BLKS FVB/N CBA DBA/2 NOD SJL 129S1 B6 NZW KK C3H A AKR BALB SWR B6 Supplemental figure 1. Body weight of mouse strains. N-sizes for number of mice are given in Fig 1. All values in supplemental figures are mean  SEM.

Supplemental figure 2. Strains with larger cortical areas have larger MCA artery trees. N-size for number of mice is the grand total of n-sizes given in Fig 1.

Supplemental figure 3. Genetic variation in collateral length among 8 strains examined. Top panel, For each strain, average length of collaterals interconnecting the MCA and ACA trees. Collateral length differed among the strains (p<0.001, ANOVA). See legend for Figure S5 for definition of collateral length and method of measurement. Middle panels, Strains with larger native (baseline) collateral number or diameter have longer collaterals. This correlation, although weak, suggests these anatomical “size” parameters may be affected by polymorphisms in a common genetic element(s) controlling collateral formation/growth. Bottom panel-left, In contrast to the Poiseuille-Hagen equation for flow/resistance, infarct volume tends to be smaller in strains with longer collaterals. A possible explanation is that longer collaterals have greater length from which penetrating arterioles can emanate during formation of the collateral circulation during vascular development. Such a difference would favor a smaller infarct volume after MCAO, which could offset the effect of longer collaterals. This hypothesis is supported by the significant correlation between collateral length and penetrating arteriole number shown in the bottom panel-right. (See Figure S5 for penetrating arteriole number for individual strains.) N-sizes for number of mice are given in figure 1. Average values for collateral length for a given strain were regressed against values for infarct volumes estimated from figure 1 of a previous study by others.9.1 B6 NZW KK C3H A AKR BALB SWR

Hematocrit BLKS FVB/N CBA DBA/2 NOD SJL 129S1 B6 NZW KK C3H A AKR BALB SWR Supplemental figure 4. Hematocrit of mouse strains. Means of each strain shown were obtained from the Mouse Phenome Database ( from the Peters et al study (2001).

L l Tortuosity Index, TI = L (Span) l (length) Penetrating arteriole MCA ACA Increasing infarct volume A B C DE F Supplemental figure 5. Collateral length (B), collateral span (C), tortuosity index (D) and number of penetrating arterioles, per collateral (E) and per length of collateral branching off of the collaterals (F), in 8 inbred strains, before (white bars, native collaterals) and 6 days after MCAO (black bars, remodeled collaterals). A, Diagram defining the measurements made from digital images using a tablet PC to trace collateral length, combined with ImageJ software. In each mouse, all collaterals between the MCA and ACA were analyzed to obtain mean values. The beginning points of a collateral, defined as its two points of departure from the end of a distal-most arteriole of the MCA and ACA tree, were identified as the points of increased tortuosity (characteristic of native pial collaterals), compared to their parent distal-most arterioles. B-F, Strains arranged in order of increasing infarct volume (see Fig 4A and its legend). B, White and black bars: Native and remodeled collateral length (6 days after MCAO) varied weakly with (Legend continued on next page)

strain (ANOVA), but did not correlate with infarct volume. Black bars: Collateral tortuosity increased during remodeling due to greater increase in length relative to span (see below). C, White bars: Collateral span (ie, a straight line connecting the ends of the collateral) varied with strain but did not correlate with infarct volume. Comparison of the strain-pattern for collateral span and length suggests, logically, that variation in collateral span necessarily associates with variation in collateral length. Black bars: Collateral span, as defined above (ie, “beginning points…”) could have increased during remodeling because: (1) the tips of the distal-most arterioles giving rise to a given collateral became slightly tortuous (“tip tortuosity”), widening the assigned beginning points of the collateral; (2) the collateral may have elongated, causing “buckling” of the tips of the parent distal-most arterioles. D, Collateral tortuosity increased significantly in two strains, but did not correlate with infarct volume. E, White bars: The number of penetrating arterioles branching off of a collateral varies with strain; strains with fewer penetrating arterioles per collateral have larger infarct volumes. Black bars: The number of penetrating arterioles did not change after remodeling. F, White bars: The number of penetrating arterioles, normalized to collateral length, varied with strain. Black bars: The tendency toward reduction reflects the increase in collateral span due to “tip tortuosity”. N-sizes for number of mice for each strain are given in Figure 1. One-way ANOVA and linear regression (the latter against infarct volume values shown in Figure 3) were performed on the white and black bar data separately, for all panels. *, **, *** p < 0.05, 0.01, grouped 2-tailed t-tests, comparing values after MCAO versus before MCAO (right hemisphere data versus left hemisphere data). We wished to compare the above tortuosity data with data in the literature. However, we were unable to find comparable data or published images that would allow us to quantify collateral tortuosity, as done above, after MCAO, with the exception of a study providing a high-resolution image of an adult rat pial circulation 20 days after unilateral MCAO, wherein maximal dilation and latex casting were employed (Coyle, 1984). We thus analyzed randomly selected collaterals in figure 1B of that study (20 days after unilateral MCAO, 10 collaterals each from the ligated and non-ligated hemispheres interconnecting the MCA and ACA), with methods used in Supplemental Figure 5. Tortuosity index for the non-ligated and ligated hemispheres were (mean  SEM, respectively): 1.34  0.03 and 1.61  0.08 (20.2% increase). This is greater than the increases in the B6 and C3H (12.2% and 10.1% increase, respectively). The greater increase in rat may reflect a species difference or that the measurement after MCAO was taken at 20 days in rat and at 6 days in mice. However, the rat measurements are only derived from a single rat. As determined from the rat images of Coyle, collateral length and span on the non-ligated side did not differ significantly from values calculated from the image of a control rat (figure 1A of Coyle) that did not receive MCAO (we measured 8 collaterals from each hemisphere, at random). This indicates that unilateral MCAO did not cause morphometric changes in the contralateral hemisphere of the same rat—consistent with our findings in mice (See first paragraph of Results).