MRI of MMP-9 gene activity in vivo after GCI. A, MRI protocol. Phosphorothioate-modified ODN with randomized sequence (sODN-Ran) or antisense sequence to mmp-9 mRNA (sODN-mmp9) was linked to SPION-NA and delivered at the time indicated according to the experimental protocol; we intracerebroventricularly infused SPION-mmp9 or SPION-Ran (40μg Fe/kg) at 3 h of reperfusion, and acquired MRI scans for R2* maps (for SPION retention) and DWI. B, R2* subtraction maps of SPION-mmp9: individual R2* maps of SPION-mmp9 minus the averaged SPION-Ran maps obtained from three mice in this study (multislice images for each animal displayed in rows). C shows group average R2* maps obtained from animals infused with SPION-mmp9 (top row) and SPION-Ran (bottom row). D depicts outlines of ROIs for statistical analyses performed according to the work of Paxinos and Franklin (2001). All are rostral view. E, Bar graphs from ROI analysis that quantitatively compares the difference between SPION retention at 10 h of reperfusion after GCI (60 min) in animals that received SPION-mmp9 (open bars, N = 3) and SPION-Ran (closed bars, N = 3), *p < 0.05 by t test.

Immunohistochemistry of MMP-9 antigen in C57black6 after 60 min GCI (n = 5). Protocol is shown in A. Regional metabolic disturbance was identified by hDWI 1 d after GCI (rostral view of one animal is shown in B). Postmortem brain tissue sections were obtained after MRI and assayed for MMP-9 antigen using rabbit polyclonal anti-MMP-9 IgG followed by secondary FITC-anti-rabbit IgG, and each sample was counter-stained for endothelia using Cy-3-griffonia simplicifolia lectin I and Hoechst stain for nuclei (C-F). The same GCI mouse without primary antibodies is shown (G). MMP-9 antigen is present in endothelia (H). No detection of anti-MMP-9 IgG is shown when the antibodies are deleted (I).

A, A rostral view of regions of metabolic disturbance as depicted by DWI or ADC in two representative mice at 1 d (first row), compared with Gd-enhanced MRI acquired at 8 weeks (second row). ADC maps with highlighted areas correspond to regions of significant metabolic disturbance (ADC drop below threshold) and predict disruption of the BBB at 8 weeks (post- and pre-Gd image of Gd-MRI, second rows). Our previous studies have shown that BBB leakage varied (25% of the animals during the first 3 h of reperfusion after 60 min BCAO using Gd-DTPA), but all animals developed BBB leakage by 4–12 weeks (Liu et al., 2008). B, C, Temporal changes in the VMD of the cortex and striatum. Cerebral metabolic disturbance is defined as a region with ADC below the threshold ADC of normal mice (2 SEM below average ADC in normal mice: striatum threshold was 4.69 × 10⁻⁴ mm²/s; cortex threshold was 5.41 × 10⁻⁴ mm²/s). The areas of metabolic disturbance were obtained from multiple MR slices (representative slide shown above the B for cortex and C for striatum), and VMD was determined by summation of these areas in the corresponding brain regions and multiplying by the thickness of each MRI slice (1 mm). The VMD value at each time point was compared with the VMD at the end point at 6 d by t test (*p<0.03). The corresponding number of animals (N) used at each time period is included at the bottom of the bar graph. We did observe 4 of 12 of the mice had significant rADC below the threshold (supplemental material, available at www.jneurosci.org) in the hippocampus; we did not include them in the volume calculation for metabolic disturbance.

A modified ex vivo hybridization assay to determine the presence of intracellular mRNA in a total of 12 mice (Cui et al., 1999). The protocol is the same as in Figure 2A, except FITC-sODN was delivered. We delivered FITC-labeled sODN (6 nmol/kg, i.p.) to the mice at 1 h of reperfusion, because the BBB is open in 75% of GCI-treated mice (Liu et al., 2008). One day after GCI, we obtained DWI and prepared postmortem samples from the region where hDWI is identified in the MRI (A, C, E, G). The sODN-mmp9 sequence has been shown to bind to mmp-9 mRNA of the mouse (Zechel etal.,2002).Samples were counterstained with Hoechst for nuclei and Cy3-griffonia simplicifolia lectin I for the vessel endothelia (B,D-F). The presence of enhanced FITC stain indicates the location of probe retention. H and I show detection of MMP-9 antigen (red) and endothelial cells (green), using the same antibodies as in Figure 4 (labeling was reversed to show difference from hybrids). Arrows show the regions where photography was taken. CPU, Caudate putamen.

MMP-9 activity knockdown using siDNA (sODN-mmp9). Phosphorothioate-modified ODN with randomized sequence (sODN-Ran) or antisense sequence to mmp-9 mRNA (sODN-mmp9) was infused at 1 and 3 h; all samples were collected at 10 h of reperfusion. Protocol is shown in A. B and C show two paired observations of MMP-9 activities in postmortem samples (20μg of protein per lane) from bilateral frontal regions of individual animals, as assessed by zymography (You et al., 2007). The intracerebroventricular infusion was at 3 h reperfusion in B (sODN = 20 nmol/kg, brain extracts were obtained from left hemisphere) or at 1 h reperfusion in C (sODN = 120 nmol/kg, brain extracts were obtained from bilateral frontal brain at 1–2 bregma). Positive control (PC) samples showing human MMP-9/2 protein (Millipore Bioscience Research Reagents #CC073) and C57black6 mouse (TBI) were included as markers along with mouse controls (SO or CGI without intracerebroventricular infusion). D shows changes in cerebral rADC [mm²/s (× 10⁻⁴); three groups of mice are shown: mice with GCI but without intracerebroventricular infusion (open bars, n = 11), mice that received sODN-mmp9 (dark bars, n = 6; *p = 0.004, t test compared to mice with sODN-Ran), and sODN-Ran (gray bars, n = 6)]. E compares striatal VMD (mm³, mean ±SEM) in two groups of mice that received sODN-mmp9 (dark bar, 8.3±1.3) or sODN-Ran (gray box, 12.8±1.0) 1 d after BCAO (n = 6 each). *p = 0.004, t test. There was no significant difference between mice after BCAO with or without intracerebroventricular infusion of sODN-Ran (data not shown). The control antigen (β-actin) in B was stained with primary antibody mouse anti-β-actin (Sigma A-5441) followed by a secondary antibody: Cy5-conjugated donkey anti-mouse antibody (Jackson ImmunoResearch). Scanned with Typhoon 9400 (GE Healthcare). The control antigen in C was the same except the secondary antibody was HRP-conjugated sheep anti-mouse antibody (GE Healthcare detection: ECL Western blotting detection of reagents), exposed to x-ray film (Kodak MioMax MR film).