In the present study, we investigated the functional competence of the BSCB in G93A SOD1 mice modeling ALS at different stages of disease. We observed microscopic evidence of BSCB impairment in cervical and lumbar spinal cords, areas of motor neuron degeneration, of ALS mice at initial disease symptoms and, more severely, at late stage disease. Our data show EB leakage in cervical/lumbar spinal cord microvessels in G93A mice at early (13 weeks of age) and late (17–18 weeks of age) stage disease. More leakage was found in lumbar spinal cords of mice at terminal stage disease. Additionally, basement membrane disruption was noted at both early and late stage disease, as shown by the loss of laminin staining in the G93A mice. Downregulation of Glut-1 and CD146 expressions in spinal cord endothelial cells was also found in G93A mice at early and late stage disease and may relate to altered endothelial lining leading to vascular leakage. Small numbers of delineated astrocytes were also established. These results confirm our previous ultrastructural findings 
on disruption of the BSCB showing functional incompetence of BSCB structural elements in ALS mice.
Significant death of motor neurons in G93A mice occurs at the onset of clinical disease (90 days) and by end-stage disease (136 days), mice show up to a 50% loss of cervical and lumbar motor neurons 
. In G93A mice, motor deficits have been observed in tests of muscle strength and coordination as early as 8 weeks of age 
. These results extend those of our previous studies 
, showing initial signs of disease, such as tremor, weight loss, and reduced hindlimb extension, in G93A mice at about 13 weeks (90 days) of age. At this age, numerous vacuolizated motor neurons were found in the cervical lumbar spinal cord and most motor neurons in the lumbar spinal cord showed signs of degeneration.
The primary BBB/BSCB function is control of the CNS homeostasis by selective transport of molecules and cells from the systemic compartment. Substances with a molecular weight higher than 400 Da generally cannot cross the barriers by free diffusion. However, certain endogenous large molecules, such as insulin, leptin, transferrin, and insulin-like growth factors, enter the brain from blood via specific endothelial carrier-mediated or receptor-mediated transporters (reviewed in 
). Recently, IgG was detected in the perikarya of motor neurons of the lumbar spinal cord in mice 24 hours after intraperitoneal injection of IgG derived from sera of ALS patients 
. The injected IgG was found in the axon terminals of the lumbar ventral horn motor neurons, localizing in the microtubules and rough endoplasmic reticulum. Furthermore, IgG was similarly detected in spinal cord motor neurons of ALS patients. There was also evidence of IgG intake in endothelial cells in affected areas of the spinal cord in both ALS patients and mice injected with human ALS IgG. Engelhardt et al. 
suggest that “there may be multiple antibodies targeting a variety of epitopes of motor neurons in ALS”. The molecular weight of IgG is 150,000 Da and it is unlikely that these molecules could cross an intact brain capillary endothelium even by receptor-mediated transcytosis. However, Pirttila et al. 
showed that insulin-like growth factor (IGF)-1, IGF binding protein-2, or nitric oxide were not elevated in CSF of ALS patients, suggesting that there is not a major disruption in the BCSFB. In another report 
, an ALS mouse model with a permissive BBB was created by crossing G93A mouse with the mdr1a/b
knockout mouse and showed that cyclosporine A (CsA), which cannot cross an intact BBB, BSCB or BCSFB, reached the CNS when delivered intraperitoneally into this combined mouse model. Since the authors did not investigate BBB or BSCB condition in the original transgenic G93A mice, it is possible that disruption or dysfunction of these barriers occur in ALS. The 1200 Da molecular weight of CsA is much smaller than that of IgG. Further investigation is needed to resolve this apparent discrepancy in ability to cross the BBB/BSCB.
Our finding of Evans blue extravasion in early symptomatic G93A mice may suggest that large molecules such as IgG and other blood proteins appear in the spinal cord due to vascular leakage, one possible mechanism accelerating motor neuron damage. However, it is unclear if BBB/BSCB disruption appears prior to motor neuron degeneration or as result of motor neuron dysfunction. Also, differences between the BBB and BSCB in endothelial protein concentrations may impact observed pathological changes in G93A mice. It has been shown that microvascular endothelial cells, isolated from murine spinal cord, morphologically similar to BBB endothelial cells, express reduced amounts of several prominent BBB proteins such as tight junction-associated proteins ZO-1 and occluding, adherens junction-associated proteins beta-catenin and VE-cadherin, and the efflux transporter P-glycoprotein 
Reduction in immunofluorescent labeling of basement membrane of affected G93A mice suggests possible membrane disruption. The basement membrane is part of the extracellular matrix and is composed of collagens, proteoglycans, elastin and several glycoproteins, of which laminin is the most abundant 
. Reduced laminin labeling was observed in cervical and lumbar spinal cords of both early and late symptomatic G93A mice, possibly indicating vascularization changes leading to capillary wall permeability. Interestingly, Ono et al. 
showed fragmented and widely separated collagen bundles in capillaries and decreased amounts of collagen in postmortem posterior half of the lateral funiculus and in the anterior horn of cervical enlargements from patients with sporadic ALS. Although the role of these aberrations in the pathogenesis of ALS remains to be determined, the authors suggested that abnormalities of collagen in the perivascular spaces of capillaries “may be secondary to neuronal degeneration as a underlying mechanism in ALS”.
It is well known that glucose transport through the BBB (BSCB) is mediated by glucose transporter isoform 1 (Glut-1) 
. Glut-1 is associated mainly with the brain capillary endothelial cells and is asymmetrically distributed between the luminal and abluminal membranes 
. This asymmetric intracellular pool of glucose transporter may provide for rapid transport of glucose across the abluminal plasmalemma to the brain parenchyma 
. The alteration of Glut-1 may be related to the pathogenesis of microvascular permeability as has been shown, for example, in cerebral edema (reviewed in 
). In the present study, we found low and mostly absent expression of Glut-1 in capillaries of both cervical and lumbar spinal cords of G93A mice at early and late stages of disease. This downregulation of Glut-1 expression in the endothelial cells of the BSCB may be related to altered endothelial lining leading to vascular leakage. Alternatively, decreased Glut-1 expression may result from aggravated alterations of the BSCB in G93A mice. Although additional experiments such as quantitative analysis of Glut-1 distribution and density in the endothelial plasma membranes are needed to elucidate the regulatory mechanisms of Glut-1 expression in the spinal cord, the present study indicates that alteration of Glut-1 could be involved in the pathogenesis of ALS.
Another of our findings was that endothelia surrounding capillaries were partially revealed by CD146 antigen expression in the cervical and lumbar spinal cords of G93A mice at initial and, more markedly, at late stages of disease. Moreover, small numbers of delineated astrocytes were established. These results may indicate that degeneration or, at least, partial dysfunction, of non-neuronal cells in ALS occurs. Evidence of widespread inflammatory reactions in ALS already exists. The presence of monocyte/macrophage cells, activated microglia, and reactive astrocytes was established in the spinal cord tissue of most ALS patients 
. In a mouse model of ALS, immune/inflammatory responses 
are present even before any evidence of motor dysfunction 
. Strategically, astrocytes are located at the interface between the blood vessels and the brain as well as in the spinal cord parenchyma, influencing both the entry of blood cells into the CNS and the activity of invading cells once they have entered the brain parenchyma (reviewed in 
). It has been shown that activated astrocytes and microglia “in response to signals derived from the immune system or generated within the CNS” produce various inflammatory molecules that may increase the permeability of the endothelial cell barrier 
. Inhibition of microglia activation, for example, as recently shown in vitro
and in vivo
using minocycline, may protect the brain after ischemic stroke by improving BBB viability and integrity 
. It is possible that glial cell activation in ALS could lead to vessel leakage. Additionally, decreased numbers of delineated astrocytes and their perivascular end-feet at the blood capillaries could affect vessel permeability.
Thus, our results confirm our previous ultrastructural findings on disruption of the BSCB showing functional incompetence of BSCB structural elements in ALS mice. A breakdown in the BSCB is clearly indicated by EB leakage in cervical/lumbar spinal cord microvessels in G93A mice at early and late stages of disease. Laminin labeling suggests that basement membrane of vessels in the spinal cords of the diseased G93A mice may be affected. Additionally, downregulation of Glut-1 and CD146 expressions in the endothelial cells of the BSCB may be related to altered endothelial lining leading to vascular leakage. Degeneration of astrocytes could influence BSCB integrity. Importantly, is BSCB breakdown a primary or secondary mechanism to motor neuron degeneration in G93A mice? Demonstrating BSCB disruption prior to the onset of disease symptoms and other pathological processes would indicate that BSCB disruption plays a primary role in ALS pathogenesis.