CXCR7 antagonism prevents axonal injury during experimental autoimmune encephalomyelitis as revealed by in vivoaxial diffusivity

C57BL/6 mice (Jackson Labs, Bar Harbor, ME) were maintained in pathogen free conditions (Department of Comparative Medicine, Washington University, St. Louis, MO) and studies were performed in compliance with the guidelines of the Washington University School of Medicine Animal Studies Committee. Antibodies utilized include rabbit anti-CD3 (Dako, Glostrup, Denmark), rat anti-GFAP (Invitrogen, Carlsbad, CA), rabbit anti-GST-π (Assays Designs-Enzo Life Sciences, Inc. Farmingdale, NY), rat anti-MBP (Abcam, Cambridge, MA) and mouse monoclonal SMI-32 (Covance, Emeryville, CA). IgG from rabbit, rat (Invitrogen, Carlsbad, CA) and mouse (BD Pharmigen, San Diego, CA) were used as isotype controls. Secondary detection was done using goat anti-rat, anti-rabbit and anti-mouse conjugated to Alexa-555 and goat anti-rabbit, and anti-rat conjugated to Alexa-488 (Molecular Probes-Invitrogen, Carlsbad, CA).

Autoreactive Th1 cells directed to myelin oligodendrocyte glycoprotein peptide (MOGp) were generated as previously described [12]. Briefly, naïve C57BL/6 mice were immunized with 50 μg msMOGp35-55 (GenScript, Piscataway, NJ) and 50 μg Mycobacterium tuberculosis H37Ra peptide emulsified in Freund's adjuvant (both from Difco Laboratories, Detroit, MI). After 14 days, polarized Th1 cells were harvested from spleen using a nylon wool column. MOG-reactive Th1 cells were restimulated and expanded in culture according to standard protocols prior to adoptive transfer to naïve recipients [12]. 5 × 10⁶ Th1 cells were incubated with 25 × 10⁶ stimulators-cells obtained from naïve splenocytes. Incubation proceeded in the presence of MOG, anti-IL-4, IL-2 (both generated in the lab) and IL-12 (BD Pharmigen, San Diego, CA) in RPMI enriched media. Cells were incubated at 37°C for 7 days. After separation with Histopaque (Sigma Aldrich, Saint Louis, MO) 1 × 10⁷ Th1 cells were restimulated with 5 × 10⁷ stimulators-cells for other 7 days as described above. Th1 cells underwent a third stimulation of only 4 days in the presence of MOG and IL-2. Following Histopaque separation cells were resuspended in HBSS at 1 × 10⁷ MOG-reactive Th1 cells/300 ul per mouse ratio. Adoptive transfer of 10 × 10⁶ MOG-reactive Th1 cells per mouse was done retro-orbitally. Recipient mice were monitored for clinical manifestations of EAE by following their body weight and graded for disease with the following score system: 1, tail weakness; 2, difficulty righting; 3, hind limb paralysis; 4, forelimb weakness; 5, moribund or dead.

Mice underwent daily subcutaneous injection of either vehicle (10% Captisol) or CCX771 (ChemoCentryx, Mountainview, CA) at doses of 5 or 10 mg/kg of body weight beginning 12 hours after adoptive transfer (5 mice per treatment group). A cohort of mice started with vehicle and changed to CCX771 at a dose of 10 mg/kg of body weight when achieved a score of 1. Dosing with CXCR7 antagonist or vehicle and monitoring for clinical progression continued for 28 days.

Axial scout images of the spine were acquired for proper localization of spinal cord level. Multiple transverse slices covering spinal cord lumbar enlargement were obtained using a Stejskal-Tanner [16, 17, 20, 26] spin-echo diffusion weighted sequence with the following acquisition parameters: TR ~1500 msec. (determined by the respiratory rate of the mouse), TE of 37 msec., number of excitations equals 2, slice thickness of 1 mm, spinal cord field of view of 1 cm × 1 cm, data matrix of 128 × 128 (zero-filled to 256 × 256). Diffusion-sensitizing gradients were applied in six orientations: (Gx, Gy, Gz) equal to (1, 1, 0), (1, 0, 1), (0, 1, 1), (-1, 1, 0), (0, -1, 1), and (1, 0, -1) with a gradient strength of 9 G/cm, duration (δ) equal to 7 msec., and separation (Δ) of 18 msec., to obtain b values of 0 and 0.750 s/mm². Acquisition time was approximately one hour for each spinal cord scanning session.

Using a weighted linear least square method, diffusion tensors and DTI parameters were generated. In this study, three DTI parameter maps were used, relative anisotropy (RA), axial diffusivity (λ||), and radial diffusivity (λ⊥) [27]. Regions of interest (ROIs) encompassing the ventrolateral white matter (VLWM) were drawn manually on relative anisotropy (RA) maps projecting to axial and radial diffusivity maps to assess axon and myelin integrity respectively using ImageJ v1.37 software (developed at the U.S. National Institutes of Health and available on the Internet at http://​rsbweb.​nih.​gov/​ij/)[28‐30]. Injured VLWM region was identified using the baseline axial diffusivity threshold derived from the spinal cord of naïve mice.

Murine spinal cords were isolated for histological analysis after imaging. Briefly, anesthetized animals were intracardially perfused with PBS and fixed in 4% PFA followed by overnight post-fixation in 4% PFA. Spinal cords were cryoprotected in 30% sucrose prior to embedding in OCT media for cryosectioning. For lumbar segment identification luxol fast blue (LFB) was done following standard procedure and visualized on the Zeiss Axioskop MOT Fluorescent microscope and AxioVision software (both from Carl Zeiss International, Jena, Germany).

Immunodetection of CD3+ lymphocytes within CNS parenchyma was performed via co-labeling with GFAP. Sections from lumbar segments of the spinal cord were permeabilized and blocked in 0.1% Triton X-100 and 10% goat serum for 60 minutes at room temperature. Sections were incubated with primary antibody overnight at 4°C, washed in PBS, incubated with secondary antibodies for 60 minutes at room temperature and counterstained with ToPro-3 (Invitrogen, Carlsbad, CA) to detect nuclei. Immunostained sections were visualized on the Zeiss LSM 510 META Confocal Laser Scanning Microscope (Carl Zeiss International, Jena, Germany). Measurement of CD3+ pixels within the ROIs encompassing parenchyma and meninges was done using the public domain NIH Image program ImageJ.

To facilitate immunodetection of myelin basic protein (MBP) and glutathione S-transferase-π (GST-π) frozen sections from lumbar segments L2/L3 were submitted to antigen retrieval in 0.1% trypsin, 0.1% CaCl in 0.05 M Tris pH 7.4 at 37°C for 10 min. Then sections were permeabilized and blocked in 0.1% Triton X-100 and 3% goat serum for 60 minutes at room temperature. Primary antibody was incubated overnight at 4°C. Slides were washed in 0.2% fish skin gelatin in PBS followed by incubation of secondary antibodies for 60 minutes at room temperature. After washing in 0.2% fish skin gelatin in PBS, slides were nuclear stained with DAPI (Invitrogen, Carlsbad, CA) and coverslip with Vectashield (Vector Laboratories, Inc. Burlingame, CA) before being visualized on the Zeiss Axioskop MOT Fluorescent microscope and AxioVision software (both from Carl Zeiss International, Jena, Germany).

For immunodetection of neurofiliament H frozen sections from lumbar segments L2/L3 were permeabilized and blocked in 0.1% Triton X-100 and 10% goat serum for 60 minutes at room temperature. Primary antibody was incubated overnight at 4°C. Slides were washed in PBS followed by incubation of secondary antibodies for 60 minutes at room temperature. After washing in PBS, slides were nuclear stained with ToPro-3 (Invitrogen, Carlsbad, CA) and coverslip with Shandon-Mount (Thermo Fisher Scientific, Pittsburgh, PA) before being visualized on the Zeiss LSM 510 META Confocal Laser Scanning Microscope (Carl Zeiss International, Jena, Germany). Measurement of SMI-32 positive signal area within the ROIs encompassing the VLWM was done using the public domain NIH Image program ImageJ.

We recently demonstrated that CXCR7 antagonism ameliorates EAE induced by adoptive transfer of encephalitogenic Th1 cells by decreasing both leukocyte entry into the CNS parenchyma and demyelination [12]. Because both loss of myelin and interactions between immune and neuronal cells lead to axonal damage [31, 32], we wondered whether CXCR7 antagonism might affect the extent of recovery. Mice underwent adoptive transfer of MOG-reactive Th1 cells and were then administered daily injections of saline, vehicle, 5 or 10 mg/kg of CCX771, a specific antagonist of CXCR7 [12], and monitored for disease progression for a period of 28 days. In order to examine whether limiting inflammatory infiltrates during ongoing disease impacts on recovery, half of the vehicle-treated mice began administration of CCX771 when they attained a clinical score of 1. Consistent with our previous report [12], CXCR7 antagonism led to a dose dependent decrease in the peak severity of EAE even in mice with established disease (Figure 1A). Evaluation of disease severity during the recovery phase, 21 to 28 days post-transfer (blue bracket in Figure 1A), revealed that CCX771-treated mice exhibited a dose-dependent increase in recovery with significantly lower clinical severity scores than vehicle- or saline-treated control groups (Tables 1 and 2 and Figure 1B). All animals from both groups exhibited a highest severity score of 2 during the recovery phase, 21 to 28 days post-transfer (Figure 1B, recovery phase for vehicle and saline SEM = 0). Of interest, animals treated with CCX771 that attained a clinical score of 1 exhibited the same mean severity score upon recovery phase, 21 to 28 days post-transfer (blue bracket in Figure 1A), as those that began treatment at the time of adoptive transfer, despite a significant difference in their peak clinical disease severities during effector phase, 7 to 21 days post-transfer (red bracket in Figure 1A). (Figure 1A Two-way ANOVA interaction: F = 2.20, DFn = 108, DFd = 560, P < 0.0001; antagonist treatment: F = 93.85, DFn = 4, DFd = 560, P < 0.0001; day post-adoptive transfer: F = 59.06, DFn = 27, DFd = 560, P < 0.0001; Figure 1B Two-way ANOVA interaction: F = 1.71, P = 0.1659; disease phase: F = 164.57, P < 0.0001; treatment: F = 22.21, P < 0.0001 and Table 1: One-way ANOVA effector phase: F = 8.250, P > 0.0004; Table 2: One-way ANOVA recovery phase: F = 18.45, P < 0.0001). These data suggest that CXCR7 antagonism not only modifies disease severity during EAE but enhances overall recovery regardless of initial impairments, as accessed via clinical scoring.

In vivo DTI was performed to assess microstructural changes within multiple transverse slices of the lumbar enlargement of mouse spinal cords (Figure 2A) in CCX771-, vehicle- or saline-treated mice after recovery from peak EAE, plus age-matched naïve control mice. Relative anisotropy (RA) maps were generated within the manually defined regions of interest (ROIs) of VLWM (Figure 2B). Axon and myelin injury in VLWM, most severe in the control groups, was apparent in EAE mice as evidenced by the intensity changes in radial and axial diffusivity maps (Figure 2B). Statistical analysis of changes in radial diffusivity failed to show significant differences in the VLWM among study groups, suggesting no differences in myelin integrity (Figure 2C, One-way ANOVA F = 1.696, P = 0.1740). In contrast, analysis of changes in VLWM axial diffusivity detected significant decreases in groups of mice that received low dose CCX771 (5 mg/kg), vehicle or saline versus those that received high dose (10 mg/kg) and naïve mice (Figure 2D). Axial diffusivity of 10 mg/kg CCX771-treated mice resembled the values of the naïve group, while 5 mg/kg CCX771-treated mice resembled those obtained from control groups (Figure 2D, One-way ANOVA F = 3.232, P = 0.0227). Lastly, RA of VLWM showed no difference between vehicle- or saline-treated mice and CCX771-treated mice (Figure 2E, One-way ANOVA F = 5.272, P = 0.0021).

The extent of axonal preservation was assessed according to the axial diffusivity distribution [16] from the naïve spinal cords (Figure 3A) to distinguish the injured from the normal appearing VLWM (Figure 3B). No differences in the extent of injured VLWM were detected between 10 mg/kg CCX771-treated and naïve mice while 5 mg/kg CCX771-, vehicle-, and saline-treated mice exhibited gradations of injury within the VLWM (Figure 3C).

We next inquired if changes in axial diffusivity after CCX771 treatment correlated with disease recovery. To answer this question we evaluated the mean clinical score only during the recovery phase, 21 to 28 day post-adoptive transfer, as a function of axial diffusivity VLWM and injured VLWM. Both analyses showed a statistically significant linear correlation of clinical score during the recovery phase with axial diffusivity (Figure 4A, R² = 0.5413, F = 33.04, DFn, DFd = 1.0, 28.0, P < 0.0001) and injured VLWM (Figure 4B, R² = 0.6520, F = 43.10, DFn, DFd = 1.0, 23.0, P < 0.0001). In addition, linear correlation between the temporal-cumulative clinical score and injured VLWM at the time of DTI evaluation was significant (Figure 4C, R² = 0.7284, F = 61.69, DFn, DFd = 1.0, 23.0, P < 0.0001). These data strongly support a role for CXCR7 as a disease-modifying molecule during the recovery phase of EAE via axonal preservation.

Prior studies evaluating the effect of CXCR7 antagonism during EAE revealed decreased demyelination in mice treated with CCX771 compared with vehicle-treated animals at the peak of disease [12]. Because we did not observe any differences in measurements of radial diffusivity between treatment groups after disease recovery, we performed luxol fast blue (LFB) staining of spinal cords at lumbar segments L2/3. Consistent with our DTI data, there were no apparent differences between any treatment groups in the VLWM myelin density as detected by this method (Figure 5A). Of interest, mice treated with 10 mg/kg CCX771 exhibited myelin-density staining that was comparable to naïve animals (Figure 5A, upper panels). After 28 days post-adoptive transfer, persistent inflammation was still detectable within the VLWM spinal cords of saline- and vehicle-treated mice and, to some extent, of mice treated with 5 mg/kg CCX771, but not in mice treated with 10 mg/kg CCX771 (Figure 5A, lower panels versus middle and right upper panels). We also observed pathological features consistent with hypertrophic and degenerating axonal bundles, as reported in human and rodent models of CNS degeneration and trauma (Figure 5A, lower panels, arrowhead) [33‐39]. However, these features were absent in the naïve and 10 mg/kg CCX771-treated mice, where more normally appearing axons were present (Figure 5A, mid-upper panel, arrow).

To better assess myelin protein expression and oligodendrocyte numbers, we examined myelin basic protein (MBP) and GST-π+ immunoreactivity. MBP expression is frequently used for immunohistochemical analysis of demyelination and remyelination in EAE [40‐43] while GST-π is a marker for mature oligodendrocytes [44‐49]. Consistent with the LFB staining, MBP expression and GST-π+ immunoreactivity within the VLWM was comparable among all treatment groups. Of note, MBP staining exhibited more heterogeneity within the spinal cords of mice with EAE that did not receive higher doses of CXCR7 antagonist. These findings suggest that CXCR7 antagonism preserves axons independent of myelin recovery.

To assess the phenotype of immune cells that persist within the CNS parenchyma during recovery, spinal cord tissues from our five treatment groups and naïve controls were evaluated for meningeal and parenchymal CD3+ cells in conjunction with GFAP expression, the latter to delineate meningeal and parenchymal borders. All treatment groups exhibited meningeal CD3+ cells, which were increased in numbers in vehicle- and saline-treated mice compared with mice treated with 5 mg/kg or 10 mg/kg CCX771 (Figure 6A). Parenchymal infiltration of CD3+ cells was also increased in vehicle- and saline-treated mice compared with mice treated with CCX771, which exhibited few parenchymal CD3+ cells (Figure 6A). Quantitative analysis of meningeal and parenchymal CD3+ staining within both the parenchyma and meninges of mice in each treatment group revealed significant differences between the two CNS regions in mice treated with low dose CCX771 (5 mg/kg) and in vehicle- or saline-treated controls but not in mice treated with high dose CCX771 (10 mg/kg) (Figure 6B, Two-way ANOVA interaction: F = 2.49, DFn = 5, DFd = 118, P = 0.0.347; antagonist treatment: F = 13.52, DFn = 5, DFd = 118, P < 0.0001; area of CNS (parenchyma or meninges): F = 52.86 DFn = 1 DFd = 118, P < 0.0001). Comparison of CD3 staining within the spinal cord parenchyma, but not the meninges, across treatment groups revealed highly significant differences between all treatment groups and vehicle-treated controls (Table 3 One-way ANOVA parenchyma: F = 10.56, p < 0.0001). These data indicate that CXCR7 antagonism prevents persistent T cell infiltration within the spinal cord parenchyma, which could potentially contribute to ongoing axonal damage.

In order to assess the contents of spared spinal cord VLWM in treatment groups after recovery from EAE, spinal cords sections encompassing L2/3, which were the same segments used for LFB staining, were examined for the extent of neurofilament H (NF-H) dephosphorylation via anti-SMI-32 immunofluorescence, as previously reported [16, 17, 50, 51]. SMI-32 immunopositivity was found predominantly in neuronal cell bodies with little detectable staining in the white matter in both groups of 10 mg/kg CCX771-treated mice and in naïve animals (Figure 7A, upper panels). A similar pattern was observed in the 5 mg/kg CCX771-treated mice group (Figure 6A, left lower panel). Extensive SMI-32+ immunoreactivity, however, was detected in the spinal cord VLWMs of either saline- or vehicle-treated mice, even in areas where myelin was present (Figure 5, middle and right lower panels), consistent with DTI findings of axonal damage (Figure 7A, middle and right lower panels). Further statistical analyses of SMI-32+ pixels/area within dorsal, lateral and ventral spinal cord white matter of mice in each treatment group revealed significant decreases between CCX771-treated and naïve versus control animals only within lateral and ventral regions (Figure 7B, 2-way ANOVA interaction: F = 0.37, DFn = 10, DFd = 71, P = 0.9568; antagonist treatment: F = 3.25, DFn = 5, DFd = 71, P = 0.0107; white matter area: F = 1.46 DFn = 2, DFd = 71, P = 0.2388). These data support the DTI findings and that CXCR7 antagonism in the setting of EAE prevents axonal damage in a dose dependent fashion.