EAE progression is associated with upregulated expression of fibronectin and α5 integrin on spinal cord blood vessels
In a previous study we demonstrated that blood vessels in the brain and cervical spinal cord of mice with EAE show upregulated expression of fibronectin and α5 integrin [3]. As the earliest and most severe pathology in the EAE model occurs in the lumbar part of the spinal cord, in the current study we first wanted to determine whether this region of the spinal cord shows similar changes in vascular fibronectin and α5 integrin expression during EAE pathogenesis. To study this process, EAE was induced in 10 week old female wild-type (WT) C57BL6/J mice by immunization with MOG35–55 peptide, a widely-accepted model of chronic progressive MS, as previously described [3]. In keeping with findings from our lab and others [3, 7, 27], WT mice began developing clinical signs 9–12 days post-immunization (tail paralysis followed by hindlimb weakness and paralysis, and eventually quadriplegic) and disease severity gradually worsened with time (Fig. 1a). Clinical severity peaked between 15 and 21 days post-immunization and improved slightly thereafter, but mice never completely recovered. To examine how vascular expression of fibronectin and α5 integrin changes in the lumbar spinal cord during the course of EAE in WT mice, we used the endothelial marker CD31 and performed CD31/fibronectin or CD31/α5 integrin dual-immunofluorescence (IF) staining on frozen sections of lumbar spinal cord at 0, 7, and 16 days post-immunization, corresponding to disease-free control, pre-symptomatic and peak symptomatic disease, respectively. As shown in Fig. 1, under disease-free control conditions, fibronectin (Fig. 1d) and α5 integrin (Fig. 1e) were expressed at only low levels by lumbar spinal cord blood vessels, but as EAE developed, vascular expression levels of both proteins increased such that by the peak stage of EAE, fibronectin and α5 integrin were expressed at much higher levels. Quantification of fluorescent intensity (Fig. 1b) revealed that compared to disease-free conditions, vascular fibronectin expression was significantly upregulated at the pre-symptomatic stage of disease (5.75 ± 0.72 compared to 1.56 ± 0.07 fluorescent units per FOV under disease-free conditions, p < 0.05), and this expression level was further increased at the peak symptomatic stage of disease (8.50 ± 1.36 compared to 1.56 ± 0.07 fluorescent units per FOV under disease-free conditions, p < 0.05). In parallel with this upregulation of fibronectin, significant endothelial upregulation of the fibronectin receptor α5β1 integrin was detected at the pre-symptomatic stage of disease (4.56 ± 0.79 compared to 1.61 ± 0.53 fluorescent units per FOV under disease-free conditions, p < 0.05), and this enhanced expression of the α5 integrin subunit was maintained at the peak symptomatic stage of disease (3.84 ± 0.45 compared to 1.61 ± 0.53 fluorescent units per FOV under disease-free conditions, p < 0.05) (Fig. 1c).
Genetic deletion of endothelial α5 integrin results in early onset EAE, correlating with worse neuroinflammation
To investigate the role of endothelial α5β1 integrin in modulating EAE pathogenesis, we used a Cre-Lox approach to generate mice lacking α5 integrin in endothelial cells (α5-EC-KO), by crossing floxed α5 integrin mice [37] with Tie2-Cre transgenic mice [20], as previously described [24]. Transgenic mice expressing Cre recombinase under the control of the Tie2 promoter, (Tie2-Cre; α5f/+) were crossed with mice in which the α5 integrin gene was floxed, i.e.; flanked by LoxP sites (α5f/f). From this breeding strategy, approximately 25% of the offspring were Tie2-Cre, α5f/f which lacked α5 integrin expression in endothelial cells (referred to as α5-EC-KO mice). Littermate mice that had two copies of the floxed α5 integrin gene but lacking the Tie2-Cre transgene (α5f/f; Tie2-Cre negative) were used as wild-type (WT) controls. Importantly, α5-EC-KO mice are viable and fertile and show no obvious defects in developmental angiogenesis or vascular function under disease-free control conditions in the adult, and thus are amenable to experimental analysis [37]. To confirm that this genetic approach was effective at deleting α5 integrin from endothelial cells in these studies, we examined α5 integrin expression in sections of spinal cord taken from mice either under disease-free control conditions or at the pre-symptomatic stage of EAE. As shown in Fig. 2a, spinal cord blood vessels in WT mice maintained under disease-free conditions showed low levels of α5 integrin expression, but this expression was markedly increased at the peak of EAE disease. In contrast, α5 integrin was undetectable on spinal cord blood vessels in α5-EC-KO mice under any condition. This demonstrates that the α5 integrin gene was totally deleted from spinal cord endothelial cells within α5-EC-KO mice and it also demonstrates that endothelial cells are the major cell type expressing α5 integrin in spinal cord blood vessels [26, 28].
To investigate how genetic deletion of endothelial α5 integrin impacts the clinical progression of EAE, disease was established in 10 week old female α5-EC-KO and WT littermate mice and disease progression compared (Fig. 2b). This showed that α5-EC-KO mice developed much earlier clinical onset of EAE relative to WT littermates (mean time of onset 8.39 ± 0.86 days post-immunization vs 13.72 ± 2.51 days for WT littermates, p < 0.05). The mean time to reach peak disease was also much shorter in α5-EC-KO mice (11.67 ± 1.09 days vs 16.94 ± 3.00 days for WT littermates, p < 0.05). This point is well illustrated in Fig. 2b which shows that in keeping with other studies, the peak clinical score of the entire WT group was reached after approximately 20 days, but in contrast, the α5-EC-KO group reached peak clinical score after just 12 days. Thus, EAE onset and progression is significantly accelerated in α5-EC-KO mice. Interestingly however, despite these differences, by day 20 the clinical scores of α5-EC-KO and WT littermate mice were largely equivalent and remained that way until the end of the experiment (day 30). This data demonstrates that lack of endothelial α5 integrin predisposes to earlier onset and accelerated progression of EAE but has no significant impact on peak disease severity or chronic disease activity.
To investigate how lack of endothelial α5 integrin impacts neuroinflammation and demyelination in this EAE model, we performed fluoromyelin/CD45 dual-IF on frozen sections of lumbar spinal cord. As shown in Fig. 3a-b, CD45 staining at the pre-symptomatic phase of EAE (7 days post-immunization) revealed that compared to WT controls, the lumbar spinal cord of α5-EC-KO mice contained significantly higher levels of CD45+ inflammatory leukocytes (4.84 ± 1.59 vs. 0.44 ± 0.05 fluorescent units per FOV, p < 0.05). At the same time-point (7 days), Mac-1 IF revealed increased infiltration of monocytes and activation of microglia in the spinal cord of α5-EC-KO mice as compared to WT littermates (13.12 ± 2.88 vs. 4.78 ± 0.18 fluorescent units per FOV, p < 0.05) (Fig. 3d-e). Fluoromyelin staining showed that demyelination was also more pronounced in α5-EC-KO mice relative to WT littermates at this time-point (4.62 ± 1.41 vs. 0.34 ± 0.11 fluorescent units per FOV, p < 0.05) and accumulation of CD45+ inflammatory leukocytes correlated strongly with erosion of myelin (see arrow in Fig. 3a). Interestingly however, while leukocyte infiltration and demyelination were much greater at the peak symptomatic stage of EAE (16 days post-immunization) compared to pre-symptomatic, levels between α5-EC-KO and WT littermates were not appreciably different. Thus, in this EAE model, absence of endothelial α5 integrin results in earlier onset of clinical disease, correlating with increased leukocyte infiltration and demyelination during the pre-symptomatic phase of disease, but by the symptomatic phase of EAE this difference had largely disappeared.
Spinal cord blood vessels in α5-EC-KO mice show enhanced vascular leak at an early stage of disease
As α5-EC-KO mice show earlier onset of EAE and increased levels of leukocyte infiltration and microglial/monocyte activation during the early pre-symptomatic stage of disease, we next examined whether the vascular integrity of spinal cord blood vessels was compromised in these mice. Using fibrinogen leak as a marker of vascular disruption, CD31/fibrinogen dual-IF showed that under disease-free conditions, there was no vascular leak in either WT or α5-EC-KO mice. However, during the pre-symptomatic (7 days post-immunization) phase of disease, while negligible extravascular leak of fibrinogen (Fbg) was detected in WT littermate control mice, α5-EC-KO mice showed obvious leak at this time-point (Fig. 4a). Quantification revealed that 7 days post-immunization, fibrinogen leak was significantly greater in α5-EC-KO mice compared to WT littermate controls (3.44 ± 0.94 compared to 0.78 ± 0.42 fluorescent units per FOV, p < 0.05), though interestingly, later during the peak symptomatic phase of disease (16 days post-immunization), vascular leak of fibrinogen in α5-EC-KO mice and WT littermate controls was largely equivalent (Fig. 4b).
The pre-symptomatic angiogenic response is markedly attenuated in α5-EC-KO mice
In a previous study, we showed that in the pre-symptomatic phase of EAE, CNS blood vessels launch a strong vascular remodeling response that involves active endothelial proliferation leading to increased vascularity [3]. In light of our finding that endothelial α5β1 integrin plays an important angiogenic role, driving endothelial proliferation during mild hypoxia [24], we next investigated whether lack of this integrin might be stunting vascular remodeling/repair during EAE progression. To examine this, we performed CD31/Ki67 dual-IF on frozen sections of lumbar spinal cord taken from mice at different stages of EAE. This showed that in the pre-symptomatic phase of EAE, the time window during which most endothelial proliferation occurs [3], compared to disease-free control conditions (negligible endothelial proliferation), spinal cords of WT littermate mice contained numerous dual-positive CD31+/Ki67+ cells per FOV (Fig. 5a and d). However, in α5-EC-KO mice, the number of dual-positive CD31+/Ki67+ cells in the spinal cords of pre-symptomatic mice was markedly reduced (12.75 ± 6.75 vs. 70.55 ± 24.10 CD31+/Ki67+ dual-positive cells/mm2 in WT littermate controls, p < 0.05) (Fig. 5a and b). Later, during the symptomatic phase of EAE, endothelial proliferation had fallen to much lower levels with no obvious difference between the WT and α5-EC-KO strains. In keeping with the strong angiogenic response during the pre-symptomatic phase of EAE, WT mice showed a significant increase in blood vessel density, both at the pre-symptomatic (601.65 ± 85.70 compared to 366.20 ± 55.56 CD31+ vessels/mm2 under disease-free conditions, p < 0.05) and symptomatic (656.56 ± 50.50 compared to 366.20 ± 55.56 CD31+ vessels/mm2 under disease-free conditions, p < 0.05) phases of disease. Significantly, the impaired endothelial proliferation response observed in α5-EC-KO mice during the pre-symptomatic phase resulted in marked reduction in spinal cord blood vessel density compared to WT littermates, both at the pre-symptomatic (408.90 ± 60.50 compared to 601.65 ± 85.70 CD31+ vessels/mm2 in WT littermate controls, p < 0.05) and symptomatic phases of EAE (525.25 ± 40.40 compared to 656.56 ± 50.50 CD31+ vessels/mm2 in WT littermate controls, p < 0.05).
Under pro-inflammatory conditions, α5 integrin null brain endothelial cells showed reduced proliferation
Following on from our observation that in EAE-affected mice, spinal cord blood vessels in α5-EC-KO mice contained fewer proliferating endothelial cells than WT mice, we wanted to test directly whether absence of α5 integrin impacts endothelial proliferation in a pro-inflammatory environment. To examine this, we isolated primary brain endothelial cells (BECs) from α5-EC-KO and WT littermate mice and cultured them on fibronectin for 24 h, at which point TNF-α was added to mimic inflammatory conditions and a BrdU incorporation assay was performed. As shown in Fig. 5e, TNF-α significantly enhanced the proliferation of WT BECs (36.8 ± 4.2 vs. 20.4 ± 2.3% under control conditions, p < 0.05) but α5 integrin null BECs showed much lower proliferation rates (8.8 ± 1.2 vs. 20.4 ± 2.3% for WT BECs under control conditions, p < 0.05) and were largely unresponsive to TNF-α (10.1 ± 2.5 vs. 8.8 ± 1.2% under control conditions, NS). These in vitro results support our in vivo observations and are consistent with the idea that endothelial α5β1 integrin confers vasculoprotection during EAE progression, in part by promoting endothelial proliferation and vascular repair.