Introduction
The inflammatory disease multiple sclerosis (MS) and its animal model experimental autoimmune encephalomyelitis (EAE) are characterized by peripheral immune cell infiltration into the central nervous system (CNS), which ultimately leads to demyelination and axonal damage in the brain and spinal cord [
38]. Immune cell infiltration in MS/EAE is a consequence of focal blood–brain barrier (BBB) breakdown, which enables the uncontrolled influx of peripheral cells and molecules. The BBB is a tightly regulated vascular barrier, composed of specialized endothelial cells (ECs), connected through intercellular tight junctions and adherens junctions, thereby controlling traffic of cells and molecules across the BBB [
79]. Disorganization of these junctional proteins together with upregulation of cell adhesion molecules, such as intercellular adhesion molecule-1 (Icam-1) and vascular adhesion molecule-1 (Vcam-1), is associated with peripheral cell adhesion and migration into the CNS in MS and EAE [
7].
Interleukin (IL)-1 is an important inflammatory cytokine, strongly implicated as an effector molecule in MS and EAE [
23]. The IL-1α and IL-1β isoforms signal through a common receptor complex, composed of the IL-1 receptor type 1 (IL-1R1) and its accessory protein (IL-1RAcP). The downstream signaling cascade induces activation of the mitogen-activated protein kinases (MAPK) and mobilization of the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) family of transcription factors, which trigger the transcription of proinflammatory genes [
74]. Macrophages, dendritic cells and neutrophils are the major sources of IL-1β production in EAE [
22,
37,
42,
80]. Levesque and colleagues identified neutrophils and monocyte-derived macrophages as the primary cell types expressing IL-1β in EAE upon their transmigration into the spinal cord parenchyma [
39]. In contrast to IL-1α, mice deficient for
Il1b are resistant to EAE, identifying IL-1β as a critical mediator of EAE development [
39,
61]. Likewise,
Il1r1-deficient mice are also resistant to EAE, which correlates with the failure to induce autoreactive T helper (T
H) 17 cells [
19,
54,
61,
77]. Importantly, mice with a global
Il1r1 deficiency are less susceptible to EAE, as compared to mice with a T cell-specific
Il1r1 deletion, suggesting that other cell types may play a pathogenic role in response to IL-1 in EAE [
61]. Furthermore, using bone marrow chimeric mice, it has been shown that IL-1R1 signaling on radiation-resistant cells promotes EAE development [
39,
55]. Together, these results identify a pathogenic role for IL-1β signaling in EAE that goes beyond the induction of autoreactive CD4 T cells.
Cellular targets of IL-1 within the CNS have been controversially debated. On one hand, BBB endothelial cells (BBB-ECs) have been reported to be the major cell type expressing IL-1R1 [
35,
39,
45], thereby driving leukocyte recruitment and compromising BBB integrity [
16,
63]. Astrocytes and microglia have also been found to respond to IL-1β and were, therefore, suggested to be key players during neuroinflammation [
11,
15,
57,
64]. IL-1β induces a rapid proinflammatory response in astrocyte cultures, leading to the expression of cytokines, chemokines, adhesion molecules, and matrix metalloproteinases [
6,
36,
78,
81]. Furthermore, IL-1β induces astrocytic expression of angiogenic factors, including vascular endothelial growth factor A (VEGF-A), which is involved in vessel plasticity and angiogenesis, thereby representing an important driver of BBB disruption in EAE [
9,
75].
A comprehensive understanding in mechanisms of IL-1-mediated signaling in CNS-resident cells, including BBB-ECs, astrocytes and microglia, during the course of EAE is far beyond completed. To study the cell type-specific role of IL-1 signaling, we made use of conditional gene targeting to obtain novel mouse strains, where the IL-1R1 is specifically deleted in different CNS-resident cell populations. In this study, we show that the deletion of IL-1R1 in BBB-ECs reduced EAE severity, whereas IL-1 signaling in astrocytes or microglia was redundant for EAE development. Differential gene expression analysis of BBB-ECs at the preclinical stage of EAE identified that IL-1 signaling suppressed the expression of heme oxygenase-1 (HO-1), a core effector molecule that prevents tissue damage under stress conditions [
53]. We found that the deletion of HO-1 specifically in BBB-ECs resulted in an enhanced EAE disease. Targeting the expression of HO-1 exclusively to BBB-ECs in vivo revealed its protective function specifically on BBB-ECs. Moreover, our findings suggest a functional crosstalk between IL-1 signaling and HO-1, counter-regulating the expression of adhesion molecules that promote the pathogenesis of EAE. Our results, therefore, identify BBB-EC HO-1 as a potential target for therapeutic interventions in the course of neuroinflammatory processes associated with BBB breakdown.
Materials and methods
Mice
Il1r1fl/fl mice [
2] were crossed to
Gfap-Cre [
10],
Cx3cr1-CreERT2 [
85],
Slco1c1-CreERT2 mice [
65], to obtain IL-1R1
GFAP, IL-1R1
CX3CR1, IL-1R1
SLC. Cre-negative
Il1r1fl/fl (IL-1R1
WT) littermates and complete
Il1r1 deficient mice (IL-1R1
−/−) [
60,
61] were used as control animals.
RiboTag (
RPL22HA) mice [
66] were crossed to the IL-1R1
SLC mouse strain to obtain IL-1R1/HA
SLC mice, or crossed to the
Slco1c1-CreERT2 mice (HA
SLC mice), which were used as controls. Conditional
ROSA-flox-human-HMOX-1 overexpression mice (
hHO-1oefl/fl) [
13], were bred to
Slco1c1-CreERT2 mice to obtain hHO-1oe
SLC mice. Cre-negative
hHO-1oefl/fl (hHO-1oe
WT) were used as control mice.
HO-1fl/fl (B6J.129P2-Hmox1 < tm1Mym > , purchased from RIKEN) were crossed to
Slco1c1-CreERT2 mice to obtain HO-1
SLC mice, whereas Cre-negative
HO-1fl/fl (HO-1
WT) mice were used as controls. Mice were housed and bred under specific pathogen-free (SPF) conditions, with a 12 h light/dark cycle and unlimited access to food and water at the University Medical Center Mainz. For all experiments, 10–14-week-old age- and sex-matched littermate animals were used, in accordance with the guidelines of the central animal facility institution (Translational Animal Research Center, University of Mainz).
Tamoxifen treatment
To induce Cre expression in BBB-ECs, 5–7-week-old Slco1c1-CreERT2-carrying mice and their Cre-negative littermate controls were intraperitoneally injected with tamoxifen (2 mg in olive oil containing 5% ethanol) on 5 consecutive days. Cx3cr1-CreERT2 mice and their Cre-negative littermates received two subcutaneous tamoxifen injections (2 mg in olive oil containing 5% ethanol) on postnatal days 12 and 14.
EAE induction and scoring
Mice were immunized subcutaneously at the tail base with 50 μg MOG35-55 peptide (GenScript) emulsified in complete Freund’s adjuvant (CFA, BD Biosciences) supplemented with 10 mg/ml Mycobacterium tuberculosis H37RA (BD Biosciences). In addition, mice were intraperitoneally injected with 100 ng of pertussis toxin (PTx) (List Biological Labs) on day 0 and day 2 post-immunization. Mice were observed daily, to monitor body weight and EAE clinical symptoms on a scale from 0 to 4 as follows: 0, no disease; 0.5: limb tail; 1: paralyzed tail; 1.5: weakened righting reflex; 2: no righting reflex; 3: partial paralysis of hind legs; 3.5: paralysis of one hind leg; 4: paralysis of both hind legs.
Lymphocyte cell isolation from CNS and lymphoid tissue
Mice were sacrificed with CO2 prior to dissection of CNS and lymphoid organs. For CNS lymphocyte isolation, CNS tissue was dissected from mice transcardially perfused with 0.9% NaCl solution (Sigma-Aldrich) and digested with 2 mg/ml collagenase II (Gibco) and 2 μg/ml DNase I (Roche) for 20 min at 37 °C and subsequently homogenized with a 18-G needle. Cells were then separated using a 70–37–30% Percoll (Sigma-Aldrich) gradient centrifugation for 40 min, 500×g at 10 °C. Cells at the 70/37% interphase were carefully collected and washed in PBS containing 2% FCS (Thermo Scientific) (PBS/FCS) prior to 10 min centrifugation at 500×g. To specifically characterize MOG-specific T cells, cells were plated in 96-well U-bottom plates and re-activated with 20 μg/ml MOG35-55 peptide and brefeldin A (Sigma-Aldrich) for 6 h at 37 °C. Afterwards, cells were harvested and stained for flow cytometry analysis. T cell expression of CD154 (CD40L) was assessed as indication of recent activation, serving as a surrogate marker for MOG antigen specificity. CD40L+ cells were further analyzed for their cytokine expression. For peripheral lymphoid cell isolation, lymph nodes and spleen were mechanically dissociated in PBS/FCS and filtered through a 40 μm cell strainer. Erythrocytes were removed by Ammonium-chloride-potassium lysis (1 M Tris, 1 M MOPS, 20 mM EDTA, 2% SDS, pH 7.7) and lymphoid cells afterwards washed with PBS/FCS. Cells were counted and equal amounts were re-activated in vitro in the presence of MOG35-55 peptide to assess the proportion of MOG-specific T cells as indicated above. For IL-1β detection, cells were incubated for 4 h in the presence of 2 μM monensin (BioLegend) and 500 ng/ml LPS (Sigma),
Endothelial cell isolation for flow cytometry
For CNS EC-isolation, mice were sacrificed and transcardially perfused as described above. The dissected CNS tissue was digested with 2 mg/ml papain (Sigma-Aldrich) solution containing 40 μg/ml DNase I for 30 min at 37 °C. During incubation, tissue was mechanically homogenized using the gentleMACS™ Dissociator (Miltenyi). The resulting cell suspension was filtered through a 70 μm cell strainer and centrifuged with a 20% Percoll gradient for 30 min, 300×g at 15 °C. The pellet, containing a mix of CNS cells, was used for flow cytometry staining.
Flow cytometry analyses
Before antibody staining, Fc receptors were blocked for 20 min to prevent unspecific antibody binding using Fc-block (BioXCell). Single cell suspensions of isolated lymphocytes from the CNS, lymph nodes and spleen were stained for 30 min at 4 °C on the cell surface with antibodies against CD90.2 APC-Cy7 (53-2.1, rat monoclonal, 1:1000, eBioscience), CD4 PerCP (GK1.5, rat monoclonal, 1:200, Biolegend), CD44 FITC (IM7, rat monoclonal, 1:200, eBioscience), CD45 BV510 (30-F11, rat monoclonal, 1:200, Biolegend), CD11b PECy7 (M1/70, rat monoclonal, 1:1000, eBioscience), Ly6c V450 (AL-21, rat monoclonal, 1:100 BD Bioscience), Ly6c PerCP (HK1.4, rat monoclonal, 1:200 Biolegend), Ly6G PE (1A8, rat monoclonal, 1:2000 Biolegend).
Afterwards, where indicated, cells were fixed and permeabilized with Cytofix/Cytoperm (BD Bioscience) and stained for 30 min at 4 °C with intracellular antibodies. To specifically gate on MOG-specific cells, cells were stained with anti-CD154 (CD40L) APC (MR1, hamster monoclonal, 1:200, Biolegend) and cytokine production was assessed by staining with IL-17A V450 (eBio17B7, rat monoclonal, 1:300, eBioscience), IFN-y PECy7 (XMG1.2, rat monoclonal, 1:400, eBioscienc) and GM-CSF PE (MP1-22E9, rat monoclonal, 1:200, eBioscience).
For EC staining, single cell suspension of isolated CNS cells were stained for 30 min at 4 °C for the following surface markers: CD31 PE, CD31 APC or CD31 FITC (all MEC 13.3, rat monoclonal, 1:100, BD), Icam-1 APC (YN1/1.7.4, rat monoclonal, 1:300, Biolegend), Vcam-1 PECy7 (MVCAM.A, rat monoclonal, 1:300, Biolegend), IL-1R1 PE (JAMA-147, rat monoclonal, 1:200, Biolegend). For intracellular staining, CNS cells were fixed and permeabilized using Cytofix/Cytoperm kit and stained for 30 min at 4 °C against Darc PE (Met1-Pro61, goat monoclonal, 1:30, R&D) and HO-1 (HO-1-1, mouse monoclonal, 1:300, Enzolifescience). Afterwards, samples were washed with PBS/FCS and incubated for 30 min at 4 °C with secondary antibody IgG Alexa488 (goat anti mouse, 1:200, Jackson) for HO-1 staining. Stained cells were acquired with a FACSCanto II cytometer (BD Biosciences) using FACS Diva software (BD Biosciences). Flow cytometry data were analyzed with FlowJo software version 9 or higher (TreeStar). For all analysis, doublets (FSC and SSC properties) and dead cells (dye inclusion) were excluded.
Use of the RiboTag method to isolate RNA
BBB-EC ribosomes from spinal cord of EAE immunized mice were isolated using RiboTag technology as described previously [
66]. Briefly, spinal cord tissue was removed, weighed and homogenized in homogenization buffer (200 U/ml RNasin, Promega; 1 mg/ml heparin, Sigma-Aldrich; 100 μg/ml cycloheximide, Sigma-Aldrich; protease inhibitor mixture, Sigma-Aldrich; 1 mM Dithiothreitol, Boehringer Mannheim GmbH), using 2–3% weight per volume for immune-precipitation (IP). The homogenate was centrifuged at 1000×
g for 10 min at 4 °C and the supernatant was mixed with HA tag antibody (C29F4, rabbit monoclonal, 2 μg/ml, cell signaling) and rotated for 4 h at 4 °C. Afterwards, the spinal cord homogenate was mixed with 200 μl Protein G magnetic beads (Invitrogen) overnight at 4 °C. The following day, the samples were placed into a magnetic rack and supernatant was removed before washing the beads three times for 10 min in high salt buffer (50 mM Tris, pH 7.5, 300 mM KCl, 12 mM MgCl2, 10% NP-40, 1 mM dithiothreitol and 100 μg/ml cycloheximide which was supplied by Sigma-Aldrich). Immediately after removal of the final high salt wash buffer, lysis buffer (Qiagen) was added and samples were vortexed. Magnetic beads were removed by placing samples again into a magnetic rack and mRNA was extracted.
RNA isolation
mRNA was prepared using the Rneasy® Plus Micro Kit from Qiagen according to the manufacturer’s guidelines. The concentration of nucleic acids was determined by measuring the absorption at 260 nm and 280 nm using NanoDrop™ (Thermo Fisher Scientific). Equal amounts of RNA for all samples were used for following assays.
Quantitative real-time PCR
cDNA was synthesized using 200–1000 ng of total RNA with the superscript II reverse transcriptase (Invitrogen) and subsequently used for qPCR, which was performed with the StepOnePlus™ Real-Time PCR System (Life Technologies) using SYBR Green (Promega). Fold enrichment was calculated using the Delta–Delta CT method normalized to hypoxanthin-guanin-phosphoribosyltransferase (HPRT) as house-keeping reference. Primer for HPRT, Hmox1, Icam1, Vcam1 and IL-1R1 were purchased from Qiagen. Ackr1 primer (forward: 5′-CTT CAC CTT GGG ACT CAG TGT-3′; reverse: 5′-GAC TGG CAG CCC TAA GAG G-3′) were self-designed using Primer Express 3.0 software and were synthesized by Metabion.
Next-generation sequencing (NGS)
RNA of RiboTag isolated tissue was quantified with a Qubit 2.0 fluorometer (Invitrogen) and the quality was assessed on a Bioanalyzer 2100 (Agilent) using an RNA 6000 Nano chip (Agilent). Samples with an RNA integrity number (RIN) of > 8 were used for library preparation. Barcoded mRNA-seq cDNA libraries were prepared from 200 ng of total RNA using NEBNext® Poly(A) mRNA Magnetic Isolation Module and NEBNext® Ultra™ RNA Library Prep Kit for Illumina® according to manufacturer’s guidelines. Quantity was assessed using Invitrogen’s Qubit HS DNA assay kit and library size was determined using Agilent’s 2100 Bioanalyzer HS DNA assay. Barcoded RNA-Seq libraries were onboard clustered using HiSeq® Rapid SR Cluster Kit v2 using 8 pM and 59bps were sequenced on the Illumina HiSeq2500 using HiSeq® Rapid SBS Kit v2 (59 Cycle). The raw output data of the HiSeq® were preprocessed according to the Illumina standard protocol.
NGS data analysis
Quality control on the sequencing data was performed with the FastQC tool (version 0.11.2,
https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). RNA sequencing reads were aligned to the ENSEMBL Mus_musculus.GRCm38 reference genome. The corresponding annotation (ENSEMBL v76) was also retrieved from ENSEMBL FTP website. The STAR aligner (version 2.4.0j) was used to perform mapping to the reference genome. Alignments were processed with the featureCounts function of the Rsubread package [
41], using the annotation file also used for supporting the alignment. Differential expression analysis was performed with DESeq2 package (version 1.22.1) [
48], setting the false discovery rate (FDR) cutoff to 0.1. Accurate estimation of the effect sizes (in terms of log fold change) is performed using the apeglm shrinkage estimator (version 1.4.1) [
88]. Gene expression profiles were plotted as heatmaps (color-coded
z-scores for the expression values, after regularized logarithm transformation) with the R programming language (
https://www.R-project.org/) and the heatmap package (version 1.0.12).
Principal component analysis was performed using the pcaExplorer package version2.8.0 [
51]. To highlight the differences of the expression values between the two groups, MA-Plots were generated with the R programming language. Further analyses included Gene Ontology pathway enrichment of DEG using topGO [
5] (v2.34.0) and goseq (v1.34.0) [
86] (with all expressed genes set as background) were performed using the ideal package (version 1.6.0) [
52].
Evans Blue assay
2% Evans Blue (Sigma-Aldrich) solution was intraperitoneally injected at onset of EAE disease. After 6 h, perfused spinal cords were weighed and homogenized in 1 ml 50% trichloroacetic acid (Sigma-Aldrich) and centrifuged for 10 min at 1000×g and 4 °C. Evans Blue concentration in the supernatant was measured on a plate reader (Infinite M200, Tecan Life Sciences) (excitation at 620; emission at 680) and quantified according to a freshly prepared standard curve.
Mice were intravenously injected with 5 mg 20 kDa FITC-Dextran (Sigma-Aldrich) in 0.9% NaCl at day 10 after EAE induction (before clinical onset). After 15 min, perfused spinal cords were weighed and homogenized in 1 ml 60% trichloroacetic acid using a Dounce homogenizer (Kimble Chase Life Science and Research Products, LLC). Samples were centrifuged at 1000×g for 10 min. Fluorescence was measured in the supernatant on a fluorescence plate reader (Tecan Life Technologies) with 470 nm excitation and 520 nm emission, and quantified according to standard curve per mg of tissue.
Immunofluorescence
For fluorescence microscopy, mice were deeply anesthetized with isoflurane and perfused with 0.9% NaCl followed by 30 ml of 4% paraformaldehyde (PFA) (Sigma-Aldrich) in PBS. Spinal cord was removed and post-fixed overnight in 4% PFA followed by 3 days incubation in 30% sucrose (Sigma-Aldrich). Sections (10 μm) were blocked using 1/1000 immunoblock (Roth) or 5% normal goat serum (NGS) respectively, and 0.5% Triton-X100 (Sigma-Aldrich) in PBS for 1 h at RT. The following primary antibodies were incubated at 4 °C overnight: CD31 (rabbit polyclonal, 1:100, Abcam), HA tag (C29F4, rabbit monoclonal, 1:800, Cell signaling), IgG biotinylated (rabbit polyclonal, 1:500, Vector), CD3 (HH3E, rat monoclonal, 1:100, Dianova), Laminin (rat polyclonal, 1:200, DAKO), CD11b (5C6, rabbit monoclonal, 1:200, Bio Rad formerly Serotec), Iba-1(rabbit polyclonal, 1:200, Synaptic Systems). After washing, sections were stained for 1 h at RT in the dark with the following secondary antibodies: anti-goat IgG Alexa488 (1:250, Invitrogen), anti-rat IgG Alexa 568 (1:250, Invitrogen), anti-mouse IgG Alexa488 (1:200, Jackson) or using streptavidin Cy3 (1:200, Jackson) for biotinylated antibodies. For Myelin staining, sections were stained for 20 min at RT with FluoroMyelin™ Red Fluorescent Myelin Stain (Thermo Fisher).
Afterwards, slides were mounted in Vectashield containing DAPI (Vector). Images were acquired with Olympus Fluoview 100 or Zeiss sp8 and analyzed in Fiji.
In vitro culture of bEnd.3 cell line
The bEnd.3 cells (BEND3) (ATCC® CRL-2299™, RRID:CVCL_0170) were cultured in DMEM with 10% FCS, according to the standard protocol and maintained in 37 °C and 10% CO2. Cells were split every 2–3 days and used at approximately 90% confluency.
Isolation and culture of primary endothelial cells
Cortices from 4–10-week-old WT mice were dissected. Outer vessels and meninges were removed using a dry cotton swap. Tissue was pooled and homogenized in a Dounce homogenizer (Kimble Chase Life Science and Research Products, LLC) in HBSS with 0.1% BSA (Sigma-Aldrich) and 1 mM HEPES (Gibco). The homogenate was mixed with 30% dextran (Sigma-Aldrich) in HBSS, BSA, HEPES and centrifuged at 3000×g for 25 min at 10 °C. The pellet was collected and dextran solution was again centrifuged as before. After second centrifugation, the first and second pellet were pooled and washed with homogenization buffer (HBSS with 0.1% BSA and 1 mM HEPES) and filtered through a 70 μm cell strainer. Capillary enriched filtrate was centrifuged at 1000×g for 10 min at 10 °C and subsequently digested in 2 mg/ml collagenase/dispase (Roche), 10 μg DNase I (Roche) and 147 μg/ml Na-Tosyl-l-lysyl-chloromethane hydrochloride (Sigma-Aldrich) in HBSS for 30 min at 37 °C. Afterwards, vessels were filtered and washed through a 20 μm nylon filter and seeded on 48-well culture dishes coated with Matrigel (ECM Gel from Engelbreth-Holm-Swarm murine sarcoma, supplied by Sigma Aldrich). Culture medium was supplemented with 20% FCS, 2% non-essential amino acids and 5 μ/ml gentamycin. 1 ng/ml human fibroblast growth factor (FGF) (Sigma-Aldrich) and 4 μg/ml puromycin (Gibco) were added for the first 48 h of culture. Prior to each medium change (every second day), medium was supplemented with fresh FGF and cells were kept in culture for 6 days.
In vitro endothelial cell stimulation assay
Cultured cells were incubated with the indicated concentration of hemin (Sigma-Aldrich) to induce HO-1 expression. Simultaneously, cells were stimulated with different concentration of IL-1β (Sinobiological), as indicated, for 16 h. All stimulation experiments were performed in DMEM containing 2% FCS and 25 mM HEPES at 37 °C and 10% CO2. For downstream analysis, cells were either lysed for RNA isolation according to the manufacturer’s protocol of the RNeasy® Plus Micro Kit (Qiagen) or trypsinized to proceed with flow cytometry staining.
Statistical analysis
Statistical analysis was performed with Prism (Graph Pad, Version 5.0b). Differences were evaluated by unpaired, two-tailed Students t test, by one-way ANOVA followed by tests with Bonferroni correction, or two-way ANOVA, as indicated. All values are represented as mean ± SEM and p values were considering significance threshold p < 0.05*; p < 0.01**; and p < 0.001***.
Data availability
The sequencing datasets generated during the current study have been deposited in the Gene Expression Omnibus (GEO) archive, and are available under the accession number GSE146378.
Discussion
The deletion of
Il1b or
Il1r1 in mice leads to EAE resistance, emphasizing the importance of IL-1β as a critical mediator for the induction of CNS autoimmunity [
39,
49,
68]. While it is well established that peripheral IL-1β acts primarily on T
H cells during the induction phase of EAE [
61,
77], the cellular targets of this cytokine in the CNS are not well described. Although existing literature supports the notion that IL-1β may act directly on astrocytes and microglia [
11,
58,
64], the response of these cells to IL-1β and the resulting consequences for EAE pathogenesis are still under debate. Using transgenic mouse lines in which
Il1r1 is deleted in specific CNS cellular subsets, we show that IL-1 signaling in astrocytes and microglia is dispensable for the pathogenesis of EAE. In contrast,
Il1r1 deletion specifically in BBB-ECs reduces EAE disease severity. Our study confirms that EC-IL-1 signaling plays a key role in mediating leukocyte migration across the BBB during neuroinflammation, which is a decisive factor for disease initiation [
16,
17,
40,
55]. Importantly, we found that IL-1 signaling at the onset of EAE suppresses the expression of HO-1 in BBB-ECs and that HO-1 overexpression specifically in BBB-ECs reduces EAE severity, similar to what we observed in BBB-EC-specific
Il1r1 knockout mice. The protective effect of HO-1 is at least partially mediated by downregulation of IL-1β-driven gene expression, as illustrated for the adhesion molecules
Icam1 and
Vcam1 as well as the atypical chemokine receptor
Ackr1, all of which are also downregulated in the absence of IL-1R1 in BBB-ECs during EAE. While reduced expression of Icam-1 and Vcam-1 partly accounts for the diminished leukocyte infiltration, downregulation of the atypical chemokine receptor 1 (Ackr1), also known as Duffy antigen receptor for chemokines (Darc), most likely contributes to EAE amelioration by preventing chemokine shuttling from the abluminal to the luminal surface of endothelial cells, thus counteracting the pro-migratory function of Darc during neutrophil infiltration [
27,
56].
Our study is the first to demonstrate a protective role for BBB-EC-derived HO-1 in the development of CNS autoimmunity. It was previously shown that global
Hmox1 deletion in mice results in enhanced EAE severity, which is associated with the inhibition of MHCII expression by APCs and accumulation of CD4
+ and CD8
+ T cells within the CNS [
18]. While the protective effect of HO-1 is driven at least partially through the end product of heme degradation, namely CO [
18], this does not rule out that other products that emerge in this pathway, such as biliverdin or bilirubin, act in a similar manner. Indeed, biliverdin exerts potent immunoregulatory effects in T
H cells [
84] while bilirubin, the product of biliverdin conversion by biliverdin reductase, reduces chronic and acute EAE disease in rats [
47].
The role of HO-1 as an anti-inflammatory and anti-oxidative molecule has been studied also in other disease models. HO-1 supports BBB integrity in the context of cerebral malaria, an often lethal neuroinflammatory syndrome that develops in response to
Plasmodium infection [
62] and when expressed in spinal cord ECs it protects the BBB against oxidative injury and limits the infiltration of leukocytes into the CNS [
43]. Other studies have demonstrated that HO-1 activation in cultured ECs inhibits the NF-κB-driven
Vcam1 transcription, which agrees with our in vivo and in vitro findings [
70,
72].
The mechanism by which EC-IL-1 signaling inhibits HO-1 expression and vice versa, how HO-1 inhibits IL-1 signaling in ECs is not completely understood. Several studies have shown that activation of the transcription factor Nrf2 correlates with the suppression of NF-κB signaling [
12,
34,
72]. Moreover, it was shown, that transcription driven by canonical NF-κB may regulate the Nrf2-mediated expression of the antioxidant response element (ARE) by different mechanisms, suggesting a crosstalk between the Nrf2/HO-1 and the IL-1 signaling pathways. The activity of Nrf2 is regulated by the repressor protein Keap1 [
32]. Under conditions of oxidative stress, Keap1 loses its ability to repress Nrf2 activity, which leads to Nrf2 nuclear accumulation, inducing the expression of stress-preventing genes such as
HMOX1 [
87]. Others have shown that the NF-κB p65/RelA subunit can physically associate with Keap1 to promote its translocation into the nucleus, resulting in the inactivation of Nrf2 [
87]. The most well-established interplay between Nrf2 and p65/RelA represents their competition for the binding site of the transcriptional coactivator CREB-binding protein (CBP), which both transcription factors require to propagate their signal [
44,
76,
82]. Moreover, the overexpression of p65/RelA can limit the availability of CBP for complex formation with Nrf2, thus favoring increased NF-κB-driven gene expression [
44]. Furthermore, it was shown that NF-κB can recruit the corepressor histone deacetylase 3 (HDAC3), causing local histone hypoacetylation of CBP or the Maf transcription factor protein (MafK), thereby potentially hampering Nrf2 signaling [
76]. Collectively, the interplay of HO-1 and the IL-1 signaling pathway can occur through a variety of molecular interactions. We suggest that both NF-κB and HO-1 expression are essential for the EC response to inflammation and that an imbalance between the two pathways is associated with increasing inflammation. Our study complements the existing data by identifying IL-1β as a key regulator by suppressing HO-1 expression in BBB-ECs at a certain threshold, which results in increased autoimmune inflammation. Moreover, we show that EC-HO-1 is highly important to control EAE pathogenesis, mediated by the reduced expression of IL-1β-driven genes.
Pharmacological induction of HO-1 was reported to improve a variety of inflammatory diseases [
3,
71], which suggests HO-1 as a potential target for therapeutic intervention also in MS. In this line, other studies suggested HO-1 to be beneficial for MS outcome, as this enzyme is increased within the CNS of MS patients [
69]. Other studies could show that HO-1 expression in PBMCs of MS patients is reduced during disease progression and that during exacerbation of the disease there was a significant downregulation of HO-1 [
24]. Furthermore, treatment with steroids increased the expression of HO-1, suggesting that HO-1 is protective against MS. The investigation of specific activators of the Nrf2/HO-1 pathway might be useful to improve anti-inflammatory mechanisms in MS. As a promising example, the Nrf2 activator dimethyl fumarate (DMF, Tecfidera
®) is currently used to treat the relapsing–remitting form of MS, with rapid and sustained clinical and neurological efficacy [
26,
33]. Importantly, DMF was shown to induce the downregulation of adhesion molecules, i.e. Vcam-1 and Icam-1 [
20,
21], which was correlated with Nrf2 activation [
26,
28,
67] and, therefore, supports the idea of a functional crosstalk between the NF-κB and Nrf2/HO-1 pathways. However, identification of a potent and specific Nrf2 activator or the targeting of its suppressors to treat MS and other inflammatory diseases requires further detailed investigations. We believe that a precise understanding of this crosstalk will help to target and manipulate the balance between Nrf2 and NF-κB signaling, which eventually represents a promising approach for improving MS therapies.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.