144 Chapter 6 A B C D E F G Figure 1. Spinal cord dissection and immunohistochemistry. (A) The distance between thoracic level 2 and lumbar level 5 was recorded after identifying all available spinal cord nerve roots, and axial blocks were dissected at each level. (B, C) Ten-micrometer-thick sections were immunostained for myelin basic protein on all available control (B) and multiple sclerosis sections (C). (D, E) Sequential sections were also stained with hematoxylin and eosin (D) and viewed at ×20 magnification (E) to count the number of neurons in each anterior horn. (G, H) Spinal cord sections were also immunostained for synaptophysin. Scale bar for spinal cord dissection (A), 1cm; scale bars for low-magnification images (B–D, F), 1mm; scale bars for high-magnification images (E, G), 50μm. defined as the transmitted light divided by the incident light, synaptophysin images were opened using ONCOtopix software (Visiopharm, Hørsholm, Denmark) and measured in each anterior horn using the “intensity” tool. Light intensity is based on transformation of images into gray scale ranging from 0 to 255. Given that 0 represents absolute black and 255 represents white, low-intensity values represent the most intense staining. Thus, to render readings more intuitive (higher values meaning higher synaptic density) the results are reported as 1/T. To assess the synaptic bouton area, images of synaptophysin and synapsin-1 staining were opened using NDP.view2 (U12388-01NDP.view2, Hamamatsu Photonics) viewing software and inspected by zooming in from objective ×1.25 to ×80 magnification. Given the size of motor neurons, a ×20 objective was most suitable for assessment of synaptic bouton area. Using the “freehand region” selection tool, the total synaptic bouton area was outlined on a mean of 5 neurons (area size: at least 200μm2) per anterior horn (see Fig 2). Statistics Linear mixed models, with random subject intercepts to allow for within-subject dependency, were used both to compare measures between slice types and to examine association between measures. Where residuals suggested the possibility of heteroscedasticity or nonnormality, robust standard errors or bootstrap was used to confirm or correct confidence intervals and p values.
RkJQdWJsaXNoZXIy MTk4NDMw