical benefit on the course of This lysosomal pathway is particularly active in DC fol- EAE was not reflected by a clear peripheral immuno- lowing stimulation with a maturation stimulus [50]. logical mechanism, these findings may therefore point Hence, we hypothesize that the MOG protein is rapidly toward an immunological effect of antigen-loaded tolDC processed and presented by DC, hampering the demon- within the CNS. Hence, for future experiments, we sug- stration of MOG protein expression following Sig-MOG- gest to focus on immunomonitoring within the CNS, LAMP mRNA electroporation in DC. In our opinion, e.g., the modulation of autoreactive T cells in spinal cord Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 17 of 20 lesions, to further elucidate the underlying mechanisms first signs of disease in the EAE model might be redun- that mediate tolDC-induced tolerance in vivo by MOG dant, given the fact that administration of the cell-based antigen-loaded tolDC. vaccine was performed when only a focused response to- In this study, we developed and used a new nonlinear ward the disease-inducing epitope could be observed. In statistical approach to model the clinical course of EAE. the human setting, however, myelin reactivity is more This method has distinct advantages over currently used complex, with both disease-initiating epitope(s) and sub- methods [54]. Firstly, and in contrast to the Kruskal- sequent epitope spreading cascade being strongly pa- Wallis test that is classically used for the statistical ana- tient-dependent [13, 15]. Therefore, one can hypothesize lysis of EAE data, there is no need to reduce the daily that loading of tolDC with one single epitope, although EAE scores to one summary score, such as mean, me- sufficient in the experimentally focused murine setting, dian, or area under the curve, to address the research may be insufficient in the human setting. Consequently, question. Our model is capable of addressing differences mRNA electroporation will probably reach its full value in the extent of disease among treatment groups without as antigen-loading technique in the multi-epitope setting combining the daily EAE scores in one summarizing of MS. For future research, we suggest additional score. Consequently, no information on the disease pro- preclinical assessments to validate our findings using files is lost. Secondly, the model properly accounts for relapsing-remitting EAE models which have been dem- incompleteness in the data under the assumption of onstrated to be associated with extensive epitope spread- missing at random (MAR), making deletion of mice ing, e.g., the PLP139–151 C57BL/6 EAE model [45] and from the analysis unnecessary. Lastly, the model has the MBP SJL/J EAE model [55]. In doing so, the direct good predictive ability of the individual disease profiles effect on epitope spreading of treatment with tolDC, and provides a natural interpretation of the results. electroporated with single or multiple myelin mRNA Nevertheless, there are also some drawbacks of this constructs, could be evaluated. Confirmation of the clin- model, i.e., (1) the assumption of normality and homo- ical benefit of mRNA-electroporated tolDC in various scedasticity underlying the model are not met; (2) the murine models of EAE, together with additional in- predicted scores might potentially be larger than 6, since depth evaluation of the immunological mechanisms the clinical score is treated here as a continuous variable; underlying this effect, would further substantiate the and (3) although the model performs well in predicting value of mRNA-electroporated tolDC as a tolerance-in- the average clinical score, it does not reflect small fluctu- ducing treatment approach. ations in the clinical score. However and taken together, our new nonlinear statistical approach is more appropri- Conclusion ate to analyze the dynamic profile of EAE than the Krus- In conclusion, we validated the mRNA electroporation kal-Wallis test. For the sake of completeness, we did technique in vitro as an efficient tool to generate mye- compare the medians between the different treatment lin-presenting tolDC and demonstrated in vivo a clinical groups of EAE mice in our study using the Kruskal-Wal- benefit of MOG mRNA-electroporated tolDC treatment lis test and reached the same statistical conclusions as in EAE mice. This protective effect was accompanied by with our new nonlinear model (data not shown). a decrease in the MOG35–55-specific pro-inflammatory There were a few limitations to this study. First of all, response in the peripheral immune system and was most of the ex vivo experiments have been performed in likely driven by suppression of CNS inflammation. Add- a small subgroup of animals. Secondly, although animals itionally, we extensively characterized the cytokine re- were observed for a considerable period following dis- sponse and epitope spreading cascade in the chronic ease induction, an even longer period of follow-up could MOG35–55-induced EAE model. This study paves the provide more insight into the long-term effect of tolDC way for additional research into mRNA-electroporated treatment on the course of EAE. Thirdly, whereas a su- tolDC treatment in the human setting of MS, with perior clinical effect of MOG antigen-loaded tolDC over patient-specific multi-epitope disease induction and non-antigen-loaded tolDC was demonstrated in this progression. This will allow fully evaluating and appreci- study, MOG35–55-pulsed tolDC and MOG mRNA-elec- ating the value of mRNA electroporation as antigen- troporated tolDC were equally effective. Conceptually, loading technique with the potential of covering all the hypothesized advantages of mRNA electroporation potential epitopes and tackling epitope spreading. as an antigen loading technique—i.e., the induction of presentation of naturally processed epitopes in a HLA- independent manner, with the potential to tackle epitope Additional files spreading—will likely be most evident in the multi-epi- tope setting of MS. Indeed, presentation of additional Additional file 1: Figure S1. Graphical overview of the culture protocol for murine bone marrow-derived dendritic cells. (PDF 271 kb) epitopes by MOG mRNA-electroporated tolDC at the Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 18 of 20 Additional file 2: Figure S2. Graphical overview of the protein structure Ethics approval and consent to participate of murine MOG. (PDF 357 kb) The procedures described in this study were approved by the Ethics Committee for Animals of the University of Antwerp (study number 2014-93 Additional file 3: Figure S3. Nucleotide sequence of the Sig-full length and 2017-08) or by the Ethics Committee on Animal Experimentation of the MOG-LAMP-3 and Sig-extracellular MOG-LAMP-3 constructs. (PDF 385 kb) Germans Trias i Pujol Research Institute and by the Generalitat de Catalunya Additional file 4: Figure S4. Graphical overview of the experimental (protocol number 9469). All experiments were performed in strict accordance set-up of the in vivo experiment. (PDF 279 kb) with EU and governmental regulations. Additional file 5: Figure S5. Representation of individual clinical scores. (PDF 349 kb) Consent for publication Not applicable. Additional file 6: Figure S6. Flow cytometric evaluation of the proportion of Treg, Breg and NK cells in splenocytes. (PDF 289 kb) Competing interests Additional file 7: Table S1. Amino acid sequences of the myelin This work was partially supported by research funding from Roche Belgium. peptides used for screening for inter- and intramolecular epitope The authors declare no further conflict of interest. spreading. (PDF 317 kb) Additional file 8: Table S2. Clinical data of MOG35–55 EAE mice. Author details 1 (PDF 393 kb) Laboratory of Experimental Hematology, Faculty of Medicine and Health Sciences, Vaccine and Infectious Disease Institute (VaxInfectio), University of Antwerp, Antwerp University Hospital (UZA), Wilrijkstraat 10, 2650 Edegem, Abbreviations Belgium. 2Division of Neurology, Antwerp University Hospital, Edegem, BDL: Below detection limit; CCR5: C-C motif chemokine receptor 5; Belgium. 3Division of Immunology, Germans Trias i Pujol University Hospital CNRQ: Calibrated normalized relative quantity; CNS: Central nervous and Research Institute, Campus Can Ruti, Badalona, Spain. 4Department of system; cpm: Counts per minute; DC: Dendritic cell; EAE: Experimental Cellular Biology, Physiology and Immunology, Universitat Autònoma de autoimmune encephalomyelitis; eGFP: Enhanced green fluorescent Barcelona, Bellaterra, Cerdanyola del Vallès, Spain. 5Center for Cell Therapy protein; EP: Electroporated; FBS: Fetal bovine serum; and Regenerative Medicine, Antwerp University Hospital, Edegem, Belgium. 6 GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; GM- Bio-Imaging Lab, University of Antwerp, Antwerp, Belgium. 7Center for CSF: Granulocyte-macrophage colony-stimulating factor; HLA: Human Statistics, I-Biostat, Hasselt University, Diepenbeek, Belgium. 8Centre for leukocyte antigen; iDC: Immature dendritic cell; IFN-γ: Interferon Health Economics Research and Modelling Infectious Diseases (CHERMID), gamma; IL: Interleukin; IMDM: Iscove’s modified Dulbecco’s medium; Vaccine and Infectious Disease Institute (Vaxinfectio), University of Antwerp, LAMP: Lysosomal-associated membrane protein; Antwerp, Belgium. 9L-BioStat, I-BioStat, KU Leuven, Leuven, Belgium. LPS: Lipopolysaccharide; MAR: Missing at random; MBP: Myelin basic 10 Laboratory of Medical Microbiology, Faculty of Medicine and Health protein; mDC: Mature dendritic cell; MOG: Murine oligodendrocyte Sciences, Vaccine and Infectious Disease Institute (VaxInfectio), University of glycoprotein; MRI: Magnetic resonance imaging; MS: Multiple sclerosis; Antwerp, Antwerp, Belgium. 11Born Bunge Institute, Translational MSD: Meso Scale Discovery; NS: Not significant; PBS: Phosphate-buffered Neurosciences, Faculty of Medicine and Health Sciences, University of saline; PFA: Paraformaldehyde; pi: Post induction; PLP: Proteolipid protein; Antwerp, Antwerp, Belgium. PMA: Phorbol myristate acetate; RPMI: Roswell Park Memorial Institute medium; SD: Standard deviation; SFC: Spot-forming cells; SI: Stimulation index; Received: 3 October 2018 Accepted: 9 July 2019 Sig: Lysosomal-targeting signal sequence; TNF: Tumor necrosis factor; tolDC: Tolerogenic dendritic cell; VEGF: Vascular endothelial growth factor A; VitD3: 1α,25-Dihydroxyvitamin D3 References 1. Hemmer B, Kerschensteiner M, Korn T. Role of the innate and adaptive Acknowledgements immune responses in the course of multiple sclerosis. Lancet Neurol. The authors would like to thank Dr. Gudrun Koppen (Flemish Institute for 2015;14(4):406–19. Technological Research) and Gitte Slingers (Hasselt University) for their 2. Hohlfeld R, Dornmair K, Meinl E, Wekerle H. The search for the target antigens guidance in performing the Meso Scale Discovery (MSD) analysis. of multiple sclerosis, part 1: autoreactive CD4+ T lymphocytes as pathogenic Additionally, we would like to thank Dr. Isabel Pintelon, Dr. Dominique De effectors and therapeutic targets. Lancet Neurol. 2016;15(2):198–209. Rijck, Dr. Karen Sterck, and Dr. Sophie Thys (Cell Biology and Histology—Core 3. Rommer PS, Zettl UK. Managing the side effects of multiple sclerosis Facility Antwerp Centre for Advanced Microscopy, University of Antwerp) for therapy: pharmacotherapy options for patients. Expert Opin Pharmacother. their help with performing the histological experiments. 2018;19(5):483–98. 4. Mansilla MJ, Contreras-Cardone R, Navarro-Barriuso J, Cools N, Berneman Z, Authors’ contributions Ramo-Tello C, et al. Cryopreserved vitamin D3-tolerogenic dendritic cells JD, ZNB, and NC conceived the idea. JD, MJM, PP, EMC, ZNB, and NC planned the pulsed with autoantigens as a potential therapy for multiple sclerosis experiments. JD, MJM, MDL, WPL, JBN, HDR, IN, Ja.D., and IW carried out the patients. J Neuroinflammation. 2016;13(1):113. experiments. JD, AJK, JVA, VV, AVDL, and MV performed the MRI experiments. ZP, 5. Mansilla MJ, Selles-Moreno C, Fabregas-Puig S, Amoedo J, Navarro-Barriuso GM, and NH developed the non-linear statistical model. JD, MJM, PP, EMC, ZNB, J, Teniente-Serra A, et al. Beneficial effect of tolerogenic dendritic cells and NC contributed to the interpretation of the results. All authors provided critical pulsed with MOG autoantigen in experimental autoimmune feedback and helped shape the research, analysis, and manuscript. All authors read encephalomyelitis. CNS Neurosci Ther. 2015;21(3):222–30. and approved the final manuscript. 6. Zhou Y, Leng X, Li H, Yang S, Yang T, Li L, Xiong Y, Zou Q, Liu Y, Wang Y. Tolerogenic dendritic cells induced by BD750 ameliorate proinflammatory T cell responses and experimental autoimmune encephalitis in mice. Mol Funding Med. 2017;23:204–14. This work was supported by the Methusalem Funding Program from the 7. Yang J, Yang Y, Ren Y, Xie R, Zou H, Fan H. A mouse model of adoptive University of Antwerp, by an applied biomedical research project of the Institute immunotherapeutic targeting of autoimmune arthritis using allo-tolerogenic for the Promotion of Innovation by Science and Technology in Flanders (IWT- dendritic cells. PLoS One. 2013;8(10):e77729. TBM 140191), and by the Belgian Charcot Foundation. Furthermore, the authors 8. Stoop JN, Harry RA, von Delwig A, Isaacs JD, Robinson JH, Hilkens received research funding from Roche Belgium. Judith Derdelinckx holds a PhD CM. Therapeutic effect of tolerogenic dendritic cells in established fellowship from the Research Foundation Flanders (FWO). collagen-induced arthritis is associated with a reduction in Th17 responses. Arthritis Rheum. 2010;62(12):3656–65. Availability of data and materials 9. Verginis P, Li HS, Carayanniotis G. Tolerogenic semimature dendritic cells The datasets used and/or analyzed during the current study are available suppress experimental autoimmune thyroiditis by activation of from the corresponding author on reasonable request. thyroglobulin-specific CD4+CD25+ T cells. J. Immunol. 2005;174(11):7433–9. Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 19 of 20 10. Grau-Lopez L, Raich D, Ramo-Tello C, Naranjo-Gomez M, Davalos A, 32. Piddlesden SJ, Lassmann H, Zimprich F, Morgan BP, Linington C. The Pujol-Borrell R, et al. Specific T-cell proliferation to myelin peptides in demyelinating potential of antibodies to myelin oligodendrocyte relapsing-remitting multiple sclerosis. Eur J Neurol. 2011;18(8):1101–4. glycoprotein is related to their ability to fix complement. Am J Pathol. 11. Saez-Torres I, Brieva L, Espejo C, Barrau MA, Montalban X, Martinez-Caceres 1993;143(2):555–64. EM. Specific proliferation towards myelin antigens in patients with multiple 33. Faul F, Erdfelder E, Lang AG, Buchner A. G*power 3: a flexible statistical sclerosis during a relapse. Autoimmunity. 2002;35(1):45–50. power analysis program for the social, behavioral, and biomedical sciences. 12. Greer JM. Autoimmune T-cell reactivity to myelin proteolipids and Behav Res Methods. 2007;39(2):175–91. glycolipids in multiple sclerosis. Mult Scler Int. 2013;2013:151427. 34. Noristani HN, Lonjon N, Cardoso M, Le Corre M, Chan-Seng E, Captier G, et 13. Goebels N, Hofstetter H, Schmidt S, Brunner C, Wekerle H, Hohlfeld R. al. Correlation of in vivo and ex vivo (1) H-MRI with histology in two Repertoire dynamics of autoreactive T cells in multiple sclerosis patients and severities of mouse spinal cord injury. Front Neuroanat. 2015;9:24. healthy subjects: epitope spreading versus clonal persistence. Brain. 2000; 35. Rorden C, Brett M. Stereotaxic display of brain lesions. Behav Neurol. 2000; 123(Pt 3):508–18. 12(4):191–200. 14. Tuohy VK, Kinkel RP. Epitope spreading: a mechanism for progression of 36. Molenberghs G, Kenward MG. Missing data in clinical studies. New autoimmune disease. Arch Immunol Ther Exp. 2000;48(5):347–51. York: Wiley; 2007. 15. Tuohy VK, Yu M, Yin L, Kawczak JA, Johnson JM, Mathisen PM, et al. The 37. Nuyts AH, Lee WP, Bashir-Dar R, Berneman ZN, Cools N. Dendritic cells in epitope spreading cascade during progression of experimental autoimmune multiple sclerosis: key players in the immunopathogenesis, key players for encephalomyelitis and multiple sclerosis. Immunol Rev. 1998;164:93–100. new cellular immunotherapies? Multiple Sclerosis (Houndmills, Basingstoke, 16. Quintana FJ, Patel B, Yeste A, Nyirenda M, Kenison J, Rahbari R, et al. England). 2013;19(8):995–1002. Epitope spreading as an early pathogenic event in pediatric multiple 38. Hilkens CM, Isaacs JD, Thomson AW. Development of dendritic cell-based sclerosis. Neurology. 2014;83(24):2219–26. immunotherapy for autoimmunity. Int Rev Immunol. 2010;29(2):156–83. 17. Yu M, Johnson JM, Tuohy VK. A predictable sequential determinant 39. Mahnke K, Schmitt E, Bonifaz L, Enk AH, Jonuleit H. Immature, but not spreading cascade invariably accompanies progression of experimental inactive: the tolerogenic function of immature dendritic cells. Immunol Cell autoimmune encephalomyelitis: a basis for peptide-specific therapy after Biol. 2002;80(5):477–83. onset of clinical disease. J Exp Med. 1996;183(4):1777–88. 40. Oh J, Shin JS. The role of dendritic cells in central tolerance. Immune 18. Willcox N, Baggi F, Batocchi AP, Beeson D, Harcourt G, Hawke S, et al. network. 2015;15(3):111–20. Approaches for studying the pathogenic T cells in autoimmune patients. 41. Steinman RM, Nussenzweig MC. Avoiding horror autotoxicus: the Ann N Y Acad Sci. 1993;681:219–37. importance of dendritic cells in peripheral T cell tolerance. Proc Natl Acad 19. Sweenie CH, Mackenzie KJ, Rone-Orugboh A, Liu M, Anderton SM. Distinct T Sci U S A. 2002;99(1):351–8. cell recognition of naturally processed and cryptic epitopes within the 42. Westerberg LS, Klein C, Snapper SB. Breakdown of T cell tolerance immunodominant 35-55 region of myelin oligodendrocyte glycoprotein. J and autoimmunity in primary immunodeficiency—lessons learned Neuroimmunol. 2007;183(1–2):7–16. from monogenic disorders in mice and men. Curr Opin Immunol. 20. Peakman M, Stevens EJ, Lohmann T, Narendran P, Dromey J, 2008;20(6):646–54. Alexander A, et al. Naturally processed and presented epitopes of the 43. Thewissen K, Nuyts AH, Deckx N, Van Wijmeersch B, Nagels G, D'Hooghe M, islet cell autoantigen IA-2 eluted from HLA-DR4. J Clin Invest. 1999; et al. Circulating dendritic cells of multiple sclerosis patients are 104(10):1449–57. proinflammatory and their frequency is correlated with MS-associated 21. Anderton SM, Viner NJ, Matharu P, Lowrey PA, Wraith DC. Influence of a genetic risk factors. Multiple Sclerosis (Houndmills, Basingstoke, England). dominant cryptic epitope on autoimmune T cell tolerance. Nat Immunol. 2014;20(5):548–57. 2002;3(2):175–81. 44. Zhou F, Ciric B, Zhang GX, Rostami A. Immune tolerance induced by 22. Adorini L. Immunodominance. In: Delves PJ, editor. Encyclopedia of intravenous transfer of immature dendritic cells via up-regulating numbers immunology. 2nd ed. Oxford: Elsevier; 1998. p. 1290–2. of suppressive IL-10(+) IFN-gamma(+)-producing CD4(+) T cells. Immunol 23. Berthelot L, Laplaud DA, Pettre S, Ballet C, Michel L, Hillion S, et al. Blood Res. 2013;56(1):1–8. CD8+ T cell responses against myelin determinants in multiple sclerosis and 45. McRae BL, Vanderlugt CL, Dal Canto MC, Miller SD. Functional evidence for healthy individuals. Eur J Immunol. 2008;38(7):1889–99. epitope spreading in the relapsing pathology of experimental autoimmune 24. Martin R, Jaraquemada D, Flerlage M, Richert J, Whitaker J, Long EO, et al. encephalomyelitis. J Exp Med. 1995;182(1):75–85. Fine specificity and HLA restriction of myelin basic protein-specific cytotoxic 46. van Zwam M, Huizinga R, Heijmans N, van Meurs M, Wierenga-Wolf AF, T cell lines from multiple sclerosis patients and healthy individuals. J Melief MJ, et al. Surgical excision of CNS-draining lymph nodes reduces Immunol. 1990;145(2):540–8. relapse severity in chronic-relapsing experimental autoimmune 25. Van Tendeloo VF, Ponsaerts P, Lardon F, Nijs G, Lenjou M, Van Broeckhoven encephalomyelitis. J Pathol. 2009;217(4):543–51. C, et al. Highly efficient gene delivery by mRNA electroporation in human 47. McNeil LK, Price L, Britten CM, Jaimes M, Maecker H, Odunsi K, et al. A hematopoietic cells: superiority to lipofection and passive pulsing of mRNA harmonized approach to intracellular cytokine staining gating: results and to electroporation of plasmid cDNA for tumor antigen loading of from an international multiconsortia proficiency panel conducted by dendritic cells. Blood. 2001;98(1):49–56. the Cancer immunotherapy consortium (CIC/CRI). Cytometry A. 2013; 26. Van Driessche A, Ponsaerts P, Van Bockstaele DR, Van Tendeloo VF, Berneman 83(8):728–38. ZN. Messenger RNA electroporation: an efficient tool in immunotherapy and 48. Britten CM, Gouttefangeas C, Welters MJ, Pawelec G, Koch S, Ottensmeier C, stem cell research. Folia Histochem Cytobiol. 2005;43(4):213–6. et al. The CIMT-monitoring panel: a two-step approach to harmonize the 27. Van Tendeloo VF, Van de Velde A, Van Driessche A, Cools N, Anguille S, enumeration of antigen-specific CD8+ T lymphocytes by structural and Ladell K, et al. Induction of complete and molecular remissions in acute functional assays. Cancer Immunol Immunother. 2008;57(3):289–302. myeloid leukemia by Wilms' tumor 1 antigen-targeted dendritic cell 49. Breithaupt C, Schafer B, Pellkofer H, Huber R, Linington C, Jacob U. vaccination. Proc Natl Acad Sci U S A. 2010;107(31):13824–9. Demyelinating myelin oligodendrocyte glycoprotein-specific autoantibody 28. Anguille S, Van de Velde AL, Smits EL, Van Tendeloo VF, Juliusson G, Cools response is focused on one dominant conformational epitope region in N, et al. Dendritic cell vaccination as postremission treatment to prevent or rodents. J Immunol. 2008;181(2):1255–63. delay relapse in acute myeloid leukemia. Blood. 2017;130(15):1713–21. 50. Gatti E, Pierre P. Understanding the cell biology of antigen presentation: the 29. Van Nuffel AM, Benteyn D, Wilgenhof S, Pierret L, Corthals J, Heirman C, et al. dendritic cell contribution. Curr Opin Cell Biol. 2003;15(4):468–73. Dendritic cells loaded with mRNA encoding full-length tumor antigens prime 51. de Andrade PB, Fraefel C, Hilbe M, Ackermann M, Dresch C. CD4+ and CD8+ T cells in melanoma patients. Mol Ther. 2012;20(5):1063–74. Transcriptional targeting of DCs with lentiviral vectors induces 30. Nair SK, Boczkowski D, Morse M, Cumming RI, Lyerly HK, Gilboa E. Induction antigen-specific tolerance in a mouse model of multiple sclerosis. of primary carcinoembryonic antigen (CEA)-specific cytotoxic T lymphocytes Gene Ther. 2013;20(5):556–66. in vitro using human dendritic cells transfected with RNA. Nat Biotechnol. 52. Mori Y, Murakami M, Arima Y, Zhu D, Terayama Y, Komai Y, et al. 1998;16(4):364–9. Early pathological alterations of lower lumbar cords detected by 31. Derdelinckx J, Berneman ZN, Cools N. GMP-grade mRNA electroporation of ultrahigh-field MRI in a mouse multiple sclerosis model. Int Immunol. dendritic dells for dlinical use. Methods Mol Biol. 2016;1428:139–50. 2014;26(2):93–101. Derdelinckx et al. Journal of Neuroinflammation (2019) 16:167 Page 20 of 20 53. Duraes FV, Lippens C, Steinbach K, Dubrot J, Brighouse D, Bendriss-Vermare N, et al. pDC therapy induces recovery from EAE by recruiting endogenous pDC to sites of CNS inflammation. J Autoimmun. 2016;67:8–18. 54. Fleming KK, Bovaird JA, Mosier MC, Emerson MR, LeVine SM, Marquis JG. Statistical analysis of data from studies on experimental autoimmune encephalomyelitis. J Neuroimmunol. 2005;170(1–2):71–84. 55. Voskuhl RR, Farris RW 2nd, Nagasato K, McFarland HF, Dalcq MD. Epitope spreading occurs in active but not passive EAE induced by myelin basic protein. J Neuroimmunol. 1996;70(2):103–11. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.