Cytokine and chemokine alterations in tissue, CSF, and plasma in early presymptomatic phase of experimental allergic encephalomyelitis (EAE), in a rat model of multiple sclerosis

Dark agouti (DA) (Harlan, Italy) female rats, 151–167-g body weight, were used in the first part of the study, placed on ad libitum food and water, and housed three per cage on a 12-h light/dark cycle. The DA rat model of EAE mimics certain aspects of the clinical course of the disease in people with RRMS [17‐19], typified by progressive, sustained demyelination, and associated axonal loss [20, 21]. The disease onset in this strain bears is characterized by neurological impairments, manifested as a flaccid tail followed by an acute attack with disturbed gait and paralysis. Most DA animals recover spontaneously from paralysis and experience remission and may then undergo one or more relapses.

A group of 42 rats was sensitized (considering group composition and two extra animals) by a medium containing 0.15 mg/ml guinea pig spinal cord tissue in complete Freund’s adjuvant (CFA, Sigma, Saint Louse, USA), 50% v/v, to which 5 mg/ml of heat-inactivated Mycobacterium tuberculosis (Difco H37Ra, DB, Milan, Italy) was added. Sensitization was performed by injecting 100 μl in both hind pads. Control rats (n = 8) and adjuvant-injected rats (CFA, 50% v/v, heat-inactivated M. tuberculosis, 5 mg/ml) (n = 15) were used. The rats were weighed daily and examined for clinical signs of EAE, according to the following semi quantitative score for neurological disability: 0 = no signs, 1 = loss of tail tone, 2 = mono or bilateral weakness of hind legs or middle ataxia, 3 = ataxia or paralysis, 4 = severe hind legs paralysis, and 5 = severe hind leg paralysis and urinary incontinence [22]. In view of the animals’ disability, wet food was included inside the cages to facilitate feeding. At 1, 5, 8, 11, and 18 DPI, eight EAE rats were randomly sacrificed. From each experimental group, five animals were used for proteomic and molecular biology experiments and three for morphology and immunohistochemistry. All animal protocols described herein were carried out in accordance with the European Community Council Directives (86/609/EEC), approved by the intramural ethical committee for animal experimentation of Bologna University and the Ministry of Health (no. 158/2013-B) and comply with the guidelines published in the NIH Guide for the Care and Use of Laboratory Animals [23].

In order to calculate the number of animals needed to study the effect of the treatment with GW2580, we performed a power analysis using the G*Power 3.1 software. To reach a power of 0.9, based on retrospective analysis of recent research done by others [24, 25], we needed a minimum of n = 6 animals per experimental group.

The experiments were designed in compliance with the ARRIVE guidelines, randomizing the procedures and applying blinded analysis.

GW2580 (LC Laboratories, Boston, USA) was dissolved in 0.5% hydroxypropyl methylcellulose and 0.1% Tween 80 [26, 27]. The experimental groups made of n = 6 animals each were as follows: vehicle (0.5% hydroxypropyl methylcellulose/0.1% Tween 80); control + GW2580; EAE; and EAE + GW2580. The rats (control + GW2580 and EAE + GW2580 groups) were under GW2580 treatment at 40 mg/kg once daily by oral gavage using flexible plastic feeding tubes FTP-15-78-50 (Instech Laboratories, Netherlands) for 1 day prior and 11 consecutive days following the immunization. All animals were weighed daily and the EAE and EAE + GW2580 groups were examined for clinical signs of EAE till the 18 DPI (last day of the experiment) following the semi quantitative score for neurological disability as already explained in the first part of the study. Animal protocols described herein were carried out in accordance with the European Community Council Directives (86/609/EEC), approved by the intramural ethical committee for animal experimentation of Bologna University and the Ministry of Health (no. 607/2016-PR).

On post-immunization days 1, 5, 8, 11, and 18, three rats were scarified and the lumbar spinal cord (LSC) tissues of the control, EAE, and adjuvant groups were fixed in 4% paraformaldehyde and picric acid and saturated aqueous solution in 0.1 M Sörensen buffer pH 7 for 24 h and then washed with 5% sucrose solution every day till the fixative was completely removed. The tissues were frozen with CO₂ gas and kept at −80 °C until processed. Coronal sections (14-μm thickness) were then prepared (Microm HN550, Bio-Optica, Milan, Italy) and processed for morphological studies. For immunohistochemistry, the sections were first rehydrated and then incubated for 1 h with PBS–0.3% Triton X-100, 2% normal serum goat, and 1% BSA, followed by incubation with the primary antibodies diluted in the pre-absorption solution overnight at 4°C. The primary antibodies and dilutions used were as follows: rabbit anti-CD44 (1:100, Cluster of differentiation 44, Acris Antibodies, Inc, San Diego, USA), rabbit anti-CD163 (1:100, Cluster of differentiation 163, Bioss Inc, Woburn, USA), mouse anti-CD86 (1:250, Cluster of differentiation 86, Novus Biological Europe, Cambridge, UK), mouse anti-NF200 (1:200, Neurofilament, Sigma, Saint Louse, USA), rabbit anti CSF1 (1:100, colony-stimulating factor 1, Novus Biological), mouse anti-OX42 (1:250, AbD Serotec Inc., Raleigh, USA), mouse anti-NG2 (1:100, membrane-spanning chondroitin sulfate proteoglycan, Millipore, Merck S.p.a., Milan, Italy), mouse anti-CNPase (1:250, 2′, 3′-cyclic nucleotide 3′-phosphodiesterase, Millipore, Merck S.p.a., Milan, Italy), rabbit anti-GFAP (1:500, glial fibrillary acidic protein, Dako, Milan, Italy), and mouse anti-GFAP (1:500, Sigma, Saint Louse, USA). FluoroMyelin™ Fluorescence Myelin Staining (Molecular Probes, Eugene, OR) was also performed to stain the myelin sheaths, following the manufacturer’s specifications. After rinsing in PBS, the sections were incubated at 37 °C for 30 min with the secondary antibodies: DyLight488-conjugated affinity-pure goat anti-mouse IgG (ThermoScientific, Milano, Italy), Rhodamine Red™-X-conjugated affinity-pure donkey anti-rabbit IgG (Jackson Immunoresearch, West Grove, PA, USA), DyLight488-conjugated affinity-pure donkey anti-rabbit IgG (ThermoScientific, Milano, Italy), and Red™-X-conjugated affinity-pure donkey anti-goat IgG (Jackson Immunoresearch, West Grove, PA, USA) diluted in PBS–0.3% Triton X-100, 1:100. The sections were then rinsed in PBS and mounted in phenylenediamine solution (0.1% 1,4-phenylenediamine, 50% glycerine, carbonate/bicarbonate buffer pH 8.6, Sigma, Saint Louse, USA). The control slices were incubated with the secondary antibodies only and processed in parallel. Images were captured using a Nikon Eclipse E600 microscope equipped with digital CCD camera Q Imaging Retiga-2000RV (Q Imaging, Surrey, BC, Canada). To analyze the inflammatory infiltration, the sections were stained with toluidine blue and evaluated on three replicate sections per animal, counting the number and severity of cellular infiltrates over each whole coronal section. Cellular infiltrates were scored as follows: 0, none; 1, a few inflammatory cells; 2, organization of perivascular infiltrates; and 3, increasing severity of perivascular cuffing with extension into the adjacent tissue [28]. The inflammation score derives from the sum of infiltration scores in each cellular infiltrate.

At post-immunization days 1, 5, 8, 11, and 18, five rats were sacrificed and the total LSC was dissected, snap frozen, and stored at −80 °C until used.

The total RNA was prepared from the spinal cord using QIAzol Reagent, cleaned with RNeasy Mini kit (Qiagen; Milano- Italy), and eluted in RNase Free Water, and purity and concentration were evaluated by spectrophotometry using NanoDrop ND-2000 (ThermoScientific, Milano, Italy). Complementary DNA (cDNA) synthesis was performed using RT² First Strand kit following the manufacturer’s instructions. In brief, after incubation for 5 min at 42 °C with genomic DNA elimination mix in order to avoid any DNA contamination, a reverse-transcription mix of BC3, P2, RE2, and H₂O was used and the transcription was performed in a final volume of 20 μl by heating first at 42 °C for 15 min, then at 95 °C for 5 min. Real-time PCR amplification was performed using the Stratagene Mx3005P multiplex quantitative PCR system (Agilent Technologies). The expression of genes involved in MS progression was carried out using RT² SYBR Green qPCR Mastermix and the Multiple Sclerosis RT2 Profiler PCR Array (Qiagen). The genes are grouped according to the following: T cell activation/signaling, adaptive immunity, cytokine/chemokine inflammation, demyelination, and cellular stress. The raw data obtained was uploaded into GeneGlobe Data Analysis software (SABiosciences, Qiagen) for analysis. Relative quantification of messenger RNA (mRNA) expression was calculated using the comparative cycle threshold (C_T) method and is expressed as Log2 fold change of expression. The fold change (2^(−Delta Delta Ct)) is the normalized gene expression (2^(−Delta Ct)) in the test sample divided by the normalized gene expression (2^(−Delta Ct)) in the control sample.

CSF sampling was adapted from the method of Liu and Duff [29]. Briefly, the rats were anesthetized by isoflurane inhalation (Gas Anesthesia System-21100, Ugo Basile, Varese, Italy) and placed prone on the stereotaxic instrument letting the body of the rat laid down. A sagittal incision of the skin was made below the occiput, and the subcutaneous tissue and neck muscles through the midline were separated and held apart using a microretractor. The dura mater of the cisterna magna was then penetrated by an 8-cm long glass capillary, which had a narrowed tip with an inner diameter of about 0.5 mm so that the CSF flowed into the capillary. After collection, each sample was centrifuged at 2000×g for 10 min at 4 °C, and the supernatant aliquoted and stored at −80 °C for biochemical assays.

Proteins known to play key roles in neuroinflammation pathways were selected. For this purpose, Bio-Plex Pro™ Rat Cytokine 24-plex Assay (Bio-Rad; Milano, Italy) was used. The kit included EPO, G-CSF (CSF3), GM-CSF (CSF2), GRO/KC, IFN-γ, IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-10, IL-12p70, IL-13, IL-17A, IL-18, M-CSF (CSF1), MCP-1 (CCL2), MIP-1α (CCL3), MIP-3α (CCL20), RANTES (CCL5), TNF-α, and VEGF.

The simultaneous quantification of the different proteins in CSF and plasma was performed using xMAP technology and a MAGPIX Luminex platform. This technology makes use of different populations of color-coded beads of monoclonal antibodies specific to a particular protein, thus allowing simultaneous capture and detection of specific analytes from a sample. All the beads from each set are read off, which further validates the results. Using this process, xMAP Technology allows multiplexing of up to 50 unique bioassays within a single sample, both rapidly and precisely [30, 31]. In brief, after the incubation of a specific monoclonal antibody conjugated bead population with 50 μl of CSF/plasma samples for 1 h at RT, washed beads were incubated with detection antibody solution at RT for 30 min, then with the streptavidin–phycoerythrin conjugated solution (RT, 10 min). After washing, beads were resuspended in the assay buffer, shaken for 1 min and then a reading performed on the MAGPIX instrument. The results were analyzed with xPONENT 4.2 ® software and expressed as pg/ml.

Student’s t test to compare means of two experimental groups, one-way ANOVA followed by Dunnett’s multiple comparison tests, and two-way ANOVA followed by Bonferroni post-test were used. Data are presented as mean ± standard error of the mean, and significance was set at P ≤ 0.05. All statistical analyses were performed using GraphPad Prism 6.0 (GraphPad Software).

For the normalization of gene expression on the RT² PCR Profiler Array, five housekeeping genes, ribosomal protein, large, P1, hypoxanthine phosphoribosyltransferase 1, ribosomal protein L13A, lactate dehydrogenase A, and actin beta were used. The CT was determined for each sample and normalized to the average CT of the five housekeeping genes. A comparative CT method was used to calculate relative gene expression. Data are represented as Log2 fold change relative to control. The P values were calculated on the basis of a Student’s t test of the replicate 2^(–Delta Ct) values for each gene in the control group and treatment groups, and P values less than 0.05 was considered significant.

The clinical profile of EAE is reported in Fig. 1a, b in which the clinical score (a) and body weight graphs (b) are shown. Clinical signs of neurological disabilities in EAE started at 7–8 DPI and reached the higher score at 11 DPI (acute phase). A remission phase is then observed, from 12 to 15 DPI, followed by a rapid increase in the clinical score (relapsing phase). Figure 1b shows the body weight gain, which decreases in EAE groups from 2 DPI compared to the control group (P < 0.001). A difference between EAE and adjuvant groups is also observed from 7 to 12 DPI (P = 0.0139), in close correspondence with the evolution of the symptoms of the disorder, which is followed by a partial recovery.

Histopathology was performed in the LSC at each time point investigated, focusing on inflammation and demyelination. Representative images of toluidine blue staining (C-H) and immunofluorescence visualization of OX42-IR microglia cells (I-L), GFAP-IR astrocytes (M-P), and myelin (T-W) are reported in Fig. 1. Toluidine staining illustrates the massive cellular infiltrate starting from 11 DPI. The semiquantitative evaluation reveals a severe and diffuse cellular infiltrate (P < 0.0001) in the LSC at the acute and the remission phases (11, 18 DPI) (Fig. 1q). Significantly, microglial reaction starts at 8 DPI (P = 0.0001) (Fig. 1r), when OX42-IR cells take on a rounder morphology (see high-magnification inserts in Fig. 1i, j, EAE t8, compared to the control) and converge around blood vessels, reaching a peak at 11 DPI (P < 0.0001). The astroglial reaction was also analyzed by GFAP-immunostaining and a higher immunoreactivity was observed at 8 DPI (P = 0.0005) (Fig. 1m, n, control, compared to EAE t8). Demyelination signs appear at 8 DPI as revealed by the slight decrease in FluoroMyelin staining intensity in the gracile fasciculus (gr) of the LSC, (Fig. 1t, u, control, compared to EAE t8; see arrows), which is very extensive at 11 DPI (Fig. 1t, v, EAE t11, compared to the control; see plus) and has partially recovered at 18 DPI (Fig. 1t, w, control, compared to EAE t18; see asterisks).

The mRNA expression of genes that encode for T cell activation/signaling, adaptive immunity, cytokines/chemokines, and cellular stress involved in neuroinflammation and demyelination processes were studied in the SC by real-time PCR. The complete list of investigated genes is presented in Fig. 2. For biological averaging and variance reduction, samples from each group were pooled for microarray experiments, in fact, for very small designs, pooling dramatically improves accuracy [32‐34]. The results from the different groups are presented in a clustergram that performs non-supervised hierarchical clustering to display a heat map with dendrograms indicating co-regulated genes across groups, criteria for significance are reported in the table of magnitude gene expression (Fig. 2). Several pro-inflammatory cytokines and chemokines such as IL-1β and CCL12 were highly up-regulated in the SC at 8 DPI and reached a peak at 11 DPI. IFN-γ, CXCL11, and LTA were the most up-regulated genes, showing more than 22 Log2 fold change compared to the controls at 8 DPI, around 115 at 11 DPI, and more than 67 Log2 fold change at 18 DPI. The highest up-regulation was observed for IFN-γ, 68.12 Log2 fold change at 8 DPI, 179.15 at 11 DPI, and 98.7 at 18 DPI (Fig. 2). Notably, ADAM17 and CSF1 mRNAs are strongly down-regulated at 1 DPI. Most of the anti-inflammatory cytokines and chemokines were down-regulated with a profile opposite to that observed for the pro-inflammatory ones (starting to decrease from 8 DPI), including VEGF, SOD1, NTF3, APC, HEXB, while an up-regulation was observed for some anti-inflammatory mediators such as GPX1 and IL-10. With regard to genes involved in myelination, a down-regulation starting from 8 DPI was observed, including MAL, MBP and PMP22. To note, the overexpression of IFN-γ mRNA in EAE in acute phase (11 DPI) is the highest EAE-induced up-regulation observed compared to the other inflammatory mediators investigated (Fig. 2).

In order to evaluate the kinetics of inflammation biomarkers during EAE, 24 cytokines and chemokines, EPO, G-CSF (CSF3), GM-CSF (CSF2), GRO/KC, IFN-γ, IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-10, IL-12p70, IL-13, IL-17A, IL-18, M-CSF (CSF1), MCP-1 (CCL2), MIP-1α (CCL3), MIP-3α (CCL20), RANTES (CCL5), TNF-α, and VEGF, were simultaneously quantified in CSF samples at the different time points investigated (1, 5, 8, 11, and 18 DPI, Fig. 3a, c). IL-17A, IL-1β, TNF-α, and IL-10 were significantly increased at 8 DPI in EAE group compared to the control animals (P = 0.0412; P = 0.0127; P = 0.0039; P = 0.0412, respectively), reaching a peak at 11 DPI (P < 0.0001; P < 0.0001; P < 0.0001; P = 0.0074, respectively) and decreasing again at 18 DPI (ns; P = 0.0035; P < 0.0001; P = 0.0128, respectively). For both pro-inflammatory chemokines CCL2 and CCL3 and for the anti-inflammatory cytokines IL-5, the same significant profile was observed, reaching a peak at 8 DPI (P < 0.0001; P = 0.001; P < 0.0001, respectively) then starting to decrease starting from 11 DPI (P < 0.0001; P = 0.0046; P < 0.0001, respectively). The levels of CCL20 and IL-6 were significantly increased in EAE animals only at 8 DPI (P < 0.0001; P = 0.004, respectively) (Fig. 3). Unpredictably, IFN-γ, IL-13, IL-2, IL-1α, IL-4, IL-12p70, and GM-CSF were not detected at any of the time points analyzed in the CSF in our experimental conditions.

The different cytokines/chemokines quantified in CSF were also simultaneously measured in plasma samples (Fig. 3b, d) in order to both correlate with CSF changes and to evaluate the peripheral effect of adjuvant. IL-1β increased in EAE compared to control group of animals at 8 DPI (P = 0.0231). GM-CSF was the only cytokine not detectable at any of the time points analyzed, and no significant changes was observed for the rest of the panel compared to the control.

No significant variations between adjuvant and control groups were observed in CSF or plasma at the investigated times.

In order to elucidate the cell type producing the most regulated biomarkers, double-labeling experiments were performed in the LSC of the control and EAE animals. Representative images from the control and EAE animals at 8, 11, and 18 DPI are shown in Fig. 4. The following markers were investigated: OX42 as marker for microglia; CD86, as marker for M1 macrophages phenotype; CD163, as marker for M2 macrophages phenotype; CD44, as marker for T lymphocytes; NG2, as marker for oligodendrocyte precursor cells (OPCs); and CNPase, as marker for oligodendrocytes. Immunostaining was quantified by a computerized procedure to evaluate the immunoreactive area. As expected, the increase in OX42-IR observed at 8 DPI corresponds to increased expression of the M2 cytokine CD163 (Fig. 4a–d, see plus in B, M), while the M1 macrophage marker CD86 showed the highest immunoreactive area at 11 DPI (Fig. 4i–l, p). This activation was accompanied by an increase in the CNPase immunoreactive (IR) area starting from 8 DPI (Fig. 4e–h, n), which partially contrasts with the FluoroMyelin staining results (Fig. 1t–w). In the control group, CD44 was expressed by oligodendrocytes (Fig. 4e; see arrows), while this co-localization was no longer present in the EAE animals at 8, 11, and 18 DPI (Fig. 4f; see asterisk). The highest up-regulation of the different markers verified in the white matter was detected in the acute phase of EAE (Fig. 4c, g, and k), where also a significant up-regulation of CD163 (P = 0.0016), CNPase (P = 0.0197), CD44 (P = 0.0005), and CD86-IR (P < 0.0001) areas were observed at 11 DPI (Fig. 4m, n, o, and p, respectively).

No significant results were observed in EAE t1, EAE t5, and adjuvant experimental groups (results not shown).

Going one step deeper into biomarker regulation at the early phase of EAE, it was found that the pro-inflammatory mediator CSF1 (The CSF-1R is a member of the platelet-derived growth factor receptor (PDGFR) family of class III receptor tyrosine kinases that includes PDGFRα/β, stem cell factor receptor (c-Kit), and Flt3/Flk2) [35] was regulated starting from 1 DPI in EAE group compared to the control in all the biological samples analyzed (Fig. 5a). In plasma, CSF1 was significantly up-regulated only at 1 DPI (P = 0.0006), then continuously decreased till the last time point studied, 18 DPI (Fig. 6b), while in CSF, CSF1 level started to decrease at 1 DPI, becoming significant at 11 and 18 DPI (P < 0.0001, P < 0.0001) (Fig. 5c). Notably, CSF1-IR in the tissue also increased starting from 1 DPI (P = 0.0003) (Fig. 5e, k), then decreased at 5 and 8 DPI (Fig. 5g, h) ultimately reaching the highest immunoreactivity at 11 DPI (P < 0.0001) (Fig. 5i, k). The mRNA expression of CSF1 gene in SC was also analyzed (Fig. 2), and a down-regulation at 1 and 11 DPI was observed (Fig. 5d). In the attempt to identify the CSF1-producing cells at 1 and 11 DPI, double-labeling immunostaining of CSF1 with CNPase and GFAP was performed in the LSC (Fig. 5l–o). Figure 6n, o shows that CSF1 is constitutively expressed by astrocytes not only in the EAE animals but also in the control (results not shown). Interestingly, CSF1 was expressed by oligodendrocytes at 1 DPI in EAE (Fig. 5l), while this expression is no longer present as from 11 DPI (Fig. 5m).

In an exploratory study, a pathway analysis approach of the different proteins derived from the most early regulated genes at 8 DPI was implemented. This time point corresponds to the beginning of the regulation of genes in the CNS, while cellular infiltrates are not yet present and clinical signs start to appear. We proceed through the web-software STRING 10.0 (http://​string-db.​org/​) and Gene Ontology databases. The protein–protein interaction analysis (both physical and functional interactions) was performed using default parameters (high confidence, 0.7) and R.norvegicus as the organism of interest. The software allows the net of interactions including other proteins closely linked to the one analyzed to be extended, in order to obtain a better understanding of the possible pathways affected by EAE in DA rats at early onset. The analysis showed that almost all the proteins deriving from genes of interest (except for the CXC family) did not directly interact with each other (Fig. 6a). The extended network (indirect protein–protein interactions) showed that those proteins were connected in four clusters according to their involvement in the biological process (autoimmune disease), situating STAT1 (signal transducers and activators of transcription), FLT1 (vascular endothelial growth factor receptor 1), and TRAF6 (TNF receptor associated factor) as nodes of the extended net linking the different groups (Fig. 6b). To gain further insight into the biological significance of those clusters in MS, an enrichment analysis using Gene Ontology was performed. It was found that these four clusters consist of a Wnt signaling pathway (the green cluster), a cytokine-mediated signaling pathway (the yellow cluster), T cell chemotaxis (the blue cluster), and positive regulation of mitogen-activated protein tyrosine kinase (MAPK) (the red cluster). Notably, CSF1 takes part of MAPK cluster.

Since the peculiar temporal expression pattern of CSF1 in particular in the very early phase of the disease, the tyrosine kinase activity of CSF1R was inhibited by the oral administration of GW2580. GW2580 is a highly selective inhibitor of the c-FMS kinase, and through this pathway, this small molecule blocks CSF1 signaling. Treatment started 1 day before the immunization and till 11 DPI. The clinical profiles of the disease in EAE, EAE + GW2580, control + GW2580, and vehicle groups of animals are reported in Fig. 7, in which the clinical score (a) and body weight graphs (b) are shown. Clinical signs of the neurological disabilities in EAE were the same as reported in the first group of animals included in our study (Fig. 1a), which prove the reproducibility of the model. While the clinical signs started at 6–7 DPI in the EAE group, a delay in the appearance of the neurological disability was observed in the EAE + GW2580 group. The acute phase was at 11–12 DPI with a maximum score of 5, while a significant reduction of the severity of the disease was observed in the EAE + GW2580 group (P < 0.001) with a maximum score of 2.8. Interestingly, the EAE + GW2580 animals did not show any relapse phase (P < 0.001) and after the acute phase, they were recovering till the 18 DPI, last day of the experiment (P < 0.001). Figure 7b shows the body weight gain. A decrease was observed in the EAE + GW2580 groups from 2 DPI compared to the GW2580 animals (P < 0.001), which is the same profile of the body weight loss observed in the EAE group but less severe, in close correspondence with the evolution of the symptoms of the disorder, which is followed by a recovery. We also calculated the cumulative disability index, reported in Fig. 7c, observed in EAE compared to the EAE + GW2580 groups (P < 0.001).