|Home | About | Journals | Submit | Contact Us | Français|
Multiple Sclerosis (MS) is an inflammatory demyelinating disease of the central nervous system (CNS) and remyelination in MS ultimately fails. Although strategies to promote myelin repair are eagerly sought, mechanisms underlying remyelination in vivo have been elusive. CXCR2 is expressed on neutrophils and oligodendrocyte lineage cells in the central nervous system (CNS). CXCR2 positive neutrophils facilitate inflammatory demyelination in demyelination models such as experimental autoimmune encephalomyelitis (EAE) and cuprizone intoxication. Systemic injection of a small molecule CXCR2 antagonist at the onset of EAE decreased demyelinated lesions. These results left the cellular target of the CXCR2 antagonist uncertain, and did not clarify whether CXCR2 blockade prevented demyelination or promoted remyelination. Here, we show that the actions of CXCR2 on non-hematopoietic cells unexpectedly delay myelin repair. Bone marrow chimeric mice (Cxcr2+/−→Cxcr2−/− and Cxcr2+/−→Cxcr2+/+) were subjected to two distinct models of myelin injury. In all cases, myelin repair was more efficient in Cxcr2+/−→Cxcr2−/− animals. Oligodendrocyte progenitor cells (OPCs) in demyelinated lesions of Cxcr2+/−→Cxcr2−/− mice proliferated earlier and more vigorously than in tissues from Cxcr2+/−→Cxcr2+/+ animals. In vitro demyelinated CNS slice cultures also showed better myelin repair when CXCR2 was blocked with neutralizing antibodies, or was genetically deleted. Our results suggest that CXCR2 inactivation permits optimal spatiotemporal positioning of OPCs in demyelinating lesions to receive local proliferative and differentiating signals. Given that CXCR2 exerts dual functions which promote demyelination and decrease remyelination by actions towards hematopoietic cells and non-hematopoietic cells respectively, our findings identify CXCR2 as a promising drug target for clinical demyelinating disorders.
Multiple Sclerosis (MS) is an inflammatory demyelinating disease of the central nervous system (CNS) and remyelination in MS ultimately fails. In the CNS, oligodendrocyte progenitor cells (OPCs) carry out a complex, precisely timed program of migration, proliferation and differentiation, followed by programmed cell death or myelination. Understanding remyelination is crucial for devising effective methods to prevent or reduce its failure in clinical demyelinating disorders.
Chemokines act through G-protein coupled receptors to regulate cell movement, activation, proliferation and differentiation. Receptors for CXC chemokines frequently act on both circulating leukocytes and parenchymal cells in solid organs, and orchestrate complex tissue modelling during development, angiogenesis and neoplasia. CXCR2 is an ELR CXC chemokine receptor, also known as IL-8RB. It is expressed both on inflammatory myeloid cells such as neutrophils and on OPCs in the CNS. Peripheral expression of CXCR2 is important for recruitment of myeloid cells to sites of inflammation (Cacalano et al., 1994), and is implicated in cutaneous wound repair by promoting neutrophil recruitment, keratinocyte proliferation and angiogenesis (Devalaraja et al., 2000). Recently, our lab (Liu et al., 2010) as well as others (Carlson et al., 2008) identified the nonreduntant role of peripheral expression of CXCR2 on neutrophils in the promotion of demyelination in the CNS in two animal models of demyelination, and expression of CXCR2 on OPCs was not involved in the demyelinating process (Liu et al., 2010).
CXCR2 expressed on OPCs in the developing rodent spinal cord governs both migratory arrest, and (in the presence of PDGF) proliferative responses during development (Tsai et al., 2002). In postmortem tissues from MS cases, the CXCR2 ligand CXCL1 was detected in close proximity to oligodendrocytes on reactive astrocytes (Omari et al., 2005). Our lab found that CXCL1 and CXCL2 were upregulated on astrocytes near inflammatory foci in CNS tissues of mice with experimental autoimmune encephalomyelitis (EAE) (Glabinski et al., 1997). In a recent study, systemic treatment with a small molecular CXCR2 antagonist enhanced remyelination in EAE (Kerstetter et al., 2009). Further, localized inhibition of CXCR2 with this antagonist enhanced remyelination in lysophosphatidylcholine (LPC) induced demyelinating lesions. It remains uncertain which cells expressing CXCR2 contribute to remyelination after systemic or local inhibition of CXCR2. To distinguish whether periperal or CNS CXCR2 positive cells impair remyelination, we evaluated CXCR2-deficient mice and CXCR2 bone marrow chimeric mice in two different demyelination/remyelination animal models. Inactivation of CXCR2 on non-hematopoietic cells accelerated myelin repair in both models. OPCs in the demyelinated lesions of CXCR2-deficient animals proliferated earlier and more vigorously than in tissues from exposed animals. Demyelinated CNS slice cultures also showed better myelin repair when CXCR2 was blocked or genetically deleted. Our data suggest that CXCR2 exerts functions on non-hematopoitic cells of the CNS which impair remyelination.
The generation of Cxcr2-deficient mice has been describedpreviously (Cacalano et al., 1994). The animals used in these experiments had been backcrossed to SWR for at least 9 generations. Then SWR background Cxcr2 heterozyogous mice were crossed with SJL mice, producing F1 SWXJ heterozygous mice which were then intercrossed with F1 Cxcr2 heterozyogous mice, resulting in Cxcr2 knockout, heterozyogous, and wild type mice. Alternatively, the animals had been backcrossed to C57BL/6J (B6) mice for 11 generations. All comparisons in the current studies were made between littermate mice, with cohorts of mice being matched for both gender and age. Cxcr2 genotype was established using PCR based genomic DNA analyses as described previously (Tsai et al., 2002). SWR×SJL F1 (SWXJ (H-2qs)) mice halplotype was identified by flow cytometry (detail below, data not shown). All experimental mice were at the age of 8–10 wks, and were housed under pathogen-free conditions in the animal facility at the Cleveland Clinic. All protocols for animal research met the requirements of the Animal Research Committee of the Cleveland Clinic in compliance with the Public Health Service policy on humane care and use of laboratory animals.
For Cxcr2 bone marrow chimeric mice, 4–5 wks old female or male Cxcr2+/+, −/ − mice were subjected to 900 rads (using a 137Cs source from a Shepherd irradiator model 81-14R; JL Shepherd and Associates ) of total body irradiation to eliminate most of the endogenous bone marrow derived cells and stem cells. All irradiated mice were injected intravenously through the tail vein or with a retroorbital venous sinus injection with 1×107 bone marrow cells flushed and isolated from the femurs and tibias of donor heterozygous mice. Mice were maintained in gentamicin (10μg/ml) (Invitrogen) via the drinking water for the first 4 wks post-transfer. Animals were allowed to recover for 6 wks before further experiments such as EAE induction and cuprizone feeding. After 6 wks post-transfer, all mice were bled, and the extent of engraftment was determined by staining of CXCR2 expression on peripheral leukocytes, particularly granulocytes by using CXCR antibody and Gr1 antibody to define the CXCR2+ granulocyte population by using flow cytometric analysis (Liu et al., 2010). Irradiation did not affect the integrity of the blood-brain barrier, confirmed by staining brain tissue sections for murine IgG to indicate serum protein leakage into the CNS 6 wks after bone marrow transfer (data not shown), which is a sensitive indicator of BBB integrity (Pedchenko and Le Vine, 1999).
Rat myelin oligodendroglial glycoprotein (MOG35–55) and mouse proteolipid protein (PLP139–151) peptides were obtained from BIO·SYNTHESIS, and purified by HPLC. The purity of the peptide was >95%. The sequence of MOG35–55 and PLP139–151 were MEVGWYRSPFSRVVHLYRNGK and HSLGKWLGHPDKF, respectively.
Induction of EAE was performed as previously described (Liu et al., 2006) with modifications. Cxcr2+/+ and Cxcr2−/− mice or Cxcr2+/+ and Cxcr2−/− bone marrow radiation chimeric mice of 8–10 wks old were subcutaneously injected with different doses of MOG35–55 emulsified in CFA (Difco Laboratories) containing 400μg Mycobacterium tuberculosis with two injections of 200ng of pertussis toxin (PTX) (Sigma-Aldrich) on days 0 and 2 post-immunization (p.i.). Chronic EAE in SWXJ mice was induced with PLP135–151, as described previously(Liu et al., 2006). All mice were weighed, examined, and graded daily for neurological signs in a blinded manner as follows: 0, no disease; 1, decreased tail tone or slightly clumsy gait; 2, tail atony and moderately clumsy gait and/or poor righting ability; 3, limb weakness; 4, limb paralysis; and 5, moribund state. Disease relapse was determined when an increase of one EAE score unit was observed. Signs of neurological impairment were typically accompanied by an abrupt, substantial weight loss (>7%). The average day of EAE onset was calculated by adding the first day of clinical signs for individual mice and dividing by the number of mice in the group. The EAE index was calculated by adding all of the daily EAE scores to obtain a cumulative score, and dividing by day of EAE onset (Liu et al., 2006). Active immunization with MOG35–55 induced monophasic EAE in B6 mice, and was followed for 30 days. Chronic relapsing EAE induced by PLP139–151 was also monitored for 30 days. Animals were euthanized if found to be worse than grade 4. We took a conservative approach and eliminated animals that scored five on the day of sacrifice. Water-soaked food was provided on the cage floor when animals reached grade 3 or worse. Deaths before day 7 p.i. were attributed to reaction to immunization or injection of PTX, not EAE.
Mice were transcardially perfused with ice-cold 0.1M PBS, followed by 4% paraformaldehyde solution (PFA) under anesthesia. Spinal cords were rapidly dissected and cut into two pieces at three comparable levels (Cervical, thoracic and lumbar). One of them was post-fixed in PBS containing 4% PFA overnight at 4°C for immunohistochemical staining, and for the other one-millimeter-thick sections were immediately fixed in PBS containing 4% PFA and 2.5% glutaraldehyde for 1 wk at 4°C for ultramicroscopy. These sections were then post-fixed in 1% osmium tetroxide for 2 hrs at room temperature and then dehydrated in a graded series of methanol baths and embedded in araldite resin according to the previously described protocol (Liu et al., 2010). One-μm-thick sections were cut from the araldite-embedded material, stained with 1% toluidine blue for light microscopy, and delineated for ultrathin sections. The strategy for analyzing tissue damage in matched semi-thin sections of lumbar spinal cord is shown schematically in figure 3B. The total white matter area and the demyelinated area in the lumbar spinal cord were measured by ImageJ software (http://rsb.info.nih.gov/ij/). The percentage of demyelinated area was calculated by normalizing the demyelinated area against the total white matter area. The histological analysis was performed by a researcher blinded to the experimental group. For cuprizone treated mice, midsagittal 1-mm thick sections through the anterior corpus callosum were embedded in araldite for electron microscopy. For electron microscopy, 80nm sections were cut, mounted on copper grids, stained with uranyl acetate and lead citrate, and examined with a Zeiss 10C electron microscope. Representative areas were photographed (Liu et al., 2010).
Peripheral blood cells were lysed in red blood cell lysis buffer ( NH4Cl 0.15M, KHCO3 1.0M, EDTA.2Na 0.1 mM) for 5 min and were washed and resuspended in FACS buffer (1% FCS and 0.1% sodium azide in PBS). After blocking with CD16/CD32Abs at 4°C for 30 min, cells were stained for surface markers with directly conjugated Abs in FACS buffer at 4°C for 30 min. Cells were washed twice and resuspended in the 200μl of PBS for flow cytometry analysis as described previously (Liu et al., 2006). Antibodies used in our experiments were PE-I-A/I-E mAb (M5/114.15.2, B.D) and FITC I-Ak mAb (10-3.6, B.D). For peripheral blood myeloid cell expression of CXCR2, FITC-Gr1Ab (BD, Pharmingen) PE-CXCR2 Ab (R&D) were used (Liu et al., 2010). Analysis was performed with a FACSCalibur (BD Biosciences)equipped with CellQuest software (BD Biosciences), and 50,000 events were acquired. Data were analyzed with FlowJo software (Tree Star). The percentages of myeloid cells positively stained for CXCR2 were determined.
For histological and immunohistochemical analysis of CNS tissues at different stages of EAE and different time points of cuprizone treatment and lysolecithin injection, mice were transcadially perfused with ice-cold PBS, followed by 4% PFA solution, under anesthesia. Brains and spinal cords were rapidly dissected and post fixed in 4% PFA, then washed with PBS and put in cryoprotection buffer at least overnight at 4°C until the tissues sank, then 30-μm-thick coronal sections of the brain and transverse sections of spinal cord were cut on a sliding microtome (Leica Microsystems, Wetzlar, Germany) and kept in cryostorage buffer at −20°C for further experiments. For immunostaining, sections were pretreated as necessary in a water bath (90°C) for 20 min in sodium citrate antigen retrieval buffer (DAKO, pH 6.0), then cooled for 30 min. For the staining process, prepared tissues were incubated with 0.3% hydrogen peroxide in 3% Triton-100, blocked by incubation with 5% normal serum at room temperature for 1 hr, then incubated overnight at 4°C with primary antibodies at the dilution indicated: rat anti-mouse CD45 monoclonal antibodies, 1:2000 (clone MCA 1388; Serotec); rabbit anti-rat/mouse PDGFRα polyclonal antibodies (to identify OPCs) (1:2000, gift from Dr. William Stallcup, La Jolla, CA); PCNA antibody (PC10, cell proliferation marker, Sigma); PSA-NCAM antibody (1:1000, Chemicon); Olig2 antibody (1:1000, Chemicon); GST-π antibody (1:20000, Assay designs), GFAP antibody (1:8000, Sigma). On day 2, tissues were incubated with appropriate biotinylated secondary antibodies[(goat anti-rat, goat anti-mouse or goat anti-rabbit (DAKO, 1:1000), then with ABC (DAKO, 1:1000)].
Sections were washed 3 times with PBS after each incubation step (except for goat serum). All antibodies, as well as ABC, were diluted in 1% BSA in PBS. Sections were developed with DAB with hydrogen peroxide for 5 min at room temperature. Following development with DAB, tissues were rinsed in ddH2O, dehydrated and mounted. Negative controls were incubated with pre-immune IgG (Sigma). For immunofluorenscence staining, Alex 594-anti-rabbit antibody, Alex 488-anti mouse antibody, or 647-streptavidin (Invitrogen) were used instead of ABC development. The quantification of percent area occupied by immunoreactivity, the number of cells in lesions (number per unit area), or percentage of immunoreactive cells among total cells were perfomed as previously described (Liu et al, 2006, Liu et al, 2010).
Cxcr2+/+ or Cxcr2−/− bone marrow radiation chimeric mice in C57BL/6 background mice of 8–10 wks old were fed a diet of pellet mouse chow (diet TD.06172; Harlan Teklad, Madison, WI) containing 0.2% (w/w) cuprizone (Biscyclohexane oxaldihydrazone, Sigma-Aldrich, St. Louis, MO ) ad libitum for a period of 4, 5 or 6 wks to induce CNS demyelination. Mice were maintained on the cuprizone diet until perfused for analysis or returned to normal chow for 3 days, 7 days or 10 days for remyelination. Cxcr2+/+ or Cxcr2+/− mice and Cxcr2+/+ or Cxcr2−/− bone marrow radiation chimeric mice without cuprizone feeding were used as controls (Liu et al., 2010). Mice were weighed twice a week, and we did not find a continous weight loss after one week. Control mice were maintained on a normal diet for the duration of the experiment.
Cross sections of spinal cords from EAE mice, and coronal sections of brain from cuprizone treated mice were stained for myelin by using Black-Gold solution (Chemicon, Jefferson, AR). The Black-Gold histochemical staining was performed according to the manufacturer’s protocol with modifications. Briefly, free floating sections were washed with 0.9% NaCl, then incubated in 0.2% Black-Gold solution at 65°C in a water bath for 10 min, rinsed for 2 min in 0.9% NaCl, fixed for 3 min in a 2% sodium thiosulfate solution, and rinsed in 0.9% NaCl 3 times (Liu et al., 2010). Sections were mounted on slides, and dehydrated at 60°C for 30 min, cleared in xylenes, and coverslipped with Permount media (Sigma). Corpus callosum area was digitized under a 5x objective, using a 3-CCD color video camera interfaced with an Image-Pro Plus Analysis System (22.214.171.124Version, MediaCybernetics). For comparison of results obtained in CXCR2+/+ and CXCR2 −/ − bone marrow chimeric mice, levels of brain sections were carefully matched. Digitized images were analyzed with National Institutes of Health ImageJ1.34s software. A threshold procedure was established to determine the proportion of Black-Gold staining area within each gated corpus callosum area. These parameters were then held constant for each set of images obtained at equal objectives and light intensities, on slides that were processed in one session. The data represent the mean area occupied by demyelinated tissue (void of Black-Gold staining), expressed as a percentage of total corpus callosum or dorsal column area.
P10 mice pups were sacrificed and the cerebellums were removed and placed in ice-cold Hank’s buffer (Invitrogen). 400 μm thick sagittal sections were cut by microtome (Leica VT 1000S, Leica). Two to three slices were plated on BD Falcon™ 6 well insert plates (Fisher) previously coated with 1 mg/ml collagen (Sigma) with a drop of complete media (50% DMEM/F12, 25% hanks buffer, 25% horse serum and 5 mg/ml glucose) on top of it for 4–6 hrs, then 500μl of the same media was added on top of the insert chamber and 1.5 ml media in the outside insert chamber. Demyelination was performed with 0.25 mg/ml lysolecithin (Sigma) in complete media with the same concentration of control rabbit antibody, CXCR2 antibody or media alone for 17 hrs, then fresh media with or without control rabbit antibody or CXCR2 antibody was administered for 2 days to allow remyelination. For evaluation of demyelination and remyelination, the slices were immunostained for myelin (mouse anti MBP antibody) and neurofilament (rabbit anti neurofilament antibody). The slices were washed with 1x PBS and fixed in 4% PFA at room temperature for 20 min, then washed 3 times with PBST (0.05% Triton x-100 in 1 xPBS), permeabilized in 10% Triton-x-100 at room temperature for 20 min, and blocked with blocking buffer (3% H2O2, 1% BSA, and 10% goat serum) at room temperature for 2 hrs. After that, the slices were incubated with MBP antibody and neurofilament antibody at 4°C overnight, and incubated with Alex 488-goat anti-mouse and Alex 594-goat anti-rabbit antibody (1:1000, Invitrogen) at room temperature for 1 hr the following day. After three washes, slices were mounted on the slides to obtain pictures. Double blind scoring of demyelination based on the extensive of myelination. 0: no myelination; 1: half myelination; 2: complete myelination.
Data are expressed as mean ± SD. Multiple comparisons were statistically evaluated by 1-way ANOVA using Prism 4 (GraphPad Software). The Students-t-test was used for the comparisons of disease severity, severity of pathological changes, and percentage area of immunoreactivity in comparisons between Cxcr2+/+ and Cxcr2−/− mice. A x2 test was used for the comparisons of disease incidence or mortality between Cxcr2+/+ and Cxcr2−/− mice. A p value <0.05 was considered as significant. *P<0.05, **P<0.01.
Although BALB/c mice lacking CXCR2 were resistant to induction of EAE (Carlson et al., 2008), the EAE score and severity were relatively mild in Balb/c background Cxcr2+/+ control mice. To extend understanding of the role of CXCR2 in EAE, we developed additional genetic models. Cxcr2−/− BALB/c mice were back-crossed to SWR and SJL/J background for 9 generations; SWR Cxcr2+/− mice were crossed with SJL/J Cxcr2+/− mice to obtain littermate Cxcr2+/+, Cxcr2+/− and Cxcr2−/− SWXJ mice for experiments (Yu et al., 1996). Mice used for EAE studies were genotyped by analysis of tail DNA and additionally analyzed by flow cytometry to ensure the H2q/s haplotype in SWXJ mice. The Cxcr2−/− allele was also back-crossed to the C57BL/6 (B6) background for 11 generations. In the susceptible SWXJ and B6 backgrounds, Cxcr2−/− mice were relatively resistant to developing EAE, with a particularly low incidence of disease (Fig. 1, Table 1). In three separate experiments, we immunized a total of 24 Cxcr2−/− mice in SWXJ background with PLP139–151 peptides, of which 9 died shortly after immunization and were removed from analysis. Only 5/15 (33%) surviving Cxcr2−/− mice developed EAE, as compared with 20/28 (71%) Cxcr2+/+ littermates (P<0.01). EAE day of onset, kinetics, mortality, and peak severity in Cxcr2−/− mice which became ill were all equivalent to that seen in Cxcr2+/+ littermates (Fig. 1, Table 1) (P>0.05). Interestingly, recovery from the initial attack of EAE was faster and more complete in Cxcr2−/− mice and severity scores in the resolution phase were significantly lower than in Cxcr2+/+ mice (P<0.05), which caused a decreased cumulative EAE score in Cxcr2−/− compared to Cxcr2+/+ littermates (Fig. 1, Table 1). Cxcr2−/− mice on B6 background were also relatively resistant to MOG-induced EAE: 87% incidence in Cxcr2+/+ mice, while 0/5 (0%) of Cxcr2−/− mice which survived beyond day 7 after immunization developed EAE (Table 1) (P<0.01). Our findings extend results from a recent report (Carlson et al., 2008) to EAE susceptible mice and further confirm that CXCR2-deficient mice are relatively resistant to induction of EAE. Moreover, we observed better recovery in Cxcr2−/− mice after the initial EAE attack.
Cxcr2−/− mice are relatively resistant to EAE in two different strains. These results left uncertain whether resistance to EAE in Cxcr2−/− mice was mediated solely by hematopoietic cells or whether altered CXCR2 expression on resident neuroepithelial cells also played a role. Addressing these questions was challenging because the expression of CXCR2 on neutrophils as well as OPCs could mediate diverse effects on lesion associated tissue injury. To address this issue, we generated Cxcr2+/−→Cxcr2−/− and Cxcr2+/−→Cxcr2+/+ bone marrow chimeric mice for EAE experiments. Cxcr2+/+→Cxcr2+/+ mice served as chimerism controls. We lethally irradiated both Cxcr2+/+ and Cxcr2−/− mice and transferred Cxcr2+/− bone marrow. Cxcr2+/− → Cxcr2−/− mice lacked CXCR2 on OPCs (and other non-radioresistant cells). Both chimerae expressed CXCR2 on hematopoietic cells (Liu et al., 2010). Engraftment of Cxcr2−/− mice was verified by demonstrating normalization of circulating myeloid cells by flow cytometry (Liu et al., 2010).
To investigate the role of CXCR2-expressing cells in the CNS, we immunized bone marrow chimeric mice that lacked CNS CXCR2 (Cxcr2+/−→ Cxcr2−/−) and controls (Cxcr2+/−→ Cxcr2+/+) using PLP139–151. In three separate experiments, we immunized a total of 22 Cxcr2+/− → Cxcr2−/− or Cxcr2+/− → Cxcr2+/+ mice, of which 5 of the 22 Cxcr2+/− → Cxcr2−/− died shortly after immunization and were removed from analysis. We found Cxcr2+/− → Cxcr2−/− and Cxcr2+/− → Cxcr2+/+ mice had an equivalent incidence of EAE (88% and 86% respectively) (Fig. 2, Table 2), indicating that the lack of expression of CXCR2 on neutrophils causes reduced incidence of EAE in Cxcr2−/− mice. The day of onset, day of peak disease (reflecting kinetics of neurological impairment), and severity of EAE at the peak were all equivalent in both groups of mice (Fig. 2, Table 2). However, as seen previously in Cxcr2−/− mice, the pace and extent of recovery were accelerated in mice lacking CXCR2 in the CNS as seen in the cumulative EAE score, which was 27±13.9 and 17±8.7 in Cxcr2+/− → Cxcr2+/+ and Cxcr2+/− → Cxcr2−/−, respectively (P<0.05) (Table 2). Neither bone marrow chimerism nor Cxcr2 haploinsufficiency in hematopoietic cells significantly affected the course of EAE since mean EAE severity scores and kinetics in Cxcr2+/−→Cxcr2+/+ and Cxcr2+/+→Cxcr2+/+ mice were not different (Supplementary table 1, data not shown). Previous published results showed that haploinsufficiency at CXCR2 produced significantly milder EAE in Cxcr2+/− mice (Carlson et al., 2008). The differences between our results and those previously reported may be due to the use of a different EAE model, as BALB/c mice (used in the previous study (Carlson et al., 2008)) represent a relatively EAE-resistant strain and develop a mild disease.
Next, we tested the hypothesis that milder EAE severity scores reflected improved tissue repair. Toluidine blue-stained semi-thin spinal cord sections were prepared to address this possibility by identifying remyelinated segments with thin myelin sheaths. We found that altered developmental myelin thickness in Cxcr2−/− mice (Padovani-Claudio et al., 2006) precluded this approach. Therefore, we quantified lesion area by prospective morphometric analysis both at the peak of disease and at the recovery phase by toluidine blue staining on semi-thin sections to reveal myelin. Cxcr2+/− → Cxcr2+/+ (Fig. 3A; upper panel) and Cxcr2+/− → Cxcr2−/− (Fig. 3A; lower panel) mice at the peak of EAE showed similar lesion areas and pathology. There was extensive myelin debris (arrowhead), activated macrophages (arrow) and degenerating axon cylinders (asterisk) in the lesion areas of both groups (Fig. 3A, middle panel), as predicted by the equivalent disease scores (Fig. 2). However, at the recovery phase (normalized by assaying at day 10 after peak EAE score), Cxcr2+/− → Cxcr2−/− mice had significantly smaller lesion areas than Cxcr2+/− → Cxcr2+/+ mice (Fig. 3A; 3C), consistent with the hypothesis that accelerated repair underlies improved neurological recovery (Fig. 2). Importantly, Cxcr2+/−→Cxcr2−/− and Cxcr2+/−→Cxcr2+/+ mice with EAE showed equivalent degrees of inflammation at the peak and recovery phase as monitored by CD45-staining and percent area occupied by immunoreactivity (Fig. 4). We conclude that myelin repair was unexpectedly enhanced in EAE tissues of mice that lacked CXCR2 on non-hematopoietic cells, despite equivalent degrees of initial inflammation and tissue injury. These results also support the hypothesis that CXCR2 expressed in myeloid cells is critical to the induction of EAE, but not to its repair.
Demyelination and remyelination in EAE occur stochastically throughout the CNS, mediated by complex, incompletely-characterized mechanisms. For these reasons, kinetic analysis of myelin repair in EAE lesions is impractical. To address the time course of myelin repair in timed and localized lesions, we used cuprizone feeding for 6 wks (Matsushima and Morell, 2001; Murtie et al., 2005), which mediates severe demyelination of the corpus callosum, followed by myelin repair within about 2 wks following cuprizone cessation. This model also offered the advantage of an extended time course in which to analyze cellular events associated with demyelination and remyelination. We performed analysis of demyelination in Cxcr2+/−→ Cxcr2−/− and Cxcr2+/−→ Cxcr2+/+ mice and found that Cxcr2+/−→ Cxcr2−/− and Cxcr2+/−→ Cxcr2+/+ mice showed equivalent severe demyelination of the corpus callosum after 4, 5 (data not shown) or 6 wks of cuprizone feeding (Liu et al., 2010) as monitored by quantitative histochemical Black-Gold staining (Fig. 5A–C). After removing the cuprizone food following 6 wks of cuprizone feeding, Cxcr2+/−→Cxcr2−/− and Cxcr2+/−→Cxcr2+/+ mice showed equivalent severe demyelination of the corpus callosum at day 3 after removing cuprizone food (Fig. 5A, 5C) (P>0.05). However, Cxcr2+/−→Cxcr2−/− mice showed dramatically enhanced myelin staining as compared with Cxcr2+/−→Cxcr2+/+ mice at day 7 and 10 of recovery (Fig. 5A, 5C) (P<0.05). Electron microscopy also showed significantly increased numbers of myelinated axons in Cxcr2+/−→Cxcr2−/− mice, compared to Cxcr2+/−→Cxcr2+/+ mice at day 10 of recovery (Fig. 5D–E) (P<0.05), confirming the histochemical staining results in these mice. There was no difference in the corpus callosum area between Cxcr2+/−→Cxcr2+/+ and Cxcr2+/−→Cxcr2−/− mice at the times of sacrifice (Fig. 5B) (P>0.05). Collectively, these data support the hypothesis that CXCR2 expression on CNS cells impaired myelin repair.
How does CXCR2 signaling to CNS cells inhibit remyelination? There are multiple potential cellular targets (Supplementary Fig. 1). For example, GFAP-positive neural stem cells (NSC) in the adult subventricular zone (SVZ) can give rise to oligodendrocytes which participate in the repair of the demyelinated corpus callosum (Nait-Oumesmar et al., 1999; Menn et al., 2006). The SVZ neurogenic niche is comprised of slowly-dividing GFAP+ NSCs, rapidly-dividing transit-amplifying cells, and migrating cells including neuroblasts. The migrating cells are labeled by antibodies to polysialylated neural cell adhesion molecule (PSA-NCAM). A small proportion of migrating cells express the oligodendroglial transcription factor Olig2 and can give rise to glutathione S-transferase (GST)-π+ oligodendrocytes. The yield of oligodendrocytes from the SVZ is increased four-fold after lysolecithin-mediated demyelination of the nearby rostral corpus callosum and in vivo retroviral labeling showed that SVZ cells contribute to remyelinating oligodendrocytes (Menn et al., 2006). The SVZ exhibits altered cytoarchitecture consistent with activation of the NSCs in both MS and EAE (Nait-Oumesmar et al., 1999).
To address the role of CNS CXCR2+ cells in remyelination, we performed kinetic analysis of NSCs and their progeny in the SVZ and the corpus callosum across the time course of demyelination and remyelination after cuprizone feeding of radiation bone marrow chimeric mice (Fig. 6 and Supplementary Fig. 1). Cxcr2+/−→ Cxcr2+/+ and Cxcr2+/−→ Cxcr2−/− mice were fed cuprizone for 4, 5 or 6 wks before sacrifice. Additional mice were fed cuprizone for 6 wks, and recovered without cuprizone feeding for 3 days or 7 days before sacrifice. At least three Cxcr2+/−→ Cxcr2+/+ and Cxcr2+/−→ Cxcr2−/− mice were analyzed at each time point. Analyses were performed in a blinded fashion and included GFAP+ NSCs in the SVZ and PSA-NCAM+ migrating cells in the lateral corpus callosum. In the corpus callosum, we quantified PDGFRα+/PCNA+ proliferating oligodendrocyte progenitor cells, GST-π+ mature oligodendrocytes, and Olig2+ cells representing both oligodendrocyte progenitors and mature oligodendrocytes.
Results are shown schematically in supplementary Fig. 1 and Fig. 6: 1. GFAP+ cells in the SVZ were unchanged throughout the experiment, consistent with their status as slowly dividing NSCs; 2. PSA-NCAM+ cells in the lateral corpus callosum showed a dramatic biphasic increase, at the same rate and to the same extent, in mice from both groups (Supplementary Fig. 1); 3. After 4 wks of cuprizone feeding, large numbers of PCNA+ cells were observed in corpus callosum of Cxcr2+/−→Cxcr2−/− and Cxcr2+/−→Cxcr2+/+ mice (Fig. 6A). Few PCNA-immunoreactive cells were observed in the control animals (Fig. 6A). However, by dual-label immunohistochemistry using PCNA and PDGFRα, the fraction of PDGFRα cells expressing PCNA was significantly higher (Fig. 6A; 6C) in Cxcr2+/−→Cxcr2−/− mice (~52%), as compared with Cxcr2+/−→Cxcr2+/+ mice (~30%) (P<0.05). Furthermore, proliferating PDGFRα cells in the corpus callosum of Cxcr2+/−→Cxcr2−/− mice quickly gave rise to oligodendroglia, with significantly (P<0.05) larger numbers of GST-π+ oligodendrocytes observed at 3 days of recovery following 6 wks of cuprizone feeding (Fig. 6B; 6D), which is about 4 days earlier than Cxcr2+/−→Cxcr2+/+ mice. In summary, early increased local OPC proliferation and differentiation in Cxcr2+/−→Cxcr2−/− mice may contribute to the enhanced myelin repair.
Cxcr2−/− mice are valuable for evaluating the function of CXCR2, but it is crucial to perform complementary experiments using receptor blockade, given the developmental effects of genetic deletion of CXCR2. It was not feasible to administer blocking antibodies during the entire time course of remyelination during EAE or cuprizone-intoxication because chronic administration of an antibody will cause unpredictable side effects in mice, including induction of neutrolizing antibodies. Therefore, we modified an in vitro model of demyelination/remyelination in lysolecithin-treated cerebellar slice cultures (Birgbauer et al., 2004) for application in mice. In order to address the effects of blocking CXCR2 during demyelination, slices were incubated in culture to permit myelination, and analyzed by dual label immunohistochemistry for myelin protein MBP (green) and neurofilament (red) detected in nerve fibers, with myelinated fibers appearing yellow in merged images (Fig. 7A: i–j). After 17 hrs of lysolecithin exposure, cerebellar slices were extensively demyelinated with abundant green myelin debris and numerous red demyelinated fibers (Fig. 7A: a–d). Equivalent demyelination was observed in slice preparations exposed to medium (not shown); neutralizing anti-CXCR2 antibodies (Fig. 7A: b); and control rabbit IgG (Fig. 7A: a, Fig. 7B) (P>0.05). At this time point, robust demyelination was also observed in slices from littermate Cxcr2+/+ (Fig. 7A: c) and −/− mice (Fig. 7A: d). After 2 days of recovery, vigorous remyelination was detected in slices incubated with anti-CXCR2 antibodies (Fig. 7A: f) and in those from Cxcr2−/− mice (Fig. 7A: h), while control slice preparations remained significantly demyelinated (Fig. 7A: e and g, Fig 7B) (P<0.01). At the two-day time point, remyelination in control slice cultures (Fig. 7A: e and g) was not significantly different from that observed after the overnight incubation with LPC (Fig. 7A: a and c; Fig 7B) (P>0.05). By contrast, in cultures in which CXCR2 had been blocked or deleted, myelin integrity was equivalent to that observed in control cultures which had not been subjected to demyelination (Fig. 7A: i–j) (P>0.05). These results were quantified and shown as Fig. 7B. These results indicate that inactivation of CXCR2 by genetic deletion or by antibody-mediated blockade exerted similar effects, and strongly enhanced the efficiency of remyelination in vitro.
We evaluated CXCR2-deficent mice in two in vivo models and one in vitro model of demyelination/remyelination to demonstrate a selective and potent function of CXCR2 in accelerating remyelination. The findings from our study point to an important inhibitory role for CXCR2 signaling in remyelination during EAE and cuprizone-induced demyelination and in vitro LPC-induced demyelination of cerebellum slice culture. Kerstetter (Kerstetter et al., 2009) reported that systemic injection of CXCR2 antagonist decreased the demyelinated lesions in the EAE model, although it remained unclear whether systemic injection of CXCR2 antagonist prevented demyelination or promoted remyelination (Liu et al., 2010; Carlson et al., 2008). We addressed these distinctions by studying CXCR2-deficient mice. Our findings confirmed previous reports and extended these observations.
Remyelination is the best documented and most robust example of tissue repair in the human CNS and is unequivocally neuroprotective in MS (Kornek and Lassmann, 1999; Kornek et al., 2000). About 40% of post mortem MS tissues demonstrate remyelination, which can occur both early and late in the course of MS (Patrikios et al., 2006). Factors that promote remyelination in vivo have been elusive.
Inhibiting CXCR2 for the treatment of inflammatory demyelination could simultaneously suppress inflammation and promote remyelination. The translational insights from using CXCR2-deficient mice were previously believed to be limited due to differences between humans and mice in the complement of receptors for ELR+ CXC chemokines (Mestas and Hughes, 2004). However, recent identification of murine CXCR1 (Fan et al., 2007) relieved this concern, and our present results clearly show a non-redundant function for CXCR2 in remyelination. The roles of CXCR2 on hematopoietic cells were differentiated by generating and evaluating Cxcr2+/−→Cxcr2−/− and Cxcr2+/−→Cxcr2+/+ radiation bone marrow chimeric mice. Transfer of CXCR2+ hematopoietic cells rendered recipient mice susceptible to EAE, regardless of whether the host genotype was Cxcr2−/− or Cxcr2+/+. In the SWXJ background, robust EAE-associated neurological impairment and histopathological changes were observed, despite CXCR2 haploinsufficiency of the donor marrow. Similarly, on the C57BL/6 background, Cxcr2+/−→Cxcr2−/− and Cxcr2+/−→Cxcr2+/+ mice demonstrated severe demyelination of the corpus callosum.
Furthermore, comparison of Cxcr2+/−→Cxcr2+/+ mice with wild-type C57BL/6 mice demonstrated equivalent demyelination and repair after cuprizone feeding, showing that the radiation chimerism did not materially alter the sequence of tissue injury and restoration in this model (Liu et al., 2010). Therefore, our experimental strategy afforded an opportunity to evaluate the selective role of CXCR2 in myelin repair.
Despite finding an important role of CXCR2 in myelin repair in three different models, the detailed cellular requirements for remyelination of EAE and cuprizone-induced demyelinating lesions remain uncertain. Varied CNS cells, including OPCs, astrocytes and neurons have been reported to express CXCR2, either in vitro or in vivo. Limited levels of CXCR2 mRNA impaired localization of the CXCR2 message by in situ hybridization, and available antibodies were found to be non-specific since they equally detected immunoreactivity in the CNS of Cxcr2−/− and Cxcr2+/+ mice (Liu et al., 2010). By PCR on sorted oligodendrocytes, we found that oligodendrocytes highly express CXCR2 (data not shown). We favor the possibility that OPC expression of CXCR2 accounts for our results, as we observed enhanced OPC proliferation in mice lacking CNS CXCR2 in both EAE and cuprizone-induced demyelination. This result seemed initially counter-intuitive, as CXCL1, a CXCR2 ligand, synergistically augmented proliferation of OPCs in vitro, in the presence of PDGF. In recent years, the neuro-restorative properties of the acute lesion environment in rodent EAE and cuprizone-induced demyelination have been extensively studied (Franklin, 2002) and characterized as comprising IGF-1 (Komoly et al., 1992) as well as LIF, CNTF, PDGF-AA, TGF-beta1 and FGF-2. Factors that promote remyelination are made by resident astrocytes and microglia as well as infiltrating hematopoietic cells. ‘Cocktails’ of growth factors enhance repair of cuprizone lesions upon intracerebral microinjection (Kumar et al., 2007), and growth-factor secretion by macrophages stimulates myelin production by brain-aggregate cultures (Diemel et al., 2004). Surprisingly, some inflammatory cytokines that govern lesion formation also promote repair of cuprizone-induced demyelination, demonstrating the intimate inter-relationship between demyelination and remyelination, and the restorative properties of acute inflammatory lesions (Mason et al., 2001; Gao et al., 2000). We propose that OPCs from Cxcr2−/− mice accumulated in the growth-promoting microenvironment of inflammatory demyelinating lesions with increased efficiency. As a corollary, we hypothesize that CXCR2-bearing OPCs were arrested at lesion edges, where CXCR2 ligands were preferentially expressed (Omari et al., 2005; Glabinski et al., 1997). These results indicate that the factors which orchestrate migration, proliferation and differentiation during development of the oligodendrocyte lineage in the mammalian CNS do not enable efficient myelin repair in the models under study here and suggest that therapeutic intervention at this level may provide tangible benefits.
Taken together, CXCR2 is the first chemokine receptor shown to contribute robustly to both the inflammatory processes that cause demyelination and to the pathways that control remyelination (Liu et al., 2010). Our studies address the fundamental role of CXCR2 in demyelination and remyelination, and provide information needed for effective clinical trials of CXCR2 blockade in MS. Future studies will address how the presence of CXCR2 impairs remyelination, and define CNS cell types whose expression of CXCR2 leads to inefficient remyelination.
This research was supported by grants from the National MS Society (RG 3580 to RMR), the National Institutes of Health (NS32151 to RMR). We thank W. Stallcup (Burnham Institute for Medical Research, La Jolla, CA) for generously providing antibody to PDGF receptor alpha and Dr. Chris Nelson (Dept of Neurosciences, Lerner Research Institute, Cleveland Clinic) for assistance with preparing the manuscript.