(A) Schematic illustrating intermittent myelination of cortical axons, designating each internode by the number of flanking myelin sheaths (0, yellow; 1, magenta; or 2, cyan). (B) Example maximum intensity projections of lost (magenta) and new (green) myelin sheaths from cuprizone-treated mice. Top row, example of a lost, but not replaced, sheath, and a novel sheath not present at baseline (green arrowhead; also see Video 8). Remaining rows are examples of lost (magenta sheaths and arrowheads) and replaced internodes (green sheaths and arrowheads) that had 0 (see also Video 9), 1, or two neighboring sheaths at baseline. Yellow arrowheads in the two neighbor example indicate nodes of Ranvier. (C) 2D rendering of traced myelin sheaths that passed through a 100 μm x 100 μm x 100 μm volume within the top 0–100 μm zone. The fate of each sheath within the volume was determined as in Figure 5. Sheath color illustrates whether it had 0, 1 or two neighbors as in A. (D, G) Comparison of mean proportion of internodes with at least one neighbor (lavender), isolated (yellow), or undefined (gray) within a 100 μm x 100 μm x 100 μm volume at baseline and 8 weeks later, from control (D, N = 5) and cuprizone-treated (G, N = 5) mice. Volumes from both control and cuprizone-treated conditions have the same relative proportion of isolated vs. ≥1 neighboring internode at both time-points. (E, H) Comparison of the mean proportion of internodes with 0 or ≥1 neighbor that are stable or novel (control, (E) or replaced, not replaced, or novel (cuprizone, (H). There is no significant difference in the proportion of isolated vs. ≥1 neighbor population between stable and novel sheaths in control (E), and between lost and novel sheaths in cuprizone (H), but relatively more internodes with ≥1 neighbor were replaced in cuprizone-treated cortex (H). F, I: Myelination matrix illustrating average number of internodes categorized by number of neighbors at baseline and at final imaging time-point for control (F), and cuprizone-treated mice (I). Internodes with more neighbors at baseline are more likely be replaced (largest average # of internodes in bottom right of matrix). See Supplementary file 1 for statistical tests and significance levels for each comparison.