Background
In multiple sclerosis (MS), a multifocal inflammatory demyelinating disease of the central nervous system (CNS), the distribution of lesions can vary widely between patients, resulting in distinct clinical phenotypes. In some patients, lesion burden is dispersed fairly evenly across CNS compartments, while in others it is skewed either towards the spinal cord or supratentorial white matter [
1,
2]. The histopathology of MS lesions, including the presence of specific immune subsets and factors, is also diverse. Similarly, lesion distribution and composition is heterogeneous in experimental autoimmune encephalomyelitis (EAE), which is widely used as an animal model of MS.
Although little is known about the factors that determine the composition or location of lesions in MS, several studies have demonstrated a pivotal role of IFNγ in EAE. Adoptive transfer of Th1-polarized, myelin oligodendrocyte glycoprotein (MOG)-reactive T cells derived from C57BL/6 wildtype (WT) mice into syngeneic wildtype (WT) hosts results in a high incidence of “conventional” EAE (cEAE), which manifests as an ascending paralysis secondary to monocyte-rich inflammatory infiltration of the thoracolumbar spinal cord. Conversely, transfer of the same population of Th1 cells into IFNγ receptor knockout (IFNγRKO) hosts, or transfer of IFNγKO CD4
+ T cells into WT hosts, results in a high incidence of atypical EAE (aEAE), characterized by gait imbalance and brainstem or cerebellar inflammation [
3‐
5]. IFNγ induces expression of chemokines, such as CCL2 and CXCL10 [
6], and endothelial adhesion molecules, such as VCAM-1 and P-selectin [
7,
8], in the CNS. Collective induction of these molecules promotes neuroinflammation by facilitating the passage of T cells and monocytes into the CNS parenchymal white matter. In fact, cEAE is dependent on spinal cord expression of CCR2 and α4β1 integrin, the receptors for CCL2 and VCAM-1, respectively [
5,
9,
10]. In contrast, aEAE lesion development correlates with production of ELR
+ CXC chemokines in the brainstem and is neutrophil dependent [
5]. The destructive effects of neutrophil infiltration in CNS white matter have been demonstrated in several different animal models [
11,
12]. Hence, the relative importance of particular cytokine and chemokine pathways varies across different models of EAE, which can translate into differential responsiveness to individual disease-modifying therapies [
3‐
5,
13,
14].
For the future development of personalized approaches to MS therapy, it will be important to acquire a detailed understanding of the spectrum of molecular and cellular pathways that underlie similar, as well as divergent, clinical courses of autoimmune demyelinating disorders. In the current study, we investigate the cellular source and spatial localization of the ELR+ CXC chemokines, CXCL1 and CXCL2, in C57BL/6 IFNγRKO mice with aEAE, the factors that trigger CXCL2 production, and the mechanism by which IFNγ regulates neutrophil infiltration of the brainstem.
Methods
Mice
Eight- to 14-week-old CD45.1 congenic and WT C57BL/6 mice were obtained from Charles River/NCI Fredrick. IFNγRKO and Rosa
mT/mG mice were originally obtained from Jackson Laboratory and bred in the University of Michigan vivarium. LysM-Cre:IFNγRflox/flox mice were obtained from Athena Soulika, University of California-Davis [
15]. All mice were housed in microisolator cages under specific pathogen-free conditions. All animal protocols were approved by the University of Michigan Committee on Use and Care of Animals.
Antibodies and reagents
For flow cytometry, the following antibodies were obtained from eBioscience: PECy7-α-CD11b (M1/70), eFluor450-α-CD45 (30-F11), and PerCpCy5.5-α-Ly6C (HK1.4). Allophycocyanin cy7-α-Ly6G (IA8) was from BD Biosciences. For immunofluorescent histology, primary antibodies included rabbit α-glial fibrillary acidic protein (GFAP) (Gibco), rat α-mouse CD45 (IBL-5/15, Millipore), goat α-mouse CXCL2 (R&D Systems), goat α-mouse CXCL1 (R&D Systems), rat α-mouse Ly6G (IA8, BD Biosciences), and hamster α-mouse CD3ε (BD Biosciences). All secondary antibodies were from Life Technologies, including AlexaFluor594 donkey-α-goat IgG, AlexaFluor488 goat-α-rat IgG, goat α-hamster, and AlexaFluor647 goat-α-rabbit IgG. For in vitro cultures, G-CSF was obtained from Amgen; recombinant mouse (rm) IFNγ, rmIL-12, rmCXCL2, rmIL-1β, and rmCXCL1 were from R&D Systems.
Induction and scoring of EAE
Donor mice were immunized subcutaneously with 100 μg MOG
35–55 (MEVGWYRSP-FSRVVHLYRNGK, Biosynthesis) in CFA (Difco) across four sites over the flanks. Inguinal, axial, and brachial lymph nodes were harvested 14 days post-immunization, pooled, homogenized, and passed through a 70-μm strainer (BD Falcon). Cells were cultured with MOG
35–55 (50 μg/mL) in the presence of rmIL-12 (6 ng/mL) and rmIFNγ (2 ng/mL). At 96 h, CD4 T cells were isolated by column separation with CD4 (L3T4) magnetic microbeads, according to the manufacturer’s instructions (Miltenyi). CD4 T cells (> 80% pure) were transferred i.p. into naïve hosts (5 × 10
6 cells/ mouse). Adoptive transfer recipients were monitored on a daily basis by an examiner who was blinded to the experimental groups. Mice were scored for severity of conventional and atypical signs of EAE using established scales [
4,
5].
Construction of bone marrow chimeric mice
Femurs and tibiae of IFNγRKO or ROSAmT/mG mice were flushed with PBS using a 26-gauge needle to obtain donor bone marrow cells. Cells were ACK lysed and suspended in cold PBS for intravenous injection into CD45.1 congenic hosts, which had been subjected to 13 Gy of irradiation (orthovoltage X-ray source) split into two doses, 3 h apart. Hosts were given 2.5 × 106 bone marrow cells from each donor source (1:1 ratio) and allowed to reconstitute for at least 6 weeks before further use. Chimerism was verified by flow cytometric analysis of peripheral blood mononuclear cells.
Histochemical procedures
Mice were perfused with 1× PBS and 4% paraformaldehyde (PFA). CNS tissues were harvested, post-fixed in 4% PFA for 96 h, decalcified in 0.5 M EDTA for 96 h, and transferred into 30% sucrose for at least 48 h prior to embedding in OCT and storage at − 80 °C. Spinal cord and brain tissues were cryosectioned at 10 μm. Sections were incubated in 1× PBS in a humidified chamber, followed by blocking solution (1× PBS 7.4 pH, 10% Normal Donkey Serum, 0.5% Triton-X100) for 1 h. Primary antibodies were applied overnight at 4 °C. For secondary antibody staining, sections were incubated with AlexaFluor594 donkey-α-Goat IgG, rinsed in PBS, and then incubated with AlexaFluor488 goat-α-rat IgG and AlexaFluor647 goat-α-rabbit IgG followed by DAPI (100 ng/mL). Sections were washed and mounted on slides (Antifade Reagent, Southern Biotech). Confocal images were acquired using a Nikon A-1 confocal microscope (Nikon PlanApoVC × 20, × 40, or × 60/1.40 oil) with diode-based laser system and NIS Elements software.
For in situ hybridization, cryosections of CNS tissue were incubated with Proteinase K (10 μg/mL, Sigma) for 20 min followed by incubation with digoxigenin-labeled cRNA probes (DIG-11-UTP; Roche). Sense and anti-sense CXCL2 probes were transcribed from a 1.08-kb mouse CXCL2 cDNA, a gift of Dr. Luc Vallières, Laval University [
16]. Hybridization was performed at 55 °C in 50% formamide with a final concentration of 100–200 ng of DIG probe/mL [
17]. Following stringency washes, hybridization signal was identified with α-Digoxigenin-AP (Roche). Colorimetric images were converted to red as positive signal and black as negative, and overlaid onto DAPI stain of in situ sections. Images were taken with a Nikon Eclipse Ti-U with a Nikon D5-U2 camera using NIS Elements software. Appropriate processing including image overlays and black level and brightness adjustments were performed in Adobe Photoshop CC2014 and applied equally to all samples and controls.
CNS mononuclear cell isolation
CNS tissue was harvested and dissected into the spinal cord and brainstem. Each specimen was homogenized in 1 mL PBS containing a protease inhibitor cocktail (Roche) and centrifuged at 800×g for 10 min. Supernatants were stored at − 80 °C for subsequent chemokine analysis. Infiltrating cells were isolated over a 27% Percoll gradient. Cells were quantified with a Cellometer AutoT4 automated cell counter, excluding dead or dying cells with trypan blue (Nexcelom).
Flow cytometry
For surface staining, cells were resuspended in PBS + 2% fetal bovine serum (FBS) containing Fc Block (50 ng/mL) and Fixable Viability Dye efluor 506 (eBioscience) prior to incubation with fluorochrome-conjugated antibodies. For intracellular staining, cells were incubated with Brefeldin A (10 μg/mL) for 4 h in the presence or absence of stimulation conditions. Cells were labeled with fluorochrome-conjugated cell surface antibodies as described above, fixed in 4% PFA, permeabilized with 0.5% saponin, and incubated with fluorochrome-conjugated α-cytokine or chemokine antibodies. Stained cells were run on a FACS Canto II flow cytometer (v6.1.3, Becton Dickenson) or sorted on the FACS Aria II using FACS Diva software. Data was analyzed using FlowJo software (v10.0.7r2, Treestar).
In vitro monocyte and neutrophil assays
Eight- to 14-week-old mice were euthanized; the femur and tibia were flushed and passed through a cell strainer (70 μm) with repeated washes. For monocyte stimulation, whole bone marrow was plated in complete media (RPMI with 10% FBS, L-Glutamine (2 mM, Gibco), Pen/Strep (1:100, Gibco), sodium pyruvate (12.5 μM, Gibco), and 2-mercaptoethanol (55 μm, Gibco)) in the presence of Brefeldin A (10 μg/mL) with lipopolysaccharide (LPS) (1 μg/mL), with or without IFNγ (2 ng/mL). Four hours later, the cells were stained for intracellular CXCL2 and subjected to flow cytometric analysis. Neutrophils were purified from bone marrow cell suspensions using an α-Ly6G microbead kit (Miltenyi). Purified neutrophils were plated in complete media in the presence or absence of CXCL2 (20 ng/mL), CXCL1 (20 ng/mL), granulocyte colony-stimulating factor (G-CSF) (25 ng/mL), IFNγ (2 ng/mL), or IL-1β (10 ng/mL) and isolated following 1 h in culture to examine CXCL2 mRNA expression.
RNA isolation
Cells from FACS or in vitro cell culture were spun down and resuspended in 1 mL Trizol (Life Technologies). For RNA extraction, 200 μL of chloroform was added to samples and mixed prior to centrifugation at 18000×g. Chloroform layer was moved to a fresh tube with 500 μL cold isopropanol, mixed and incubated for 15 min prior to purifying the RNA out using the RNeasy MiniKit (Qiagen) with column DNase digestion per manufacturer’s instructions.
RT- and q-PCR
RT-PCR was performed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) per manufacturer’s instructions. For q-PCR, TaqMan Universal Master Mix and primer/probe sets for CXCL2, CXCR2, and GAPDH were purchased from Applied Biosystems and run on a MyIQ system using iQ5 software (BioRad) as described in manufacturer’s instructions.
Statistical analysis
Statistical analyses were performed using GraphPad Prism software. Unless otherwise stated, all graphs are expressed as means ± SEM. Comparisons of cell numbers, or of transcript levels in myeloid cells isolated from WT versus IFNγRKO hosts, were done using an unpaired Student’s t test. Comparisons of transcript levels in paired neutrophils or monocytes/macrophages isolated from mixed BM chimeras, or in myeloid cells cultured in vitro under different conditions, were done using a paired Student’s t test. A p value < 0.05 (*) was considered significant. p < 0.01 (**) and p < 0.001 (***).
Discussion
Based on the results of this study, we propose the following hypothetical model by which an IFNγ/CXCL2 axis regulates the regional localization of inflammatory infiltrates during EAE. CXCL1 is released by choroid plexus epithelium in the early stages of both cEAE and aEAE, and promotes neutrophil migration into the fourth ventricle and meningeal space. In IFNγRKO hosts, the neutrophils are then stimulated to migrate deeper into the brainstem parenchyma by a CXCL2 concentration gradient, initially generated by activated microglia, and subsequently amplified by neutrophils and monocytes as they accumulate in the brainstem white matter. Astrocyte-derived CXCL1 might also contribute to the early infiltration of the brainstem parenchyma by neutrophils. CXCL2 production by CNS myeloid cells escalates during the progression of aEAE, driven by an autocrine and/or paracrine positive feedback loop (as suggested by the data in Fig.
4). Conversely, in WT hosts, CXCR2 and CXCL2 expression in CNS myeloid cell populations is suppressed by IFNγ that is released by reactivated myelin-reactive T cells. Consequently, neutrophils do not penetrate deep into the brainstem white matter. The inhibition of CXCR2 and CXCL2 expression by CNS-infiltrating neutrophils in response to IFNγ signaling appears to be a cell- and microenvironment-specific phenomenon, since IFNγ has been reported to actually promote CXCR2 and CXCL2 production by non-hematopoietic cells in non-CNS tissues [
23‐
25]. Reminiscent of the results of the current study, infiltrating neutrophils were recently found to be the major source of CXCL2 in a mouse model of immune complex-mediated cutaneous inflammation (ICMCI) [
26]. Expression of CXCL2 by immune complex-activated neutrophils amplified their recruitment to the skin in an autocrine/paracrine manner. The ICMCI model also resembles our Th1-mediated EAE model in that tissue resident cells were the major source of CXCL1. In both animal models, differences in the cellular source and spatial distribution of CXCL1 and CXCL2 translate into distinct roles of those chemokines in the pathogenic process.
Although the model of Th1-mediated aEAE described here is IL-17 independent, that is not the case for an alternative aEAE model induced in C3Heb/Fej mice by the adoptive transfer of IL-23 polarized, MOG
97–114-reactive CD4
+ Th17 cells [
13,
19]. The latter model of aEAE, like ours, is neutrophil dependent and associated with CXCL2 induction in the brain [
19]. Astrocytes isolated from the brain and spinal cord of C3Heb/Fej mice at peak aEAE express higher levels of CXCL2 mRNA than endothelial cells, microglia, or CNS-infiltrating myeloid cells [
19]. In contrast, infiltrating myeloid cells and microglia are a major source of CXCL2 in the brainstem of C57BL/6 IFNγRKO recipients following the transfer of encephalitogenic Th1 cells (Figs.
1 and
2). CXCL2 itself, in combination with IL-1β, is a likely candidate for the inducer CXCL2 in our model (Fig.
4). We did not detect CXCL2 in GFAP
+ astrocytes in brainstem sections of symptomatic C57BL/6 IFNγRKO hosts. However, GFAP staining was relatively sparse in our brainstem sections. In addition, astrocytes were rare among the CD45
− cells that we isolated from brainstem preparations for flow cytometric analysis (data not shown). Therefore, we cannot rule out the possibility that, in our aEAE model, astrocytes and/or other non-hematopoeitic CNS resident cells contribute to brainstem CXCL2 expression in conjunction with myeloid cells. The contention that IFNγ suppresses aEAE via effects on multiple cellular targets in parallel is supported by the increased susceptibility of both IFNγRKO ➔WT and WT➔IFNγRKO bone marrow chimeric mice to aEAE [
15].
The intracellular pathways that regulate neutrophil CXCL2 expression during aEAE remain to be elucidated. The fact that IFNγ inhibits CXCL2-induced, but not G-CSF-induced, upregulation of CXCL2 in cultured neutrophils (Fig.
4) suggests the co-existence of multiple, parallel pathways. LysM-Cre:SOCS3
fl/fl conditional knockout mice develop an atypical form of EAE that is ameliorated by neutrophil depletion or CXCR2 blockade [
27]. Neutrophils from LysM-Cre:SOCS3
fl/fl mice exhibit enhanced and prolonged activation of STAT3 in response to stimuli such as G-CSF. We are currently investigating the role of a SOCS3/STAT3 immunoregulatory circuit in our model.
Conclusions
The current study adds to a growing body of literature demonstrating diversity in the dysregulated immune pathways that drive, and modulate, CNS autoimmune disease. Clinical variants of human autoimmune demyelinating disease display different patterns of lesion distribution. For example, the spinal cord is disproportionately targeted in an opticospinal form of MS that is prevalent in Asia [
28,
29]. In contrast, lesion burden is skewed towards the cerebrum in most individuals in the Western hemisphere with relapsing-remitting MS [
30]. Differences in underlying immunopathological mechanisms and lesion distribution between MS patients could correlate with differences in environmental triggers, and ultimately translate into differences in therapeutic responsiveness to individual disease-modifying drugs [
31‐
33]. Collectively, the current study, together with previously published studies in aEAE and cEAE, suggests that actionable therapeutic targets may differ among MS patients with distinct types of immune dysregulation and clinical phenotypes. Conversely, in some cases the role of a particular mediator and/or leukocyte might be universal across individuals within the same clinical subset (as with the regulatory role of IFNγ in brainstem inflammation). These observations illustrate the complexity of the relationship between immunopathological pathways and their clinical manifestations. They underscore the importance of discovering biomarkers that correlate with underlying immune mechanisms, in addition to clinical phenomenology, to help guide the management of inflammatory demyelinating disease in a customized manner.
Acknowledgements
Claire Johnson provided technical assistance. Flow sorting experiments were performed in the Flow Cytometry Research Core at the University of Michigan, Biomedical Research Core Facilities (BRCF). Confocal microscopy was performed in the Microscopy and Image-analysis Laboratory (MIL) at the University of Michigan, Biomedical Research Core Facilities (BRCF). The MIL is a multi-user imaging facility supported by NIH-NCI, O’Brien Renal Center, UM Medical School, Endowment for the Basic Sciences (EBS), and the University of Michigan.