Our postmortem tissue analysis suggests that BSCB disruption in ALS patients leads to extravasation of erythrocytes in the spinal cord and subsequent accumulation of erythrocyte-derived hemoglobin and iron-containing hemosiderin, as well as extravasation of multiple plasma-derived proteins. We also show that BSCB breakdown in ALS subjects is associated with pericyte loss in motor neuron dense regions of the spinal cord, i.e., the cervical spinal cord anterior horn gray matter. The present study further supports the existence of alterations of the BSCB in ALS subjects. Past studies utilizing both CSF and tissue analyses have suggested possible BBB and/or BSCB disruption in a subset of human ALS subjects varying from 26 to 100 % of cases depending on both the study and the parameter being analyzed (summarized in Table ).
Prior studies suggestive of vascular disruption in sporadic and familial amyotrophic lateral sclerosis
At a molecular level, transcriptional analysis has demonstrated that the tight junction proteins of the BSCB, i.e., zonula occludens-1 (ZO-1) and occludin, are reduced in sporadic and familial ALS cases in the lumbar spinal cord suggesting a potential mechanism for barrier disruption [31
]. Elevations in matrix metalloproteinase-9 (MMP-9), an enzyme known to chronically degrade endothelial tight junctions [7
], have also been detected in ALS serum samples [8
] and postmortem brain and spinal cord specimens [41
]. MMP-9 has been linked to degradation of the tight junction proteins and extracellular basement membrane matrix proteins at the BSCB of ALS patients by an independent study [46
]. Our findings suggest that abnormalities may not be restricted to endothelial cells and that reductions in spinal cord capillary pericytes may also contribute to microvascular disruption in human subjects. Importantly, pericytes have been shown to promote endothelial tight junction protein expression, including ZO-1 and occludin [6
], facilitate tight junctional alignment [14
], and reduce endothelial vesicular uptake and transcytosis of circulating macromolecules [4
]. Pericytes under pathologic conditions have been recently demonstrated to be an important source of secreted MMP-9 leading to degradation of both endothelial tight junctions and the basement membrane resulting in vascular fragility [7
]. Whether a similar relationship exists in human ALS remains to be determined and should be addressed.
Unlike other neurodegenerative disorders with chronic CNS microhemorrhage, such as Alzheimer’s disease [13
], the spinal cord erythrocyte extravasation described in ALS patients in this study was predominately confined to the capillary level. Following extravasation, erythrocytes lyse and liberate both membranous and cytoplasmic components including free hemoglobin [33
]. Following lysis of extravasated erythrocytes, degradation of hemoglobin-derived heme within the CNS gives rise to carbon monoxide, bilirubin and iron [33
]. This results in local elevations of both free and protein-bound iron including the insoluble-iron storage protein hemosiderin [33
]. Hemosiderin deposits found in the present study were much smaller in size (~10–50 μm) than hemorrhages described at the arterial or arteriolar level and/or in Alzheimer’s disease (>100 μm) [13
]. Recent studies have demonstrated that circulating plasma penetrates five times faster than the erythrocyte core [56
] and that the plasma protein diameter is typically 3 to 3.5-fold larger than the erythrocyte core in CNS parenchyma following vascular disruption [39
]. Similarly, we found that the diameter of distribution of extravasated plasma proteins, such as fibrin, was on average 3 to 4-fold larger than that of hemoglobin. This likely reflects a more centrally localized erythrocyte core surrounded by peripheral plasma protein diffusion in human ALS spinal cord tissue.
Hemoglobin is toxic to motor neurons [54
] and neural cells expressing different SOD1 mutants [67
] in vitro through iron-dependent oxidative injury. In mouse models of ALS [46
] and in pericyte-deficient mice with disrupted PDGFRβ signaling in pericytes [63
], BSCB disruption with erythrocyte extravasation and/or motor neuron accumulation of extravasated plasma-proteins such as thrombin and fibrin precedes motor neuron injury. Whether small foci of extravasated erythrocytes can contribute to motor neuron injury in the spinal cord of ALS patients, and whether extravasated proteins (e.g., fibrin, thrombin) localize on the surface or inside motor neurons in human ALS, as found in cortical and/or motor neurons in pericyte-deficient mice [4
] or transgenic apolipoprotein E4-expressing mice [7
], require future experimental investigation.
Perivascular hemosiderin deposits account for perivascular hypointensities of T2*-weighted MRI analysis utilized clinically for the detection of intracerebral microhemorrhage [27
]. The small size of spinal cord hemosiderin deposits as described in the present study may pose a significant barrier to the sensitivity and specificity of conventional T2*-weighted MRI imaging pulse sequences in human subjects [27
]. Therefore, MRI studies in ALS patients should be interpreted with caution [68
]. For example, one brain MRI study failed to detect microhemorrhages that were <100 μm in diameter in a small number of ALS patients [59
], possibly due to limited resolution of imaging [27
]. Future studies are needed combining MRI and histopathologic analyses to better optimize pulse sequences for the detection of perivascular hemosiderin deposits in ALS patients, especially in the spinal cord, which has yet to be investigated. Whether ALS patients develop BBB disruption with erythrocyte extravasation, hemosiderin deposition, and/or pericyte reductions in motor cortex and affected brainstem motor nuclei remains to be seen.
Greater than 99 % of circulating iron is bound to hemoglobin [23
] and, therefore likely contributes to previously reported elevations in CNS iron in human ALS [35
]. In addition, we demonstrate accumulation of plasma-derived proteins including IgG, fibrin and thrombin. At present, it is unclear as to whether the presence of vascular disruption alters the clinical course of ALS cases. A possible limitation of any study of human spinal cord tissues is the post-mortem sampling, with results reflecting an end-stage process. Thus, experimental models are needed to better characterize ALS-associated vascular dysfunction and determine whether exacerbating or mitigating the BSCB breakdown and erythrocyte extravasation in accepted experimental models of ALS contributes to motor neuron injury and disease progression during ALS pathogenesis. Sensitive biomarkers should be investigated that detect ALS pericyte reductions and capillary leakage present in spinal cord tissues. Larger studies and continued development of technologies to detect spinal cord microvascular changes hold considerable promise in deducing whether vascular disruption may represent an important variant of ALS and guide therapeutic development accordingly.