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

Abstract

Background: Multiple Sclerosis (MS) is characterized by the pathological trafficking of leukocytes into the central nervous system (CNS). Using the murine MS model, experimental autoimmune encephalomyelitis (EAE), we previously demonstrated that antagonism of the chemokine receptor CXCR7 blocks endothelial cell sequestration of CXCL12, thereby enhancing the abluminal localization of CXCR4-expressing leukocytes. CXCR7 antagonism led to decreased parenchymal entry of leukocytes and amelioration of ongoing disease during EAE. Of note, animals that received high doses of CXCR7 antagonist recovered to baseline function, as assessed by standard clinical scoring. Because functional recovery reflects axonal integrity, we utilized diffusion tensor imaging (DTI) to evaluate axonal injury in CXCR7 antagonist- versus vehicle-treated mice after recovery from EAE.

Methods: C57BL6/J mice underwent adoptive transfer of MOG-reactive Th1 cells and were treated daily with either CXCR7 antagonist or vehicle for 28 days; and then evaluated by DTI to assess for axonal injury. After imaging, spinal cords underwent histological analysis of myelin and oligodendrocytes via staining with luxol fast blue (LFB), and immunofluorescence for myelin basic protein (MBP) and glutathione S-transferase-π (GST-π). Detection of non-phosphorylated neurofilament H (NH-F) was also performed to detect injured axons. Statistical analysis for EAE scores, DTI parameters and non-phosphorylated NH-F immunofluorescence were done by ANOVA followed by Bonferroni post-hoc test. For all statistical analysis a p < 0.05 was considered significant.

Results: In vivo DTI maps of spinal cord ventrolateral white matter (VLWM) axial diffusivities of naïve and CXCR7 antagonist-treated mice were indistinguishable, while vehicle-treated animals exhibited decreased axial diffusivities. Quantitative differences in injured axons, as assessed via detection of non-phosphorylated NH-F, were consistent with axial diffusivity measurements. Overall, qualitative myelin content and presence of oligodendrocytes were similar in all treatment groups, as expected by their radial diffusivity values. Quantitative assessment of persistent inflammatory infiltrates revealed significant decreases within the parenchyma of CXCR7 antagonist-treated mice versus controls.

Conclusions: These data suggest that CXCR7 antagonism not only prevents persistent inflammation but also preserves axonal integrity. Thus, targeting CXCR7 modifies both disease severity and recovery during EAE, suggesting a role for this molecule in both phases of disease.

Figure 1

CXCR7 antagonism ameliorates the clinical severity of EAE. Dose response effects of the CXCR7 antagonist, CCX771, were evaluated during induction of and after treatment of ongoing adoptive transfer EAE. (A) Animals were grouped into those receiving daily treatment with saline (yellow), vehicle (red) or CCX771 at 5 mg/kg (green) or 10 mg/kg (blue) beginning at the time of adoptive transfer or when animals reached a score of 1 (purple). Results are expressed as mean clinical score ± SEM, n = 5. Two-way ANOVA for all treatment groups showed a strong interaction of treatment and days post-adoptive transfer with disease progression: interaction: F = 2.201, P < 0.0001; antagonist treatment: F = 93.85, P < 0.0001; days post-adoptive transfer: F = 59.06, P < 0.0001. (B) Comparison between the mean highest severity scores of effector phase (red bracket on A) and recovery phase (blue bracket on A). Results are expressed as mean highest severity score ± SEM. Two-way ANOVA P values summary: interaction: F = 1.71, P = 0.1659; disease phase: F = 164.57, P < 0.0001; treatment: F = 22.21, P < 0.0001).

Figure 2

DTI analysis shows changes in ventral white matter. At the end of clinical assessment mice from all treatment groups and naïve littermates underwent in vivo DTI analysis. Spinal cord level was localized by axial scout images followed by multiple transverse slices (red arrows) to include the whole lumbar enlargement (A, slice thickness = 1.0 mm, field of view = 1 cm × 1 cm). Diffusion-sensitizing gradients were applied in six orientations: (Gx, Gy, Gz) = (1, 1, 0), (1, 0, 1), (0, 1, 1), (-1, 1, 0), (0, -1,1), and (1, 0, -1) with a gradient strength = 9 G/cm, duration (δ) = 7 ms, and separation (Δ) = 18 ms, to obtain b values of 0 and 0.750 s/mm². Regions of interest (ROIs) encompassing the ventrolateral white matter (VLWM) was drawn manually on the DTI parameter maps (B). The boundary between white matter and gray matter was identified on relative anisotropy (RA) maps. The clear gray-white matter contrast was seen in RA maps of all study groups. Radial (λ⊥) and axial (λ||) diffusivities showed heterogeneous abnormalities within the VLWM, being more severe in saline- and vehicle-treated groups. Analysis of radial diffusivity and relative anisotropy do not show differences between treatment groups, suggesting no differences in myelin integrity (C and E, One-way ANOVA F = 3.232, P = 0.0227 and F = 5.272, P = 0.0021, respectively). Meanwhile, analysis of axial diffusivity shows a similarity between 10 mg/kg CCX771-treated mice with naïve (D, One-way ANOVA, F = 3.232, P = 0.0227). Results are expressed as mean of λ⊥, λ|| or RA ± SD).

Figure 3

Axial diffusivity threshold segmentation on VLWM of mouse spinal cord reveals similarities of CCX771-treated mice with naïve. VLWM λ|| distribution from naïve spinal cord (μ = 1.7, σ = 0.36, n = 5) (A). Red line represents threshold of injured white matter from naïve spinal cords used as control. The background image is the λ|| maps (B). The red masks represent axial diffusivity threshold defined injured axon (or white matter). Saline- and vehicle-treated mice showed extensive VLWM having abnormal axial diffusivity. Statistical analysis confirmed no significant difference between naïve and 10 mg/kg CCX771-treated mice in contrast to the other groups (C, results are expressed as mean ± SD. One-way ANOVA, F = 7.855, P = 0.0002).

Figure 4

Axial diffusivity defined abnormal VLWM volume correlates with clinical severity. Linear regression analysis of mean clinical score during recovery phase as a function of axial diffusivity (A) and injured VLWM (B) (A, R² = 0.5413, F = 33.04, DFn, DFd = 1.0, 28.0, P < 0.0001; B, R² = 0.6520, F = 43.10, DFn, DFd = 1.0, 23.0, P < 0.0001). Area-under-curve (AUC) as a function of injured VLWM showed an R² value of 0.7284 (C, F = 61.69, DFn, DFd = 1.0, 23.0, P < 0.0001).

Figure 5

Myelin staining is consistent with DTI findings. Lumbar segments sections of 6 μm were stained with LFB and exhibited comparable levels of myelination around ventral horns when observed at 5X of magnification (A scale bar = 50 μm). Persistent inflammation at the meningeal surface was detected in 5 mg/kg CCX771-, vehicle- and saline-treated mice. At 40X magnification (offsets, scale bar = 20 μm) of the same lumbar segment shown similar myelin distribution for 10 mg/kg CCX771-treated mice than naïve. At 40X magnification (offsets, scale bar = 25 μm) normal axons could be identified (arrow) in mice treated with 10 mg/kg CCX771 while axonal bundles in different stages of degeneration were observable to a lesser extent in 5 mg/kg CCX771-treated mice but, more abundant and severe in vehicle- and saline-treated mice (arrowheads). Immunofluorescent detection of MBP (green) and GST-π (red) showed similar levels of myelin content and presence of oligodendrocytes in all treatment groups (B scale bar = 25 μm). As observed with LFB, DAPI nuclear staining (blue) is more evident in the 5 mg/kg CCX771-, vehicle- and saline-treated mice.

Figure 6

Injured VLWM shows evidence of persistent inflammation. Double immunofluorescence analysis of infiltrated CD3+ T-cells (green) using GFAP (red) antibody to delineate CNS parenchyma within spinal cords of mice evaluated by DTI (A). T cells were absent on spinal cords from naïve mice while some CD3+ lymphocytes were detected on CCX771-treated mice. However, these CD3+ T-cells were mostly restricted to meningeal spaces and within GFAP+ parenchyma. Vehicle- and saline-treated mice showed a higher infiltration of CD3+ lymphocytes within GFAP+ parenchyma. As observed with LFB, To-Pro-3 nuclear staining (blue) is more evident in the 5 mg/kg CCX771-, vehicle- and saline-treated mice. (A 63X magnification, scale bar = 25 μm). Statistical analysis shows significant differences in CD3+ pixel ratios within ROI's (parenchyma and meninges) among treatment groups (B results are expressed as mean CD3+ pixels per ROI ± SEM. Two-way ANOVA P values summary: interaction F = 2.49, P = 0.0347; antagonist treatment F = 13.52, P < 0.0001; area of CNS (parenchyma or meninges) F = 52.86, P < 0.0001).

Figure 7

Injured VLWM shows abnormal dephosphorylation of Neurofilament H. Immunofluorescence analysis of NF-H was assessed using SMI-32 antibody immunoreactivity to detect dephosphorylation within spinal cords after DTI evaluation (A). Higher SMI-32 staining within VLWM was observed in saline- and vehicle-treated mice in contrast to naïve controls, while CCX771-treated mice were more comparable to control mice (63X magnification, scale bar = 25 μm). Statistical analysis shows significant differences within lateral and ventral portions of the spinal cord white matter area (B results are expressed as mean SMI-32+ pixels per area ± SEM. Two-way ANOVA P values summary: interaction F = 0.3671, P = 0.9568; antagonist treatment F = 3.248, P = 0.0107; area F = 1.4661, P = 0.2382.).