sign in sign up
<<
Home , NADPH oxidase

NADPH Modifies Patterns of MHC Class II−Restricted Epitopic Repertoires through Redox Control of Antigen Processing

This information is current as Euan R. O. Allan, Pankaj Tailor, Dale R. Balce, Payman of September 28, 2021. Pirzadeh, Neil T. McKenna, Bernard Renaux, Amy L. Warren, Frank R. Jirik and Robin M. Yates J Immunol 2014; 192:4989-5001; Prepublished online 28 April 2014;

doi: 10.4049/jimmunol.1302896 Downloaded from http://www.jimmunol.org/content/192/11/4989

Supplementary http://www.jimmunol.org/content/suppl/2014/04/27/jimmunol.130289 Material 6.DCSupplemental http://www.jimmunol.org/ References This article cites 73 articles, 30 of which you can access for free at: http://www.jimmunol.org/content/192/11/4989.full#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision by guest on September 28, 2021

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2014 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

NADPH Oxidase Modifies Patterns of MHC Class II–Restricted Epitopic Repertoires through Redox Control of Antigen Processing

Euan R. O. Allan,*,† Pankaj Tailor,*,† Dale R. Balce,*,† Payman Pirzadeh,*,†,‡ Neil T. McKenna,*,† Bernard Renaux,*,† Amy L. Warren,x Frank R. Jirik,† and Robin M. Yates*,†

The chemistries within of APCs mediate microbial destruction as well as generate peptides for presentation on MHC class II. The antimicrobial effector NADPH oxidase (NOX2), which generates within maturing phagosomes, has also been shown to regulate activities of cysteine through modulation of the lumenal redox potential. Using real-time analyses of lumenal microenvironmental parameters, in conjunction with hydrolysis pattern assessment of phagocytosed , we dem- Downloaded from onstrated that NOX2 activity not only affects levels of phagosomal proteolysis as previously shown, but also the pattern of proteolytic digestion. Additionally, it was found that NOX2 deficiency adversely affected the ability of bone marrow–derived macrophages, but not dendritic cells, to process and present the I-Ab–immunodominant peptide of the autoantigen myelin oligodendrocyte glyco- (MOG). Computational and experimental analyses indicated that the I-Ab binding region of the immunodominant peptide of MOG is susceptible to cleavage by the NOX2-controlled cysteine cathepsins L and S in a redox-dependent manner. Consistent with these findings, I-Ab mice that were deficient in the p47phox or gp91phox subunits of NOX2 were partially protected from MOG- http://www.jimmunol.org/ induced experimental autoimmune encephalomyelitis and displayed compromised reactivation of MOG-specific CD4+ T cells in the CNS, despite eliciting a normal primary CD4+ T cell response to the inoculated MOG Ag. Taken together, this study demonstrates that the redox microenvironment within the phagosomes of APCs is a determinant in MHC class II repertoire production in a cell-specific and Ag-specific manner, which can ultimately impact susceptibility to CD4+ T cell–driven autoim- mune disease processes. The Journal of Immunology, 2014, 192: 4989–5001.

hagocytosed or endocytosed exogenous Ags must be oligopeptides (1). It has thus become increasingly apparent that proteolytically processed within the phagosomes and endo- tight control over the level of proteolysis within these compart- by guest on September 28, 2021 somes of the APCs to be complexed with MHC class II ments is necessary to generate and preserve the antigenic oligo- P + (MHC-II) and presented to CD4 T cells. Although intralumenal peptides required for productive Ag presentation to T cells (2, 3). proteolysis is a prerequisite for generating oligopeptides of 15– However, beyond the overarching correlation of levels of proteol- 24 aa in length for MHC-II, too much proteolysis is thought to be ysis and processing efficiency, the intricacies of the effect of rel- detrimental to Ag processing efficiencies as it can over-digest the ative protease activities upon the pattern of Ag microdissection remain less evident. Theoretically, because antigenic proteolysis is *Department of Comparative Biology and Experimental Medicine, Faculty of Vet- performed by numerous and diverse proteases (each with differing erinary Medicine, University of Calgary, Calgary, Alberta T2N 4N1, Canada; substrate and cleavage-site preferences), the relative local abun- † Department of Biochemistry and Molecular Biology, Faculty of Medicine, Univer- dance and activity of each protease could affect the pattern of sity of Calgary, Calgary, Alberta T2N 4N1, Canada; ‡Department of Chemistry, Faculty of Science, University of Calgary, Calgary, Alberta T2N 1N4, Canada; and proteolysis and hence the relative availability of different peptides x Department of Ecosystem and Public Health, Faculty of Veterinary Medicine, Uni- for loading onto MHC-II. To complicate this paradigm, identical versity of Calgary, Calgary, Alberta T2N 4N1, Canada CD4+ T cell epitopes must be generated by different APCs (notably Received for publication October 28, 2013. Accepted for publication March 28, thymic epithelial cells, dendritic cells [DCs], B cells, and macro- 2014. phages) in different tissue environments and during different in- This work was supported by the Canadian Institutes of Health Research and the + Multiple Sclerosis Society of Canada. Graduate student support was provided by flammatory states in order for effector CD4 T cells to function. the endMS training network and Alberta Innovates: Health Solutions. Whether this perceived complexity of MHC-II epitope generation Address correspondence and reprint requests to Robin M. Yates, Department of translates into modification of T cell immunity during an immune Comparative Biology and Experimental Medicine, Faculty of Veterinary Medicine, response, or whether relative epitopic abundance within Ag pro- University of Calgary, 3330 Hospital Drive NW, HRIC 4AA10, Calgary, AB T2N 4N1, Canada. E-mail address: [email protected] cessing compartments exceeds saturable levels of MHC-II (such The online version of this article contains supplemental material. that subtle modification of processing patterns would be of no consequence) is largely undetermined. Abbreviations used in this article: BMMf, bone marrow–derived macrophage; BMDC, bone marrow–derived dendritic cell; DC, dendritic cell; EAE, experimental It was recently discovered that the antimicrobial effector autoimmune encephalomyelitis; eGFP, enhanced GFP; Em, emission; Ex, excitation; NADPH oxidase (NOX2) negatively regulates the levels of pro- FSC, forward scatter; HEL, hen egg white ; MHC-II, MHC class II; MOG, myelin oligodendrocyte glycoprotein; NOX2, NADPH oxidase; PDB, Protein Data teolysis within the maturing of macrophages (4, 5) and Bank; qPCR, quantitative real-time PCR; RFU, relative fluorescent unit; rMOG, DCs (3, 6, 7). NOX2, a multiprotein complex expressed in recombinant MOG; ROS, reactive species; SSC, side scatter; TFA, trifluoro- phagocytes, is rapidly assembled on the early phagosomal mem- acetic acid. brane where it oxidizes cytosolic NADPH to convert molecular Copyright Ó 2014 by The American Association of Immunologists, Inc. 0022-1767/14/$16.00 oxygen to superoxide within the phagosomal lumen. Because www.jimmunol.org/cgi/doi/10.4049/jimmunol.1302896 4990 NOX2 MODULATES PATTERNS OF Ag PROCESSING

NOX2 activity dramatically oxidizes the otherwise reductive mi- with murine GM-CSF cDNA, respectively, as previously described (3, 5, croenvironment of the phagosomal lumen, the local cysteine cath- 16, 17). BMMfs were activated overnight (18 h) with rIFN-g (100 U/ml, epsins (which require a reductive environment for activity), but PeproTech) for Ag presentation assays (5). Female mice used for EAE studies were immunized between 8 and 10 wk and age-matched within not the non–cysteine proteases, such as the aspartic cathepsins, are experiments (10, 18, 19). temporarily inactivated (4, 5, 7). Over and above the impact on Protein and peptide synthesis general proteolytic efficiency, we predicted that the impact of

NOX2 on the activities of certain protease subsets within maturing MOG35–55 was synthesized at the University of Calgary (University of phagosomes and endosomes would affect the pattern of proteolytic Calgary’s Peptide Services, Calgary, AB, Canada). Recombinant protein digestion and thus the relative probabilities of specific oligopeptides corresponding to the extracellular domain of MOG1–125 (rMOG) was cloned with a carboxyl terminal 6xHis tag, expressed using Rosetta Blue being generated. NOX2 activity also inhibits the reduction of (DE3) Escherichia coli and purified using Ni-NTA agarose beads (Qiagen). disulfide bonds within the phagosome (5). This could limit the Endotoxin was removed to ,0.005 endotoxin unit/mg (E-Toxate kit, intraphagosomal denaturation of Ags that contain disulfide bonds Sigma-Aldrich) using Endotoxin Affisorbant agarose (polymyxin B and potentially hide vast stretches of an Ag’s polypeptide se- Separopore, bioWORLD, Dublin, OH), according to the manufacturers’ directions. quence from proteolysis. It may also limit the presentation of spe- cific oligopeptides that contain a disulfide linkage in the native Flow cytometry protein (8, 9). Taken together, we hypothesized that NOX2’s control Flow cytometry data were acquired using a FACSCalibur flow cytometer over proteolysis and disulfide reduction would affect not only the (BD Biosciences, Franklin Lakes, NJ) and analyzed using FlowJo software efficiency of Ag processing, but also the repertoire of antigenic v8.6 (Tree Star, Ashland, OR). Unless otherwise indicated, leukocyte peptides available for MHC-II presentation. populations were selected using forward scatter/side scatter (FSC/SSC) and Downloaded from samples were measured with a minimum of 2.5 3 104 counts. Unless Murine experimental autoimmune encephalomyelitis (EAE), a otherwise noted, Abs were purchased from BD Biosciences. commonly used animal model of multiple sclerosis, is primarily driven by myelin Ag-specific autoreactive CD4+ T lymphocytes Fluorometric phagosomal analysis + (10). CD4 T cells must first be activated and expanded in the Three-micrometer IgG-conjugated silica experimental particles were used periphery (primarily by DCs in secondary lymphoid organs) fol- to evaluate intraphagosomal (ROS) generation, proteolysis and cysteine activity, intraphagosomal pH, and lowing immunization with myelin-derived proteins/peptides (e.g., http://www.jimmunol.org/ disulfide reduction in populations of live BMMfs that were prepared as myelin oligodendrocyte glycoprotein [MOG]). Reactivation of + previously described (4, 5, 20–22). Quantification was performed using these expanded effector CD4 T cells by APCs within the sub- a Safire microplate reader (Tecan, Ma¨nnedorf, Switzerland), a FLUOstar arachnoid space triggers a local inflammatory response leading to Optima fluorescent plate reader (BMG Labtech, Ortenberg, Germany), or oligodendrocyte damage and death (11, 12). Further an EnVision multilabel reader (PerkinElmer, Waltham, MA) at 37˚C, at of myelin debris by APCs (particularly macrophages) increases a multiplicity of infection of two to three particles per cell, in an assay buffer containing PBS supplemented with 1 mM calcium chloride, 2.7 mM the presentation of myelin-derived Ags to effector T cells, creating potassium chloride, 0.5 mM magnesium chloride, 5 mM dextrose, and a positive feedback loop that amplifies the inflammatory process 0.25% gelatin (4, 23). Real-time traces are shown in relative fluorescent to a level that is clinically apparent. Hence, the local and recruited units (RFU), which were calculated by dividing the substrate fluorescence macrophages act as the central “gatekeepers” of T cell reactivation (SFRT) by the average calibration fluorescence (CF) for a particular time by guest on September 28, 2021 point (RFU = SFRT/CF) (7). The average rates (taken from 30 to 80 min and ultimately the initiation and propagation of clinical EAE (11, after phagocytosis) of oxidation, proteolysis, and cathepsin activities were 13–15). Thus, EAE presents a good model for studying the gen- determined by the linear portions of the real-time traces and made relative eration of MHC-II–restricted autoepitopes, from both exogenous to the internal controls as indicated in each figure (y = mx + c, where y and endogenous sources, by different APCs in different tissues. indicates relative fluorescence, m indicates gradient, and x indicates time). In the present study, to our knowledge we demonstrate for the Intraphagosomal ROS generation by NOX2 in APCs was assessed via the quantification of particle-restricted H2HFF-OxyBurst substrate (Molecular first time that the redox environment within phagolysosomes Probes) relative to Alexa Fluor 594–succinimidyl ester (calibration fluo- differentially influences local protease activities resulting in 1) rescence) as previously described (4). The intraphagosomal total protease altered patterns of antigenic processing, 2) modified activation of activity and the intraphagosomal hydrolytic activity of cathepsin B/S/L in specific CD4+ T cell clones, and 3) altered susceptibility to au- APCs were assessed by recording the rate of substrate-liberated fluores- cence relative to the calibration fluorescence using the particle-bound toimmune clinical disease perpetrated by a cysteine/cathepsin– fluorogenic DQ green Bodipy albumin (DQ-albumin) (Molecular Probes) susceptible immunodominant peptide epitope. Whereas most and (biotin-LC-Phe-Arg)2-rhodamine 110 (provided by David Russell, therapies for autoimmune disorders target T cell pathways, these Cornell University, Ithaca, NY), respectively (4, 5, 21, 23, 24). findings suggest that autoreactive T cell responses may also be Intraphagosomal pH was measured by monitoring the excitation ratio attenuated through manipulation of Ag processing chemistries. of particle-restricted CFSE (excitation [Ex]/emission [Em] of 490/520 and 450/520 nm), followed by regression to a third-order polynomial stan- dard curve (generated using experimental particles in buffers with known Materials and Methods pH) as previously described (4, 5, 21, 24). Intraphagosomal reductase activity Mice and cells was examined using the rate of fluorescence liberated from Bodipy FL L- cystine (Molecular Probes), a self-quenched cystine-based fluorogenic sub- C57BL/6 (wild-type [WT]) mice and the congenic mouse strains B6.129S- strate conjugated to dextran-coated experimental particles, relative to Alexa Cybbtm1Din/J (Cybb2/2), B6 (Cg)-Ncf1m1J/J (Ncf), B6.Cg-Tg(TcraTcrb) Fluor 594 succinimidyl ester (calibration fluorescence, Molecular Probes) 425Cbn/J (OTII), and C57BL/6-Tg(Tcra2D2,Tcrb2D2)1Kuch/J (2D2) as previously described (4, 5). Phagocytic index was determined using were purchased from The Jackson Laboratory and bred in-house under 3-mm IgG-opsonized, BSA-coated silica experimental particles labeled identical husbandry. Cybb2/2 mice lack the gp91phox (catalytic) subunit of with Alexa Fluor 594 succinimidyl ester given to BMMf monolayers. NOX2, whereas Ncf mice lack the p47phox (activating) subunit of NOX2. Trypan blue (EMD Chemicals) (0.01% in PBS) was used to quench the + OTII mice have a transgenic CD4 TCR specific for OVA323–339 in the Alexa Fluor 594 fluorescence of extracellular particles. The phagocy- context of I-Ab, and 2D2 mice express a transgenic CD4+ TCR (Vb11 tosed particles in three separate images from each well were counted b TCR/Va3.2 TCR) specific for MOG35–55 in the context of I-A . All animal using the 403 (numerical aperture, 0.75) objective on an Olympus IX70 research was performed according to protocols approved by the University fluorescence microscope (Olympus, Center Valley, PA) (4). Rates of of Calgary Animal Care and Use Committee and in accordance with the endocytosis/pinocytosis of Alexa Fluor 488–labeled 70-kDa dextran were Canadian Council of Animal Care. Bone marrow–derived macrophages measured in BMMfs. In brief, APCs were pulsed with medium containing (BMMfs) and bone marrow–derived DCs (BMDCs) were derived from 8- 500 mg/ml Alexa Fluor 488–labeled dextran for 18 h, washed, and chased to 12-wk-old male mice using L929-conditioned media or conditioned in assay buffer for 4 h. Fluorescence was measured using a FLUOstar media derived from the supernatant of Ag8653 melanoma cells transfected Optima fluorescent plate reader. Background was deducted and values The Journal of Immunology 4991 expressed relative to WT for each experiment (n = 3). Representative tative real-time PCR (qPCR), Luminex, or flow cytometry (10, 37). To images were taken using the 403 (numerical aperture, 0.75) objective on specifically measure efficiencies of reactivation of MOG35–55-specific an Olympus IX70 fluorescence microscope. CD4+ T cells by APCs within the CNS, CD4+ T cells from 2D2 mice were expanded and adoptively transferred into preclinical EAE-induced WT and Assessment of pattern of intraphagosomal proteolysis Cybb2/2 mice. In brief, splenocytes were harvested from 2D2 mice and BMMfs derived from WT and Cybb2/2 mice were exposed to IFN-g (100 cultured in T cell medium (RPMI 1640 with 10% FCS, 10 mM 2-ME) U/ml, 18 h) and given 90 min to phagocytose/process Alexa Fluor 488–hen supplemented with 0.5 ng/ml IL-12 (R&D Systems) and 20 mg/ml egg white lysozyme (HEL) (Alexa Fluor 488, Invitrogen; HEL, Sigma- MOG35–55 for 72 h as previously described (38). Following a 24-h rest- Aldrich) covalently coupled to IgG-opsonized 3-mm silica experimental ing period in T cell media without IL-12 and MOG35–55, 2D2 cells were 3 6 particles (multiplicity of infection of three to four beads per BMMf). adoptively transferred (5 10 cells/mouse in 100 ml PBS) into EAE- induced mice via i.p. injection on day 7 after induction. MOG-specific 2D2 BMMfs were lysed in 0.1% trifluoroacetic acid (TFA) in the presence of + protease inhibitor mixture 1 (Calbiochem/Merck, Darmstadt, Germany) CD4 cell recruitment to, and reactivation within, the CNS were deter- and passed through a centrifugal 10-kDa microconcentrator (Nanosep 10K mined on day 10 after induction, following isolation and flow cytometric Omega, Pall Canada, St. Laurent, QC, Canada). Peptides in the filtrate analysis of lumbar spinal cord tissue. were separated by reverse-phase HPLC using a Supelco Ascentis Express Flow cytometry of CNS tissue C18 column (10 cm length, 2.7 mm pore) (Sigma-Aldrich) using a 0–30% acetonitrile gradient with 0.1% TFA. Fluorescently labeled peptides were Spinal cord-infiltrating and resident leukocytes were isolated from EAE detected with a Waters 2475 fluorescent detector (lEx of 488 nm/lEm of mouse spinal cords at 12 d after induction, at the clinical peak of disease, or 520 nm). Relative peptide abundance was determined using area under the preclinical symptoms at 10 d for the harvest of adoptively transferred 2D2 curve for relevant peaks and made proportional to WT samples. CD4+ T cells, via a discontinuous Percoll gradient as previously described (19). Cells were immunostained for CD4 (PerCP), CD8 (PE), CD3 (FITC), In vitro assessment of Ag processing and presentation CD11b (PE), CD45 (PE, FITC), B220 (FITC), and CD11c (FITC) in the Downloaded from To evaluate specific changes to presented antigenic peptide repertoires from following combinations—CD4/8/3 (T cells), CD11b/45 (macrophages/ phagocytosed Ag processed in the presence or absence of an NOX2, four microglia), B220/CD45 (B cells), and CD11b/CD11c (macrophages/ I-Ab–restricted CD4+ T cell hybridomas lines that each responds to different DCs)—and analyzed by flow cytometry. Leukocytes were selected by FSC/ HEL epitopes (Hb1.9 [HEL ], H30.44 [HEL ], H46.13 [HEL ], SSC, and specific populations were identified using Abs against lineage- 20–35 31–47 48–62 specific markers where indicated. Absolute numbers of each cell type were and B04 [HEL74–90]) (provided by Dr. Lars Karlsson, Johnson & Johnson, San Diego, CA) (25, 26) were stably transfected with a pcDNA3 construct calculated from the percentages after gating: absolute number of cells from 3 3 that contained enhanced GFP (eGFP) under the control of a NFAT- leukocyte isolation proportion of leukocytes (FSC/SSC) proportion of

positively stained cells. CD4 and CD8 cells were identified based on CD4 http://www.jimmunol.org/ enhanced IL-2 promoter (pNFATeGFP) as previously described (25–28). + Hence, upon productive Ag presentation with the hybridoma’s cognate I-Ab– and CD8 staining on the CD3 lymphocyte population. Macrophage and microglia were differentiated by high (macrophage) or low (microglia) restricted Ag, the hybridoma would express eGFP. BMMfsfromWT + and Cybb2/2 mice were exposed to 1 mg/ml HEL for 6 h and individual CD45 expression. CD4 T cells were stained for intracellular Foxp3, hybridomas were added and incubated with the BMMfs for 16 h. Pre- IFN-g, and IL-17, and absolute cell numbers were calculated as described above. Additionally, the absolute number of infiltrating spinal cord sentation efficiency of each HEL epitope was determined by measuring b + eGFP expression by flow cytometry and is presented relative to GFP ex- MOG35–55/I-A -positive CD4 T cells after 12 d of EAE with MOG35–55 was examined. Spinal cord cells were isolated (as described above) and pression induced by nonspecific hybridoma activation by 50 ng/ml PMA b b and 1 mg/ml ionomycin (25, 26). were incubated with MOG35–55/I-A (PE) or control tetramers (CLIP/I-A ; Ag presentation efficiencies were also assessed using OVA and MOG as PE) (National Institutes of Health Core Facility, Emory University, Ags utilizing the TCR transgenic mouse models OTII and 2D2, respectively. Atlanta, GA) at 37˚C overnight, washed, and stained with anti-CD4 2 2 + b by guest on September 28, 2021 BMMfs and BMDCs from WT, Cybb / , and Ncf mice were exposed to (PerCP). CD4 cells were gated on MOG35–55/I-A tetramer or control CLIP tetramer. Leukocytes were selected by FSC/SSC, and absolute OVA323–339 (10, 25, and 50 mg/ml), OVA (10, 25, and 50 mg/ml), MOG35–55 + (25 mg/ml), rMOG (10 and 25 mg/ml), myelin sonicates (purified from numbers of tetramer cells were calculated from percentages after gating: 3 mouse brain homogenate as previously described; 25 and 50 mg/ml), and 3- absolute number of cells from leukocyte isolation proportion of leu- kocytes (FSC/SSC) 3 proportion of positively stained CD4 cells 3 pro- mm silica experimental particles with MOG1–125 adsorbed in the presence or b + absence of IgG opsonization (2.5 mg/ml, three to five beads per APC) (4, 5, portion of MOG35–55/I-A tetramer cells. Finally, spinal cord cells were 29) for a period of 6 h (30–35). APCs were washed extensively, and naive isolated (as described above) from adoptive transfer mice 10 d after EAE OTII or 2D2 splenocytes were added and incubated with the APCs for 16 h. induction/3 d after adoptive transfer of 2D2 splenocytes. Leukocytes were selected based on FSC/SSC, and from this population 2D2 CD4+ T cells Presentation efficiencies of OVA323–339 or MOG35–55 were determined by measuring surface expression of the early activation marker (CD69) and IL- were identified based on CD4/Vb11 TCR/Va3.2 TCR expression. The + + activation status of these T cells was assessed by additional identification 2Ra (CD25) on CD4 OTII/2D2 T cells (gated by FSC/SSC, CD4 ) via flow + cytometry (35). Similar to previous reports, pilot experiments revealed that of CD25 and intracellular IFN-g populations. a range of pulsed Ag concentration between 10 and 50 mg/ml gave the T lymphocyte proliferation ex vivo greatest dynamic T cell response in our experimental setup and was used to determine Ag concentrations used in all presentation experiments Lymph node cells were isolated from inguinal lymph nodes of mice 12 d (Supplemental Fig. 2A–E) (30–35). after inoculation with rMOG/CFA and cultured in a 96-well U-bottom plates at a density of 2.5 3 105 cells/well. Cells were restimulated with 25 mg/ml Induction of EAE of MOG35–55, MOG71–90, MOG101–120, or rMOG for 48 h at 37˚C before 3 EAE was induced using standard protocols as described (10, 18, 36). In being pulsed with 1 mCi [ H]thymidine (MP Biomedicals, Irvine, CA) for brief, 8- to 10-wk-old female WT, Cybb2/2, and Ncf mice were anes- 18 h at 37˚C. Cells were harvested and the incorporation of radioactive thetized with -xylazine and injected s.c. with an emulsion of 50 mg thymidine was measured by scintillation (1450 MicroBeta TriLux; MOG (24 of 25 WT, 11 of 37 NOX2-deficient), 200 mgMOG , PerkinElmer). Stimulation index is the proliferative response relative to the 35–55 71–90 maximum proliferation induced by the positive control, rMOG, after no 200 mgMOG101–120,or200mgrMOGinCFA(0.5mg/mlMycobacterium butyricum in paraffin oil) (BD Difco, Franklin Lakes NJ) in a total volume of peptide was removed as background. 200 ml split between each flank. Pertussis toxin (300 ng) (List Biological Histopathology Laboratories, Campbell, CA) was injected i.p. (pH 7, in saline) on day 0 and day 2. MOG71–90 and MOG101–120 were not sufficient to induce EAE CNS lesion location, inflammation, and extent of demyelination were in any genotype. Mice were weighed and clinically scored each day for assessed by histopathological examination of CNS tissues from MOG35–55 40 d. In brief, scores were: 0, asymptomatic; 0.5, tail weakness; 1, limp and rMOG immunized mice sacrificed at day 12 or at peak clinical score tail; 1.5, hindlimb limping; 2, hindlimb weakness; 2.5, partial hindlimb (36, 39). Formalin-fixed cross-sections and longitudinal sections of brain, paralysis; 3, complete hindlimb paralysis; 3.5, hindlimb paralysis with cerebellum, and cervical, thoracic, and lumbar spinal cord were stained forelimb weakness; 4, forelimb paralysis; 4.5–5, morbidity/death (10). with Luxol fast blue and H&E (Goodman Cancer Centre Histology Core Mice that reached a clinical score of 4 were euthanized. Control mice Facility, Montreal, QC, Canada) and examined by a boarded veterinary injected with saline/CFA/pertussis toxin did not develop clinical symptoms pathologist in a blinded fashion. Histopathological assessment of inflam- of EAE. Where indicated, mice were sacrificed 12 d after injection, or at mation and demyelination was quantified according to Racke et al. (40). In the peak of the clinical score to collect tissues (brain, spinal cord, inguinal brief, inflammation scores were: 0, no inflammatory cells; 1, few scattered lymph nodes, spleen, and serum) for analysis by histopathology, quanti- inflammatory cells; 2, organization of inflammatory infiltrates into peri- 4992 NOX2 MODULATES PATTERNS OF Ag PROCESSING vascular cuffs; 3, extensive perivascular cuffing with extension into adja- 1024) is plotted for each binding region within the first 125 aa of MOG cent subarachnoid space and CNS parenchyma; 4, extensive perivascular (rMOG). The software algorithm SitePrediction (49) was used to predict cuffing with increasing subarachnoid and parenchymal inflammation. the cleavage patterns of MOG35–55 by cysteine cathepsins using input sites Demyelination scores were: 0, no demyelination; 1, a few scattered naked from the MEROPS database. The illustration (see Fig. 7) was drafted by b axons; 2, small groups of naked axons; 3, large groups of naked axons; 4, Bill Zaun of ZaunArt (Ft. Collins, CO). The MOG35–55/I-A complex (see confluent foci of demyelination; 5, widespread demyelination. Fig. 7) is adapted from Carrillo-Vico et al. (50). Structural analysis of cathepsin/MOG complexes qPCR and qPCR array The crystal structures of cathepsins L and S bound to ligands were obtained qPCR array technology was used to screen for transcriptional changes in from the (PDB) database, 3IV2 (41) and 3OVX (42), genes of interest to EAE (51, 52). The relative expression of 84 genes respectively. The ligand molecules were eliminated and the engineered relating to myelination, T cell activation and signaling, adaptive immunity, residues were mutated back to WT residues. Missing residues from the cytokines, chemokines, inflammation, apoptosis, cell adhesion, cell stress, structures were added through structural alignments with other available immune receptors, and immune/CNS-related transcription factors were 2 2 PDB files. The 1PY9 crystal structure was used for MOG (43). Similar to examined in the lumbar spinal cord of WT and Cybb / mice using the cathepsins, missing residues were added to the structure by comparing mouse multiple sclerosis RT2 Profiler PCR array (SABiosciences). Murine to other available structures on the PDB database. For performing the lumbar spinal cords were snap-frozen in liquid N2 and total RNA was docking process, the ZDOCK server was used (44, 45). The PDB files of extracted using the phenol-based RNeasy lipid tissue mini kit (Qiagen). cathepsins (L and S) and MOG were uploaded as receptors and ligands, cDNA was synthesized from 600 ng RNA using the RT2 First Strand kit respectively. Other than the active site residues and those comprising the (Qiagen). The qPCR array was optimized on an Eppendorf Mastercycler loops around the active site of the cathepsins, the rest of the proteins were gradient 2S (Qiagen) as per the manufacturer’s instructions. In brief, blocked as not being involved in the proteolytic function of cathepsins. In custom plates and buffers were used to optimize the thermocycler, cDNA the case of MOG, the connecting tail of the protein to the cell membrane from a single mouse of interest was used in a single array, and the results Downloaded from was blocked from interacting with the active site of cathepsins. ZDOCK were verified using the Web-based PCR array data analysis (http://www. then provided the top five predicted results as output PDB files. The first sabiosciences.com/pcrarraydataanalysis.php) and confirmed with the (top) prediction was then imported in the SPDB viewer package and en- Microsoft Excel–based PCR array data analysis template (PAMM-125Z). ergy minimized with the implemented GROMOS96 force field. The visual All gene expression was quantified relative to five housekeeping genes molecular dynamics package was used for visualization and structural (including b-actin and b2-microglobulin) and passed all internal control alignment (as mentioned above). tests (genomic DNA contamination, RNA quality, PCR performance). Three independent experiments for each group (WT, 12 d/peak; Cybb2/2, 2/2 Fluorometric measurement of cathepsin activities in vitro 12 d/peak; WT/Cybb , mock injected) were tested to allow for statistical http://www.jimmunol.org/ analysis. To measure macrophage/microglial polarization to an M2 phe- Measurements of cysteine cathepsin activities from lysosomal extracts were notype, expression levels of the M2-associated marker ARG-1 were performed as described (7). In brief, lysosomes were magnetically isolated measured by conventional qPCR. cDNA was made from lumbar spinal from iron/dextran-loaded BMMfs and lysed in 20 mM sodium acetate cord RNA (described above) using iScript reverse transcriptase supermix buffer containing 0.1% Tween 80 at pH 5.0 (46). The lysosomal extracts for RT-qPCR (Bio-Rad). All primers were at 300 nM, had a single melting were added to 0.2 M potassium acetate buffer (pH 5.0) containing 1 mM curve, had efficiencies between 90 and 100%, and were designed or ver- cysteine/cystine redox buffer 600:1 (221–236 mV) (47), varying concen- ified using Primer 3 (National Center for Biotechnology Information). 18S trations of hydrogen peroxide, and cathepsin-specific fluorogenic sub- (forward, 59-AGTCGGCATCGTTTATGGTC-39, reverse, 59-CGCGGTT- strates, including cathepsin S substrate (Ac-KQKLR-AMC, Ex of 354 nm, CTATTTTGTTGGT-39) was used as an internal control and did not vary Em of 442 nm, 0.645 mg/well; Anaspec), cathepsin L substrate (Ac-His- across treatments, with the following PCR conditions (in a Bio-Rad iQ5

Arg-Tyr-Arg-ACC, Ex of 380 nm, Em of 460 nm, 5 mg/well; Calbiochem/ by guest on September 28, 2021 thermocycler): 95˚C for 5 min and 40 cycles of 95˚C for 30 s and 58˚C for Merck), and aspartic cathepsin D/E substrate (Mca-GKPILFFRLK(Dnp)- 30 s. ARG-1 (forward, 59-AGGGTTACGGCCGGTGGAGAG-39, reverse, r-NH2, Ex/Em of 328/393 nm, 1 mg/well; Anaspec)]. Fluorescent dequenching 59-CCCCTCCTCGAGGCTGTCCTTT-39) was quantified under the fol- of the substrates was measured using a FLUOstar Optima fluorescent plate lowing PCR conditions: 95˚C for 3 min, 50 cycles of 95˚C for 15 s, and reader (BMG Labtech), or an EnVision multilabel reader (PerkinElmer) 60˚C for 60 s (31). Expression is presented relative to 18S and relative to at 37˚C. Slopes of initial reaction rates were determined by curve-fitting the mock control samples. applications in Microsoft Excel and expressed relative to untreated controls. Statistics Cathepsin-mediated MOG35–55 cleavage Unless indicated, statistical analyses were completed by one-way ANOVA Recombinant cathepsin L (0.25 mg, Novoprotein, Shanghai, China) and S (or unpaired Student t test) with a Tukey test or nonparametric equivalent (5 ng, EMD Millipore, Billerica, MA) were added to 0.2 M potassium (p , 0.05). When a Bartlett’s test for equal variances failed, data were acetate buffer (pH 5.0) containing 1 mM cysteine/cystine redox buffer transformed (natural log transformation) and reanalyzed. Analyses of all 600:1 (221–236 mV) (47), varying concentrations of hydrogen peroxide, clinical data were done using a Kruskal–Wallis test (Dunn’s multiple and 10 mg MOG . Digestions were incubated for 15 min (cathepsin S) 35–55 comparisons). Statistical analyses of enzymatic activity assays (of lyso- or 60 min (cathepsin L) at 37˚C in a Bio-Rad MyCycler thermal cycler somal extracts and MOG abundance) were done by nonlinear re- (Bio-Rad, Mississauga, ON, CAN), followed by heat denaturation for 2 42–55 gression (variable slope model). All statistical analyses were completed min at 90˚C. Peptides were separated by reverse-phase HPLC on a Supelco using GraphPad Prism software (GraphPad Software, La Jolla, CA). Ascentis Express C18 column (10 cm length, 2.7 mmpore)(Sigma- Aldrich) using a 2–50% acetonitrile gradient with 0.1% TFA. Peaks were collected and corresponding peptides identified by MALDI-TOF and/ Results or liquid chromatography–tandem mass spectrometry (Southern Alberta A deficiency of NOX2 subunits influences patterns of Ag Mass Spectrometry Centre, Calgary, AB, Canada). The relative abundance of the peptides was quantified using the area under corresponding peaks. processing phox 2/2 phox Abundance of the cleavage product MOG42–55 relative to MOG35–55 BMMfs from mice deficient in either gp91 (Cybb )orp47 b (corresponding destruction of the I-A binding region) was calculated (Ncf) subunits of NOX2 did not produce a measurable respi- using the following formula: RPA = (P /S )/(P /S ), where RPA indicates x x 0 0 ratory burst within the phagosome following the uptake of relative peptide abundance, Px indicates product (MOG42–55)ofsam- ple, Sx indicates substrate (MOG35–55) of sample, P0 indicates product IgG-opsonized particles (Fig. 1A, 1B). Consistent with previous (MOG42–55) without hydrogen peroxide, and S0 indicates substrate reports, we found that the absence of NOX2 activity promoted the (MOG35–55) without hydrogen peroxide. reductive potential of the phagosome (as evidenced by the enhanced Computational analysis of I-Ab binding and protease cleavage disulfide bond reduction of phagocytosed cystine), which in turn of MOG favored activities of the cysteine cathepsins (cathepsin B, S, and L) and led to a significant increase in overall phagosomal proteolysis The immune epitope database and analysis resource MetaSVMp (a quantitative binding affinity program that employs the Immune Epitope (Fig. 1C–F, Supplemental Fig. 1A, 1B). As anticipated, NOX2 Database was used to compute relative affinities between I-Ab and regions deficiency did not alter the rate or extent of phagosomal acidifi- of MOG1–125 as previously described (48). The inverse of the IC50 (nM 3 cation, or rates of endo/phagocytosis (Supplemental Fig. 1C–F). The Journal of Immunology 4993

FIGURE 1. Phagosomes in NOX2-deficient BMMfs have diminished production of ROS, ele- vated bulk proteolysis and cysteine cathepsin activity, and altered relative abundance of peptide products compared with WT. (A and B) Phagosomal respira- tory burst, (C and D) total proteolytic activity (rate of substrate-liberated fluorescence from the particle- bound fluorogenic substrate DQ green Bodipy albu- min), (E and F) cysteine cathepsin activities, and (G) relative abundance of HEL cleavage products of WT 2/2 and NOX2-deficient BMMfs (Cybb and Ncf) Downloaded from after phagocytosis of fluorometric experimental par- ticles. (A, C, and E) Representative real-time traces where RFU are directly proportional to the degree of substrate oxidation or hydrolysis and are taken from 0 to 90 min after phagocytosis. (B, D, and F) The average rates (taken from 30 to 80 min after phago- cytosis) of oxidation, total proteolytic activity, and http://www.jimmunol.org/ cysteine cathepsin activity relative to the WT BMMf samples. (G) Relative abundance of peaks produced during HPLC analysis of peptides collected from lysed BMMfs (WT and Cybb2/2) that were incu- bated with IgG particle-restricted Alexa Fluor 488– HEL for 90 min. Relative peptide abundance is the area under the curve for each peak proportional to WT samples (see representative traces in Supplemental Fig. 1G). Data represent three inde- pendent experiments relative to WT control. (B, D, F, by guest on September 28, 2021 and G) Data are presented as means 6 SEM. *p , 0. 05 versus WT control (by ANOVA).

Next, we determined whether the NOX2-mediated inhibition of (Supplemental Fig. 1G). Most peaks (representing peptide products) selected lysosomal proteases (the cysteine cathepsins) influenced were common to both traces from WT and Cybb2/2 macrophages the pattern of proteolysis and thus the relative abundance of peptide (Fig. 1G, Supplemental Fig. 1G). However, the relative abundance fragment products of phagocytosed Ag. To achieve this we used the of many peptides (as determined by area under the curve for a given model Ag HEL, heavily derivatized with Alexa Fluor 488 succi- peak) differed markedly in the presence or absence of NOX2 (Fig. nimidyl ester and covalently coupled to opsonized experimental 1G). Additionally, a peak reliably produced by WT BMMØs was particles. BMMfs derived from WT and NOX2-deficient mice consistently undetectable in Cybb2/2 samples (Fig. 1G). These data were allowed to phagocytose and process the labeled protein cargo support the hypothesis that NOX2 modulates not only the overall for 90 min. BMMfs were subsequently lysed and the composition efficiency of phagosomal proteolysis, but also influences the relative of the fluorescent peptide products ,10 kDa was resolved by re- abundance of different peptide epitopes potentially available for verse phase HPLC. Although this experimental approach precludes MHC-II binding and presentation. the identification of the antigenic peptides generated (due to the overwhelming background of nonfluorescent endogenous pep- NOX2 activity influences MHC-II presentation in an tides), it provides a snapshot of the peptides derived from the Ag-specific, APC-specific manner phagocytosed Ag that are potentially available for MHC-II binding Because NOX2 activity modified the pattern of phagosomal an- within the endolysosomal system. The traces representing peptide tigenic proteolysis and composition of available peptides, we in- composition generated from phagocytosed protein by BMMfs vestigated whether this could, in turn, affect the presentation were highly conserved and reproducible within each genotype efficiency of particular MHC-II–restricted epitopes. To achieve 4994 NOX2 MODULATES PATTERNS OF Ag PROCESSING this, we examined the efficiency of epitope-specific CD4+ T cell activation, using OTII and 2D2 transgenic T cell models that re- spond to the I-Ab immunodominant peptides derived from OVA and MOG, respectively. In brief, BMMfs and BMDCs derived from WT and NOX2-deficient mice were allowed to endocytose unprocessed whole Ag or preprocessed antigenic peptides for a defined period. The activation of CD4+ T cells isolated from OTII and 2D2 mice and coincubated with the APCs were assessed by measuring surface expression of the early activation markers CD69 and CD25 as previously described (15, 25, 31, 33, 35, 53, 54). Deficiency in either the gp91 or p47 subunits of NOX2 did not affect the presentation of either model Ags when APCs were pulsed with preprocessed peptide (Fig. 2A, 2B, Supplemental Fig. 2A, 2C, 2F–H). Similarly, WT and NOX2-deficient APCs + were equally able to activate OVA323–339-specific CD4 Tcells when given whole OVA (Fig. 2A, Supplemental Fig. 2F). In contrast, however, the ability of BMMfs, but not BMDCs, to + activate MOG35–55-specific CD4 T cells in the absence of NOX2 was significantly diminished when given the whole recombinant Downloaded from MOG protein (rMOG, MOG1–125) (Fig. 2B, Supplemental Fig. 2G, 2H). The discrepancy between WT and NOX2-deficient BMMfs to present MOG35–55 was further increased when NOX2 activity was enhanced by delivering the rMOG protein on IgG-opsonized experimental particles (Fig. 2B, Supplemental

Fig. 2H). To determine whether the processing of endogenously http://www.jimmunol.org/ derived MOG would be similarly affected by the NOX2 activity of macrophages, myelin sonicates from whole mouse brain were prepared and given to BMMfs. Similar to findings with recombi- nant MOG, NOX2-deficient macrophages were significantly less able to process and present MOG35–55 from endogenously expressed MOG in purified myelin (Fig. 2C, Supplemental Fig. 2E, 2I). Collectively, these data suggest that NOX2 activity selectively affects the presentation of Ag through specific modulation of Ag processing. by guest on September 28, 2021 To further interrogate the ability of NOX2 to alter processing patterns of a single Ag, we measured the efficiency of WT and 2/2 Cybb BMMfs to process four distinct HEL epitopes (HEL20–35, HEL31–47, HEL48–62, and HEL74–90) from whole HEL protein and present them to epitope-specific CD4+ T cell hybridomas (Hb1.9, H30.44, H46.13, and B04, respectively) (26). To achieve this, T cell hybridomas were first stably transfected with an NFA- T-enhanced IL-2 promoter–driven eGFP construct, followed by multiple rounds of selection by FACS to enrich those hybrid- FIGURE 2. NOX2 deficiency in BMMfs does not affect efficiency of presentation of preprocessed peptide Ag, but affects processing of protein oma clones that expressed eGFP only following presentation of 2 2 Ag in an Ag-specific manner. WT, Cybb / , and/or Ncf BMMf were their cognate peptide epitope by APCs (25). When added to incubated for 6 h with (A) no peptide (NP), OVA323–339 (332–339; 10, 25 BMMfs that were previously pulsed with HEL protein, we found mg/ml), OVA (10, 25 mg/ml); (B) MOG35–55 peptide (35–55; 25 mg/ml), 2/2 f that WT and Cybb BMM s were equally able to induce eGFP rMOG1–125 (rMOG; 10, 25 mg/ml), 3-mm IgG-opsonized silica particles expression in two of the hybridomas (Hb1.9 and H46.13), indi- (multiplicity of infection of three to five) covalently coupled with cating that the processing/presentation efficiency of HEL20–35 and rMOG1–125 (rMOG Beads); (C) mouse myelin (25, 50 mg/ml); or (D) 1 mg/ HEL48–62 peptides were unaffected by NOX2 (Fig. 2D). Interest- ml HEL. Activation of the OVA323–339- and MOG35–55-specific OTII and ingly, NOX2 had a disparate effect on the processing/presentation 2D2 CD4+ T cells was determined by expression of CD69 after 16 h with + efficiencies of the two other HEL epitopes, with Cybb2/2 BMMfs the BMMf. Activation of CD4 T cell HEL-specific hybridomas showing compromised presentation of HEL (reduced eGFP (Hb1.9 responding to HEL20–35, H30.44 responding to HEL31–47,H46.13 31–47 responding to HEL , and B04 responding to HEL ; stably trans- expression in H30.44), but an enhanced presentation efficiency of 48–62 74–90 fected with pNFATeGFP) was determined by eGFP expression. (A–C) HEL74–90 (increased eGFP expression in B04) (Fig. 2D). Con- Data represent five to six independent experiments, presented relative to sistent with earlier biochemical findings (Fig. 1), these data in- the WT 25 mg/ml MOG35–55/OVA323–339 peptide control [or relative to WT dicate that NOX2 does not uniformly affect processing of a given myelin (C)]. (D) Data represent four to seven independent experiments Ag, but differentially affects relative epitopic generation from and are presented relative to a maximum eGFP expression induced by a single Ag. 50 ng/ml PMA with 1 mg/ml ionomysin. Data are presented as means 6 SEM *p , 0.05 versus internal WT controls (by ANOVA). NOX2-deficient I-Ab mice show protection from MOG-induced EAE peptide, we reasoned that NOX2-deficient I-Ab mice would be Because NOX2-deficient BMMfs have a reduced ability to pro- less susceptible to MOG-induced EAE. Indeed, following the b cess MOG to generate the vastly I-A –immunodominant MOG35–55 induction of EAE using the peptide MOG35–55 and rMOG, both The Journal of Immunology 4995

Cybb2/2 and Ncf mice showed a reduced incidence and a (Fig. 3H). Interestingly, although NOX2 deficiency also affor- delayed onset of clinical disease, as well as a reduction in the ded some protection from full-length MOG protein-induced average of peak clinical scores (Fig. 3A–E) when compared EAE, it was typically less apparent, and the onset of EAE in with WT mice. Supporting the clinical observations, NOX2- Ncf mice was not significantly delayed (Fig. 3E). We reasoned deficient mice were shown to have on average 2- to 3-fold that this may result from the generation of other potential immuno- + fewer infiltrating macrophages, microglia, B cells, CD4 T cells, genic epitopes (MOG71–90 and MOG101–120) in these mice, even and CD8+ T cells in spinal cord tissue at 12 d postinoculation though we found them to be insufficient to induce EAE on their as compared with WT (Fig. 3F, 3G). NOX2 deficiency also own. Nonetheless, consistent with the findings that NOX2-deficient + b resulted in significantly lower numbers of MOG35–55 tetramer BMMfs are less efficient in the generation of the I-A –immuno- + + b CD4 T cells within spinal cord tissue, as well as reduced CD4 dominant MOG epitope MOG35–55 in vitro, NOX2-deficient I-A TcellexpressionofIFN-g, indicating reduced MOG-specific mice show protection from MOG-induced, particularly MOG35–55- T cell activation within the CNS in the absence of NOX2 induced, EAE. Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

FIGURE 3. NOX2-deficient mice are partially protected from EAE. Clinical disease course of NOX2-deficient (Cybb2/2 and Ncf) and WT mice after active induction of EAE with (A)50mg MOG35–55 or (B) 200 mg rMOG1–125.(C) Clinical incidence (all mice that exceed a clinical score of 0), (D) maximum clinical score, and (E) time to clinical onset (score of at least 0.5) of EAE induced using 50 mg MOG35–55 (35–55) or 200 mg rMOG1–125 (rMOG). (A–E) Mice were clinically scored (0–4) daily for 40 d. Data represent four independent experiments, each containing cohorts of three to seven 2/2 2/2 mice per genotype (total mouse numbers: MOG35–55, n = 25 WT, 23 Cybb , 14 Ncf, rMOG1–125 n = 28 WT, 19 Cybb , 19 Ncf); (F–H) n = 3–4 per genotype/injection. (F) The total number of leukocytes isolated from the spinal cord (via a discontinuous Percoll gradient) of mice 12 d postinoculation with + high 50 mg MOG35–55 (35–55), 200 mg rMOG1–125 (rMOG), or CFA alone (Mock). (G) The number of macrophages (Mf, CD11b /CD45 ), microglia (MG, CD11b+/CD45 low), B cells (B220+/CD45+), CD8+ T cells (CD8+/CD3+), and CD4+ T cells (CD4+/CD3+) isolated from the spinal cord of mice 12 d + + + + postinjection with 50 mg MOG35–55.(H) The number of infiltrating CD4 T cells that express Foxp3, IFN-g, IL-17, (CD4 /Foxp3, CD4 /IFN-g, CD4 /IL- b + + + 17, respectively) or MOG35–55/I-A –positive CD4 T cells (CD4 /MOG35–55 tetramer ) isolated from spinal cord of mice 12 d after inoculation with 50 mg MOG35–55. Data are presented as means 6 SEM. *p , 0.05 versus WT internal control (clinical data, Kruskal–Wallis; Student t test). 4996 NOX2 MODULATES PATTERNS OF Ag PROCESSING

+ Clinical progression after onset and the final pathology of EAE in MOG35–55 CD4 T cell clones in the peripheral lymph nodes of clinically affected animals were not influenced by NOX2 deficiency WT and Cybb2/2 mice were measured 12 d postinoculation. 2 2 Although Cybb / and Ncf mice show reduced incidence and Using a MOG35–55-specific tetramer and peptide-driven ex vivo delayed onset of EAE, we noted that NOX2-deficient animals that proliferation, it was found that the clonal expansion of MOG35–55- + did develop EAE symptoms progressed (albeit delayed) in responsive CD4 T cells in secondary lymphoid organs was un- 2/2 a manner that was clinically indistinguishable from affected WT changed in Cybb mice following inoculation with MOG35–55 mice (Fig. 4A). Likewise, mice that were sacrificed at the peak peptide (Supplemental Fig. 3D). Because DCs are primarily + of the clinical disease (between clinical scores of 2.5 and 3.5) charged with the activation and expansion of naive CD4 T cells, exhibited comparable levels of leukocyte infiltration, inflamma- this finding is consistent with the ability of NOX2-deficient tion, and demyelination of the spinal cord (Fig. 4B–D), peripheral BMDCs to efficiently present MOG35–55 (Supplemental Fig. + cytokines (Supplemental Fig. 3A), proliferative response to 2G). Interestingly, although the peripheral expansion of CD4 MOG35–55 (Fig. 4E), ARG-1 mRNA levels (Supplemental Fig. T cells specific to the immunodominant MOG35–55 was equiva- 2/2 3B), and general CNS mRNA expression patterns of genes rele- lent, Cybb , but not WT, mice exhibited additionally expanded vant to EAE (Supplemental Fig. 3C). These findings suggest that MOG71–90- and MOG101–120-responsive T cell clones (Fig. 4E, NOX2 did not act to modify the progression of disease once it had Supplemental Fig. 3E). This finding suggests that while NOX2 passed the clinically apparent threshold, but it influenced the deficiency does not alter the generation of MOG35–55 in DCs, it stages of pathogenesis that preceded the establishment of overt actually enhances both the generation and presentation of the al- encephalomyelitis. ternative MOG epitopes, further supporting a role of redox- To determine whether peripheral responses to the MOG inoc- mediated modification to patterns of Ag processing in an in vivo Downloaded from + ulation were affected by NOX2, the degrees of expansion of setting. Nonetheless, because the initial priming of MOG35–55 CD4 http://www.jimmunol.org/

FIGURE 4. NOX2 deficiency does not alter dis- ease severity of EAE in clinically affected mice. (A) The maximum clinical score of mice that reached a clinical score of 0.5 for at least 1 d after injection with 50 mg MOG35–55 (35–55) or 200 mg rMOG1–125 (rMOG). Data represent four independent experi- by guest on September 28, 2021 ments, each containing cohorts of three to seven mice per genotype. (B) The number of macrophages (Mf, CD11b+/CD45high), microglia (MG, CD11b+/ CD45low), B cells (B220+/CD45+), CD8+ T cells (CD8+/CD3+), and CD4+ T cells (CD4+/CD3+) iso- lated from spinal cord of mice at the peak of the disease following inoculation with rMOG1–125 counted and characterized by flow cytometry (n = 3–4). (C and D) Histopathology of CNS removed from mice immunized with MOG35–55 at the peak of the disease (score .2.5). (C) Representative micrographs of lumbar spinal cord stained with Luxol fast blue (LFB) or H&E; arrows indicate areas of inflammation and demyelination. (D)His- tological scores for inflammation and demyelination (n =3).(E) Peripheral activation and peptide-driven ex vivo expansion of lymphocytes. Lymph node cells were isolated from inguinal lymph nodes of mice 12 d postinoculation with rMOG/CFA and restimulated in vitro with 25 mg/ml MOG peptides 35–55, 71–90, and 101–120 for 48 h. Stimulation index represents [3H]thymidine incorporation as measured by scintillation (n = 6). Data are presented as means 6 SEM; significant differences. *p , 0.05 versus WT internal control (clinical data, Kruskal– Wallis; by ANOVA). The Journal of Immunology 4997

T cells in response to the inoculation of exogenous MOG was unchanged by NOX2, we reasoned that the inefficiency of NOX2-deficient macrophages to process endogenous MOG in the CNS limited the progression to clinically apparent enceph- alomyelitis. + Reactivation of MOG35–55-specific CD4 T cells within the CNS during preclinical EAE is compromised in NOX2-deficient mice To interrogate whether CNS-resident NOX2-deficient APCs are compromised in their ability to present endogenous MOG in preclinical EAE, the efficiency of reactivation of adoptively + transferred MOG35–55-specific 2D2 CD4 T cells in WT and Cybb2/2 mice was examined. 2D2 CD4+ T cells were expanded in vitro and transferred into WT and Cybb2/2 mice 7 d after in- oculation with MOG35–55. CNS tissue was recovered 72 h after transfer, and recruitment and reactivation MOG35–55-specific 2D2 CD4+ T cells were determined by flow cytometry. As expected + during this preclinical phase, MOG35–55-specific 2D2 CD4 Downloaded from T cells (Va3.2+ and Vb11+ CD4+ T cells) were recruited in equivalent numbers to the CNS in WT and Cybb2/2 mice (Fig. 5A). However, these recruited MOG-responsive CD4+ T cells displayed significantly reduced expression of the activation marker CD25 and the Th1 effector IFN-g in the CNS of Cybb2/2

mice demonstrating a reduced MOG35–55-specific reactivation in http://www.jimmunol.org/ the absence of NOX2 in this tissue (Fig. 5B–D). Lysosomal cysteine cathepsins destroy critical regions in MOG35–55 in a redox-dependent manner We hypothesized that the inefficiency of NOX2-deficient macro- phages to present the immunodominant MOG35–55 peptide would be due to the increased hydrolysis of critical regions within MOG35–55 by cysteine cathepsins in the early phagosome. Using

SitePredict, a sequence-based prediction tool for protease cleav- by guest on September 28, 2021 age, we identified two potential cathepsin L/S cleavage sites b 41 42 within the critical I-A binding region of MOG35–55 (Arg /Ser and Ser45/Arg46) (49, 50, 55, 56). Pertinently, both of these cleavage sites are located on a solvent-accessible loop in the MOG protein, rendering them potentially susceptible to hydrolysis in the early phagosome. This prediction was further supported by de- termining the spatial interactions between the MOG protein and cathepsins L and S using a ZDOCK algorithm (44, 45). ZDOCK makes predictions on interactions between two protein structures based on shape complementarity, desolvation free energy, and FIGURE 5. The efficiency of reactivation of MOG35–55-specific 2D2 + electrostatic energies. By constraining the protein–protein inter- CD4 T cells is diminished within the CNS NOX2-deficient mice. (A and B + action based on the location of the cathepsin active sites, ZDOCK ) Transgenic MOG35–55-specific 2D2 CD4 T cells were isolated and expanded ex vivo using IL-12 and MOG35–55 for 72 h before adoptive has also predicted that the loop region of amino acid 41–46 on the 2/2 transfer into WT and Cybb mice 7 d after inoculation with MOG35–55. MOG can intimately interact with the active sites of both cathe- + Transferred MOG35–55-specific 2D2 CD4 T cells were isolated from the psins L and S (Fig. 6A, 6B). Using recombinant systems and mass CNS after 72 h in vivo using a discontinuous Percoll gradient and iden- spectral analysis, we confirmed that both cathepsins L and S ef- tified (CD4+,Vb11+,Va3.2+) by flow cytometry. Expression of CD25 + + + + + + ficiently degrade MOG35–55 under reductive conditions within the (CD4 ,Va3.2 , CD25 ) and IFN-g (CD4 ,Va3.2 , IFNg ) by transferred + critical 41–46 region (Fig. 6C, 6D). Because both cathepsin L and MOG35–55-specific CD4 T cells were determined by flow cytometry. (C cathepsin S are oxidatively inhibited by NOX2 products (Fig. 6E), and D) Corresponding flow cytometry plots. For (A) and (B), n = 5. Data 6 , we further demonstrated that MOG35–55 was protected from are presented as means SEM. *p 0.05) (by Student t test). cleavage by cathepsins L and S by increasing concentrations of hydrogen peroxide (Fig. 6D). Taken together, these predictions b of MHC-II Ags in an Ag- and cell-specific manner. Furthermore and experimental data demonstrate that the I-A binding region of data presented strongly suggest that this phenomenon is mediated MOG35–55 in the native protein is able to be specifically and ef- through a redox-based mechanism, where phagosomal NOX2 ficiently cleaved by cathepsin L and cathepsin S under reductive, inhibits local cysteine cathepsins which require a reductive envi- but not oxidative, conditions. ronment for optimal activity. Specifically, we found that the I-Ab– immunodominant epitope of MOG (MOG35–55), which contains Discussion a cathepsin L/S cleavage site within the I-Ab binding region, is not In this study, we have shown that NOX2, in addition to its anti- efficiently presented by NOX2-deficient macrophages. Addition- microbial function, alters the pattern of phagolysosomal processing ally, we demonstrated that NOX2-deficient mice were partially 4998 NOX2 MODULATES PATTERNS OF Ag PROCESSING Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

b FIGURE 6. MOG is cleaved within the MOG35–55 I-A binding region by cathepsin L and S in a H2O2-inhibitable manner. (A and B) Novel repre- sentations of the spatial interaction between cathepsin L/S and MOG as determined by a ZDOCK algorithm. The complex is colored based on protein secondary structure: yellow denotes b-sheets; purple denotes a helix; blue denotes 310 helix. The region of residues 35–55 on MOG has been highlighted in green. Target residues within the predicted cleavage site of cathepsin L/S on MOG are shown with sticks. Cyan, blue, red, yellow, and white sticks represent carbon, , oxygen, sulfur, and hydrogen atoms, respectively. (A) The inset presents a magnified view of the predicted target loop (Arg41, Ser42, Pro43, Phe44, Ser45, and Arg46) of MOG within the active site of cathepsin L (Cys25 and His163). (B) The inset presents a magnified view of the predicted target loop (Arg41, Ser42, Pro43, Phe44, Ser45, and Arg46) of MOG within the active site of cathepsin S (Cys25 and His164). (C) A schematic of the predicted and b experimentally confirmed cathepsin L/S cleavage site in relation to the I-A binding region (red) of the immunodominant epitope of MOG (MOG35–55), and the corresponding cleavage products that were identified by MALDI-TOF MS. (D) The ability of recombinant cathepsin L and recombinant cathepsin S to cleave

MOG35–55 was assessed using MOG35–55 in a reconstituted system with increasing concentrations of H2O2. Following incubation at 37˚C, relative quantities of remaining substrate (MOG35–55) and cleavage products (MOG42–55 and MOG35–41) were quantified by reverse-phase HPLC following peak identification by b MALDI-TOF MS. Data are presented as the abundance of the cleavage product MOG42–55 relative to MOG35–55 (corresponding destruction of the I-A binding region) and was calculated using the following formula: RPA = (Px/Sx)/(P0/S0), where RPA indicates relative peptide abundance; Px indicates product (MOG42–55) of sample, Sx indicates substrate (MOG35–55) of sample, P0 indicates product (MOG42–55) without hydrogen peroxide, and S0 indicates 2 2 substrate (MOG35–55) without hydrogen peroxide. Cathepsin L (IC50 of 35.97 mMH2O2; R = 0.91) and cathepsin S (IC50 of 60.84 mMH2O2; R = 0.99). (E) Relative rates of hydrolysis of cathepsin-specific fluorogenic substrates (cathepsin L, Ac-HRYR-ACC; cathepsin S, Ac-KQKLR-AMC; cathepsin D/E,

Mca-GKPILFFRLK-Dnp) by BMMf lysosomal extracts in increasing concentrations of H2O2. The cathepsin L (IC50 by BMMf lysosomal extracts of 2 2 42.34 mMH2O2; R = 0.96), cathepsin S (IC50 of 94.28 mMH2O2; R = 0.97), and aspartic cathepsin D/E cleavage (cathepsin D/E: does not converge). Data are from three experiments and presented as means 6 SEM. Statistical analysis was performed using nonlinear regression (variable slope model) where the inhibitor (H2O2) is log-transformed.

+ protected from MOG-induced EAE, a CD4 T cell–driven auto- to the immunodominant MOG35–55 epitope by NOX2-deficient immune disease, despite eliciting a normal peripheral immune macrophages (Fig. 7). response to the inoculated MOG Ag. Hence, we propose that the The effect of NOX2 on MOG processing appeared to be APC- partial protection from EAE in NOX2-deficient mice is brought specific. We found that when APCs were given intact MOG, NOX2 + about by ineffective reactivation of effector CD4 T cells within activity enhanced MOG35–55 presentation in BMMfs, but had no the CNS due to the inefficient processing of endogenous MOG effect in BMDCs (Supplemental Fig. 2G). It could be argued that The Journal of Immunology 4999

BMDC preparations may not be homogeneous and therefore we faithfully recapitulate the antigenic repertoires produced by DCs cannot draw absolute conclusions regarding the effect of NOX2 on in the primary immune response. MOG processing in DCs from these experiments. However, the Although we have shown that NOX2-deficient macrophages are differential effect of NOX2 on APC processing was supported by unable to efficiently process and present MOG and to efficiently + the finding that following inoculation with full-length MOG, reactivate MOG35–55-responsive CD4 T cells in the CNS, it is NOX2-deficient mice induced a normal MOG35–55-reactive pri- impossible to definitively exclude other effects that NOX2 defi- mary T cell response in the lymph node, but a reduced secondary ciency may have during EAE. Extracellular release of NOX2- + response and reactivation of expanded MOG35–55-reactive CD4 derived ROS by infiltrating leukocytes in NOX2-competent mice T cells in the CNS (Figs. 3, 5, Supplemental Fig. 3A, 3D). The could potentially contribute to the pathology of EAE, which disparity between these related cell types may be explained by would manifest as a comparatively milder clinical syndrome in lower NOX2 activity in DCs or by their enhanced processing and NOX2-deficient mice. In our study, whereas the incidence and presentation efficiency of low abundance Ags (1, 45). Alterna- onset of clinically apparent EAE was altered in NOX2-deficient tively, this disparity may be the result of the differential expression mice, animals that did succumb to EAE displayed identical clin- of cysteine cathepsins between these APCs. Although cathepsin L ical courses (following initial onset clinical signs) and the same mRNA and protein is detectable in DCs, cathepsin L activity is severity of disease (as determined by maximum clinical score, notably absent (43, 57, 58). Macrophages, alternatively, express histopathological findings, cellular infiltration, and CNS tran- high levels of the active form of cathepsin L (4, 41). Similarly, scriptional analysis) irrespective of genotype. Although the role expression of active cathepsin S is modest in DCs but highly of extracellular ROS-driven pathology cannot be excluded, our expressed in macrophages, particularly in response to classical observations do not fit with this model. Another possible con- Downloaded from activating stimuli (1, 59). Because the I-Ab-binding region of tributing factor could arise through an absence of NOX2-mediated MOG35–55 is susceptible to cleavage by cathepsins L and S (Fig. signaling, which may polarize macrophages toward a protective 6), it stands to reason that the regulation of these by M2 (alternatively activated) phenotype. Specifically, there are two NOX2 is more important to epitopic preservation in macrophages reports of a deficiency in p47phox (Ncf) potentiating an M2-like than in DCs. Nevertheless, differential representation of presented phenotype in peritoneal macrophages or microglia in response to

peptide epitopes on DCs and macrophages presents an interesting Listeria monocytogenes or LPS, respectively (61, 62). Interest- http://www.jimmunol.org/ quandary. Whereas DCs are typically charged with priming the ingly, macrophages deficient in gp91phox (Cybb2/2) did not show CD4+ T cell response to a new Ag, macrophage presentation to the same propensity for M2 polarization (61). During peak EAE, expanded CD4+ T cells at sites of inflammation is often essential we found that the M2 marker Arg-1 within the CNS was equally for the propagation and escalation of the effector immune expressed in NOX2-deficient and WT mice (Supplemental Fig. responses to persistent Ag (42, 60). Our data indicate that, in 3B), and that Th1 response predominated in both draining NOX2-deficient mice, lymph node DCs present an MHC-II–re- lymph nodes and CNS tissues in response to MOG inocula, irre- stricted peptide epitope that cannot be efficiently recreated by spective of genotype (Supplemental Fig. 3C, 3D). Additionally, peripheral macrophages from endogenous antigenic sources (e.g., mice deficient in either p47phox (Ncf) and gp91phox (Cybb2/2) myelin), and this ultimately creates an inefficiency between displayed similar levels of protection from EAE (Fig. 3), as well by guest on September 28, 2021 priming and effector Th responses. Hence, in this scenario, NOX2 as a similar pathology to WT mice at the peak of disease (Fig. 4). activity within activated tissue macrophages acts not only to slow Taken together, these data suggest that the primary protection the overall rates of phagosomal proteolysis, but also to more from EAE by NOX2 deficiency is unlikely to be the result of

FIGURE 7. An illustration of the reac- tivation of effector Th cells in the CNS of WT and NOX22/2 mice from endogenous MOG. MOG-containing myelin is phago- cytosed during CNS inflammation and pro- cessed within phagosomes. In WT macro- phages, NOX2 limits the activities of cathepsin L and S and these macrophages can efficiently present MOG35–55 to MOG35–55-specific ef- fector Th cells. In a feed-forward response, the reactivated Th cells further recruit and activate macrophages, leading to increased phagocytosis of myelin and upregulation of NOX2 and MHC-II. NOX2-deficient macro- phages cannot oxidatively inactivate cathepsin L and S within the phagosome, and thus have a higher likelihood of cleaving the immuno- b dominant MOG35–55 in the I-A binding re- gion. This prevents efficient MHC-II presen- tation to MOG35–55-specific effector Th cells and the efficient perpetuation of the autoim- b mune response. The MOG35–55/I-A complex is adapted from Carrillo-Vico et al. (50). 5000 NOX2 MODULATES PATTERNS OF Ag PROCESSING decreased ROS-mediated pathology or an undefined role of p47 in 6. Mantegazza, A. R., A. Savina, M. Vermeulen, L. Pe´rez, J. Geffner, O. Hermine, S. D. Rosenzweig, F. Faure, and S. Amigorena. 2008. NADPH oxidase controls myeloid polarization, and that their contribution to the observed phagosomal pH and antigen cross-presentation in human dendritic cells. Blood phenotype is subtle, if at all. This is additionally supported by 112: 4712–4722. Hultqvist et al. (63) who reported that p47-deficient H2q mice 7. Rybicka, J. M., D. R. Balce, S. Chaudhuri, E. R. Allan, and R. M. Yates. 2012. q Phagosomal proteolysis in dendritic cells is modulated by NADPH oxidase in developed EAE with the same incidence as WT H2 mice fol- a pH-independent manner. EMBO J. 31: 932–944. lowing induction with MOG1–125, but not with MOG79–96. This 8. Collins, D. S., E. R. Unanue, and C. V. Harding. 1991. Reduction of disulfide further indicates that protection from MOG-induced EAE in bonds within lysosomes is a key step in antigen processing. J. Immunol. 147: 4054–4059. NOX2-deficient mice is an Ag-, haplotype-specific phenomenon. 9. Hastings, K. T., R. L. Lackman, and P. Cresswell. 2006. Functional requirements The possible binding regions of the MOG protein to I-Ab (with for the lysosomal thiol reductase GILT in MHC class II-restricted antigen pro- cessing. J. Immunol. 177: 8569–8577. reasonable affinity) are strikingly limited to MOG35–55 (actual b ∼ 10. Stromnes, I. M., and J. M. Goverman. 2006. Active induction of experimental I-A binding at MOG40–46), making this region ultimately allergic encephalomyelitis. Nat. Protoc. 1: 1810–1819. immunodominant in this haplotype (Supplemental Fig. 3F) (50, 11. Goverman, J. 2009. Autoimmune T cell responses in the central nervous system. Nat. Rev. Immunol. 9: 393–407. 64). The finding that the residues 41–42 are particularly suscep- 12. Nakahara, J., S. Aiso, and N. Suzuki. 2010. Autoimmune versus oligoden- tible to cathepsin L (in a reductive microenvironment) (Fig. 6C, drogliopathy: the pathogenesis of multiple sclerosis. Arch. Immunol. Ther. Exp. 6D) may not only underpin the disconnect between the abilities of (Warsz.) 58: 325–333. 13. Chastain, E. M., D. S. Duncan, J. M. Rodgers, and S. D. Miller. 2011. The role of DCs and macrophages to activate T cells in NOX2-deficient mice, antigen presenting cells in multiple sclerosis. Biochim. Biophys. Acta 1812: 265–274. but may also give insight into why MOG escapes tolerization in 14. Almolda, B., B. Gonzalez, and B. Castellano. 2011. Antigen presentation in I-Ab mice. Because thymic epithelial cells express high levels of EAE: role of microglia, macrophages and dendritic cells. Front Biosci (Land- mark Ed) 16: 1157–1171. cathepsin L (43), efficient cleavage within MOG40–46 may lead to 15. Lodygin, D., F. Odoardi, C. Schla¨ger, H. Ko¨rner, A. Kitz, M. Nosov, J. van den Downloaded from € inefficiencies of MOG35–55 presentation and limit the deletion of Brandt, H. M. Reichardt, M. Haberl, and A. Flugel. 2013. A combination of MOG-autoreactive CD4+ T cells. Indeed, the susceptibility of fluorescent NFATand H2B sensors uncovers dynamics of T cell activation in real time during CNS autoimmunity. Nat. Med. 19: 784–790. myelin basic protein to asparagine endopeptidase led Manoury 16. Zal, T., A. Volkmann, and B. Stockinger. 1994. Mechanisms of tolerance in- et al. (65) to propose a similar mechanism of autoreactivity of duction in major histocompatibility complex class II-restricted T cells specific myelin basic protein in certain haplotypes of mice. for a blood-borne self-antigen. J. Exp. Med. 180: 2089–2099. 17. Karasuyama, H., and F. Melchers. 1988. Establishment of mouse cell lines which

The relationship between levels of proteolysis and efficiency of constitutively secrete large quantities of interleukin 2, 3, 4 or 5, using modified http://www.jimmunol.org/ Ag presentation has been the topic of many studies during the past cDNA expression vectors. Eur. J. Immunol. 18: 97–104. 18. van der Veen, R. C., T. A. Dietlin, F. M. Hofman, L. Pen, B. H. Segal, and decade (1, 2, 7, 66). However, most attention has been paid to S. M. Holland. 2000. Superoxide prevents -mediated suppression of determining the efficiency of processing rather than the patterns of helper T lymphocytes: decreased autoimmune encephalomyelitis in nicotin- processing. Because the Th arm of the immune system relies on amide adenine dinucleotide phosphate oxidase knockout mice. J. Immunol. 164: 5177–5183. the faithful reproduction of MHC-II–restricted peptides from any 19. Agrawal, S. M., C. Silva, W. W. Tourtellotte, and V. W. Yong. 2011. EMMPRIN: given Ag during education, activation, and effector function (all a novel regulator of leukocyte transmigration into the CNS in multiple sclerosis performed by different APCs), subtle changes to Ag processing and experimental autoimmune encephalomyelitis. J. Neurosci. 31: 669–677. 20. VanderVen, B. C., R. M. Yates, and D. G. Russell. 2009. Intraphagosomal chemistries could lead to a breakdown in tolerance or conversely measurement of the magnitude and duration of the oxidative burst. Traffic 10: deficiencies in the Th cell response. Interestingly, NOX2 defi- 372–378. by guest on September 28, 2021 ciencies in humans (chronic granulomatous disease) have been 21. Yates, R. M., A. Hermetter, and D. G. Russell. 2005. The kinetics of phagosome maturation as a function of phagosome/lysosome fusion and acquisition of hy- associated with altered susceptibilities to a number of autoimmune drolytic activity. Traffic 6: 413–420. disorders, including systemic lupus erythematous, rheumatoid 22. Russell, D. G., and R. M. Yates. 2007. TLR signalling and phagosome matu- ration: an alternative viewpoint. Cell. Microbiol. 9: 849–850. arthritis, colitis, and Crohn’s disease (67–73). In this study, we 23. Yates, R. M., and D. G. Russell. 2008. Real-time spectrofluorometric assays for demonstrate that the redox state of phagosomes is a determinant of the lumenal environment of the maturing phagosome. Methods Mol. Biol. 445: Ag processing fidelity and its modification by NOX2 can ulti- 311–325. 24. Yates, R. M., A. Hermetter, and D. G. Russell. 2009. Recording phagosome mately impact Th cell–driven disease processes. maturation through the real-time, spectrofluorometric measurement of hydrolytic activities. Methods Mol. Biol. 531: 157–171. Acknowledgments 25. Underhill, D. M., M. Bassetti, A. Rudensky, and A. Aderem. 1999. Dynamic interactions of macrophages with T cells during antigen presentation. J. Exp. We thank Jason Lemmon, Sibapriya Chaudhuri, and Dr. Joanna Rybicka for Med. 190: 1909–1914. their logistical support. We also thank Bill Zaun for artistic services. 26. Alfonso, C., G. S. Williams, J. O. Han, J. A. Westberg, O. Winqvist, and L. Karlsson. 2003. Analysis of H2-O influence on antigen presentation by B cells. J. Immunol. 171: 2331–2337. Disclosures 27. Shastri, N., A. Miller, and E. E. Sercarz. 1984. The expressed T cell repertoire is The authors have no financial conflicts of interest. hierarchical: the precise focus of lysozyme-specific T cell clones is dependent upon the structure of the immunogen. J. Mol. Cell. Immunol. 1: 369–379. 28. Alfonso, C., J. O. Han, G. S. Williams, and L. Karlsson. 2001. The impact of H2- DM on humoral immune responses. J. Immunol. 167: 6348–6355. References 29. Norton, W. T., and S. E. Poduslo. 1973. Myelination in rat brain: method of 1. Delamarre, L., M. Pack, H. Chang, I. Mellman, and E. S. Trombetta. 2005. myelin isolation. J. Neurochem. 21: 749–757. Differential lysosomal proteolysis in antigen-presenting cells determines antigen 30. Ciabattini, A., E. Pettini, P. Andersen, G. Pozzi, and D. Medaglini. 2008. Pri- fate. Science 307: 1630–1634. mary activation of antigen-specific naive CD4+ and CD8+ T cells following 2. Delamarre, L., R. Couture, I. Mellman, and E. S. Trombetta. 2006. Enhancing intranasal vaccination with recombinant . Infect. Immun. 76: 5817– immunogenicity by limiting susceptibility to lysosomal proteolysis. J. Exp. Med. 5825. 203: 2049–2055. 31. Hori, S., T. Takahashi, and S. Sakaguchi. 2003. Control of autoimmunity by 3. Savina, A., C. Jancic, S. Hugues, P. Guermonprez, P. Vargas, I. C. Moura, naturally arising regulatory CD4+ T cells. Adv. Immunol. 81: 331–371. A. M. Lennon-Dume´nil, M. C. Seabra, G. Raposo, and S. Amigorena. 2006. 32. Atif, S. M., S. Uematsu, S. Akira, and S. J. McSorley. 2014. CD1032CD11b+ NOX2 controls phagosomal pH to regulate antigen processing during cross- dendritic cells regulate the sensitivity of CD4 T-cell responses to bacterial fla- presentation by dendritic cells. Cell 126: 205–218. gellin. Mucosal Immunol. 7: 68–77. 4. Balce, D. R., B. Li, E. R. Allan, J. M. Rybicka, R. M. Krohn, and R. M. Yates. 33. Wood, K. J., H. Ushigome, M. Karim, A. Bushell, S. Hori, and S. Sakaguchi. 2011. Alternative activation of macrophages by IL-4 enhances the proteolytic 2003. Regulatory cells in transplantation. Novartis Found. Symp. 252: 177–188. capacity of their phagosomes through synergistic mechanisms. Blood 118: 4199– 34. Krishnamoorthy, G., A. Saxena, L. T. Mars, H. S. Domingues, R. Mentele, 4208. A. Ben-Nun, H. Lassmann, K. Dornmair, F. C. Kurschus, R. S. Liblau, and 5. Rybicka, J. M., D. R. Balce, M. F. Khan, R. M. Krohn, and R. M. Yates. 2010. H. Wekerle. 2009. Myelin-specific T cells also recognize neuronal autoantigen in NADPH oxidase activity controls phagosomal proteolysis in macrophages a transgenic mouse model of multiple sclerosis. Nat. Med. 15: 626–632. through modulation of the lumenal redox environment of phagosomes. Proc. 35. Bettelli, E., M. Pagany, H. L. Weiner, C. Linington, R. A. Sobel, and Natl. Acad. Sci. USA 107: 10496–10501. V. K. Kuchroo. 2003. Myelin oligodendrocyte glycoprotein-specific T cell re- The Journal of Immunology 5001

ceptor transgenic mice develop spontaneous autoimmune optic neuritis. J. Exp. 56. Ben-Nun, A., N. Kerlero de Rosbo, N. Kaushansky, M. Eisenstein, L. Cohen, Med. 197: 1073–1081. J. F. Kaye, and I. Mendel. 2006. Anatomy of T cell autoimmunity to myelin 36. Lal, J., and M. Smriti. 2008. Should drug-pre-filled syringes be made available oligodendrocyte glycoprotein (MOG): prime role of MOG44F in selection and with disposable spinal needles? Acta Anaesthesiol. Scand. 52: 1175–1176. control of MOG-reactive T cells in H-2b mice. Eur. J. Immunol. 36: 478–493. 37. Choi, S. H., D. Y. Lee, E. S. Chung, Y. B. Hong, S. U. Kim, and B. K. Jin. 2005. 57. Beers, C., K. Honey, S. Fink, K. Forbush, and A. Rudensky. 2003. Differential Inhibition of -induced microglial activation and NADPH oxidase by regulation of cathepsin S and cathepsin L in interferon g-treated macrophages. J. minocycline protects dopaminergic neurons in the substantia nigra in vivo. J. Exp. Med. 197: 169–179. Neurochem. 95: 1755–1765. 58. Honey, K., M. Duff, C. Beers, W. H. Brissette, E. A. Elliott, C. Peters, M. Maric, 38. Williams, J. L., A. P. Kithcart, K. M. Smith, T. Shawler, G. M. Cox, and P. Cresswell, and A. Rudensky. 2001. Cathepsin S regulates the expression of C. C. Whitacre. 2011. Memory cells specific for myelin oligodendrocyte gly- cathepsin L and the turnover of g-interferon-inducible lysosomal thiol reductase coprotein (MOG) govern the transfer of experimental autoimmune encephalo- in B lymphocytes. J. Biol. Chem. 276: 22573–22578. myelitis. J. Neuroimmunol. 234: 84–92. 59. Hsing, L. C., and A. Y. Rudensky. 2005. The lysosomal cysteine proteases in 39. Yang, K., J. L. Vega, M. Hadzipasic, J. P. Schatzmann Peron, B. Zhu, Y. Carrier, MHC class II antigen presentation. Immunol. Rev. 207: 229–241. S. Masli, L. V. Rizzo, and H. L. Weiner. 2009. Deficiency of thrombospondin-1 60. Lanzavecchia, A., and F. Sallusto. 2001. Regulation of T cell immunity by reduces Th17 differentiation and attenuates experimental autoimmune enceph- dendritic cells. Cell 106: 263–266. alomyelitis. J. Autoimmun. 32: 94–103. 61. Yi, L., Q. Liu, M. S. Orandle, S. Sadiq-Ali, S. M. Koontz, U. Choi, F. J. Torres- 40. Racke, M. K., D. Burnett, S. H. Pak, P. S. Albert, B. Cannella, C. S. Raine, Velez, and S. H. Jackson. 2012. p47phox directs murine macrophage cell fate D. E. McFarlin, and D. E. Scott. 1995. Retinoid treatment of experimental al- decisions. Am. J. Pathol. 180: 1049–1058. lergic encephalomyelitis. IL-4 production correlates with improved disease 62. Choi, S. H., S. Aid, H. W. Kim, S. H. Jackson, and F. Bosetti. 2012. Inhibition of course. J. Immunol. 154: 450–458. NADPH oxidase promotes alternative and anti-inflammatory microglial activa- 41. Martin, W. D., G. G. Hicks, S. K. Mendiratta, H. I. Leva, H. E. Ruley, and L. Van tion during neuroinflammation. J. Neurochem. 120: 292–301. Kaer. 1996. H2-M mutant mice are defective in the peptide loading of class II 63. Hultqvist, M., P. Olofsson, J. Holmberg, B. T. Ba¨ckstro¨m, J. Tordsson, and molecules, antigen presentation, and T cell repertoire selection. Cell 84: 543– R. Holmdahl. 2004. Enhanced autoimmunity, arthritis, and encephalomyelitis in 550. mice with a reduced oxidative burst due to a mutation in the Ncf1 gene. Proc. 42. Banchereau, J., and R. M. Steinman. 1998. Dendritic cells and the control of Natl. Acad. Sci. USA 101: 12646–12651. immunity. Nature 392: 245–252. 64. von Budingen,€ H. C., N. Tanuma, P. Villoslada, J. C. Ouallet, S. L. Hauser, and Downloaded from 43. Nakagawa, T., W. Roth, P. Wong, A. Nelson, A. Farr, J. Deussing, C. P. Genain. 2001. Immune responses against the myelin/oligodendrocyte J. A. Villadangos, H. Ploegh, C. Peters, and A. Y. Rudensky. 1998. Cathepsin L: glycoprotein in experimental autoimmune demyelination. J. Clin. Immunol. critical role in Ii degradation and CD4 T cell selection in the thymus. Science 21: 155–170. 280: 450–453. 65. Manoury, B., D. Mazzeo, L. Fugger, N. Viner, M. Ponsford, H. Streeter, 44. Mintseris, J., B. Pierce, K. Wiehe, R. Anderson, R. Chen, and Z. Weng. 2007. G. Mazza, D. C. Wraith, and C. Watts. 2002. Destructive processing by aspar- Integrating statistical pair potentials into protein complex prediction. Proteins agine endopeptidase limits presentation of a dominant T cell epitope in MBP. 69: 511–520. Nat. Immunol. 3: 169–174.

45. Mellman, I., S. J. Turley, and R. M. Steinman. 1998. Antigen processing for 66. Lennon-Dume´nil, A. M., A. H. Bakker, R. Maehr, E. Fiebiger, H. S. Overkleeft, http://www.jimmunol.org/ amateurs and professionals. Trends Cell Biol. 8: 231–237. M. Rosemblatt, H. L. Ploegh, and C. Lagaudrie`re-Gesbert. 2002. Analysis of 46. Pethe, K., D. L. Swenson, S. Alonso, J. Anderson, C. Wang, and D. G. Russell. protease activity in live antigen-presenting cells shows regulation of the phagosomal 2004. Isolation of Mycobacterium tuberculosis mutants defective in the arrest of proteolytic contents during dendritic cell activation. J. Exp. Med. 196: 529–540. phagosome maturation. Proc. Natl. Acad. Sci. USA 101: 13642–13647. 67. Jacob, C. O., M. Eisenstein, M. C. Dinauer, W. Ming, Q. Liu, S. John, 47. Pillay, C. S., E. Elliott, and C. Dennison. 2002. Endolysosomal proteolysis and F. P. Quismorio, Jr., A. Reiff, B. L. Myones, K. M. Kaufman, et al. 2012. Lupus- its regulation. Biochem. J. 363: 417–429. associated causal mutation in cytosolic factor 2 (NCF2) brings unique 48. Nelson, P. A., M. Khodadoust, T. Prodhomme, C. Spencer, J. C. Patarroyo, insights to the structure and function of NADPH oxidase. Proc. Natl. Acad. Sci. M. Varrin-Doyer, J. D. Ho, R. M. Stroud, and S. S. Zamvil. 2010. Immunodo- USA 109: E59–E67. minant T cell determinants of aquaporin-4, the autoantigen associated with 68. De Ravin, S. S., N. Naumann, E. W. Cowen, J. Friend, D. Hilligoss, neuromyelitis optica. PLoS ONE 5: e15050. M. Marquesen, J. E. Balow, K. S. Barron, M. L. Turner, J. I. Gallin, and 49. Verspurten, J., K. Gevaert, W. Declercq, and P. Vandenabeele. 2009. SitePre- H. L. Malech. 2008. Chronic granulomatous disease as a risk factor for auto-

dicting the cleavage of proteinase substrates. Trends Biochem. Sci. 34: 319–323. immune disease. J. Allergy Clin. Immunol. 122: 1097–1103. by guest on September 28, 2021 50. Carrillo-Vico, A., M. D. Leech, and S. M. Anderton. 2010. Contribution of 69. Lee, B. W., and H. K. Yap. 1994. Polyarthritis resembling juvenile rheumatoid ar- myelin autoantigen citrullination to T cell autoaggression in the central nervous thritis in a girl with chronic granulomatous disease. Arthritis Rheum. 37: 773–776. system. J. Immunol. 184: 2839–2846. 70. Matute, J. D., A. A. Arias, N. A. Wright, I. Wrobel, C. C. Waterhouse, X. J. Li, 51. Yan, Y., H. Laroui, S. A. Ingersoll, S. Ayyadurai, M. Charania, S. Yang, C. C. Marchal, N. D. Stull, D. B. Lewis, M. Steele, et al. 2009. A new genetic G. Dalmasso, T. S. Obertone, H. Nguyen, S. V. Sitaraman, and D. Merlin. 2011. subgroup of chronic granulomatous disease with autosomal recessive mutations Overexpression of Ste20-related proline/alanine-rich kinase exacerbates exper- in p40phox and selective defects in neutrophil NADPH oxidase activity. Blood imental colitis in mice. J. Immunol. 187: 1496–1505. 114: 3309–3315. 52. Trave´s, P. G., R. Lo´pez-Fontal, A. Luque, and S. Hortelano. 2011. The tumor 71. Olsson, L. M., A. K. Lindqvist, H. Ka¨llberg, L. Padyukov, H. Burkhardt, suppressor ARF regulates innate immune responses in mice. J. Immunol. 187: L. Alfredsson, L. Klareskog, and R. Holmdahl. 2007. A case-control study of 6527–6538. rheumatoid arthritis identifies an associated single nucleotide polymorphism in 53. Anderson, A. C., R. Chandwaskar, D. H. Lee, J. M. Sullivan, A. Solomon, the NCF4 gene, supporting a role for the NADPH-oxidase complex in autoim- R. Rodriguez-Manzanet, B. Greve, R. A. Sobel, and V. K. Kuchroo. 2012. A munity. Arthritis Res. Ther. 9: R98. transgenic model of central nervous system autoimmunity mediated by CD4+ 72. Rioux, J. D., R. J. Xavier, K. D. Taylor, M. S. Silverberg, P. Goyette, A. Huett, and CD8+ T and B cells. J. Immunol. 188: 2084–2092. T. Green, P. Kuballa, M. M. Barmada, L. W. Datta, et al. 2007. Genome-wide 54. McNeil, L. K., and B. D. Evavold. 2003. TCR reserve: a novel principle of CD4 association study identifies new susceptibility loci for Crohn disease and T cell activation by weak ligands. J. Immunol. 170: 1224–1230. implicates autophagy in disease pathogenesis. Nat. Genet. 39: 596–604. 55. Petersen, T. R., E. Bettelli, J. Sidney, A. Sette, V. Kuchroo, and B. T. Ba¨ckstro¨m. 73. Gardiner, G. J., S. N. Deffit, S. McLetchie, L. Pe´rez, C. C. Walline, and 2004. Characterization of MHC- and TCR-binding residues of the myelin oli- J. S. Blum. 2013. A role for NADPH oxidase in antigen presentation. Front. godendrocyte glycoprotein 38–51 peptide. Eur. J. Immunol. 34: 165–173. Immunol. 4: 295.