SUPPLEMENTARY INFORMATION doi:10.1038/nature09511 Supplementary Table I Blood pressure and heart rate measurements pre- and post-stroke Pre Post 7-days Systolic Diastolic BPM Systolic Diastolic BPM Systolic Diastolic BPM Control 105.0 1.6 74.1 1.0 599.5 16.6 105.1 2.4 75.6 2.9 580.4 20.0 105.2 0.2 70.3 2.9 553.6 11.8 Stroke + vehicle 103.5 2.6 75.8 2.1 587.7 9.9 104.3 0.8 75.0 1.2 657.3 27.6 106.5 0.9 72.2 0.5 565.7 15.2 Stroke + L655,708 102.9 2.5 72.7 1.3 584.0 33.2 103.1 2.1 71.9 2.3 649.8 38.8 105.8 3.6 73.7 1.8 549.3 27.5 Gabra5 -/- Stroke 108.8 2.6 76.5 1.8 582.3 20.3 108.0 1.3 78.4 2.0 650.5 11.2 104.1 2.7 75.1 2.1 590.9 7.6 Gabard-/- Stroke 107.2 3.0 74.0 3.0 568.7 10.5 102.4 1.9 75.2 0.8 682.3 36.8 104.2 2.4 72.2 2.7 561.8 17.2 Data represents the mean SEM from n=4 per group. BPM = beats per minute. www.nature.com/nature 1
RESEARCH SUPPLEMENTARY INFORMATION Supplementary Figure 1. Schematic summary of key results. Stroke increases peri-infarct GABA by reducing the level of GABA transporter GAT3/GAT4, indicated by lighter shading in a subset of GAT3/4 (blue pinwheels). The function and level of GAT1 are unaltered (orange pinwheels). Due to different subunit-associated properties, extrasynaptic GABAa receptors (green) mediate a tonic form of inhibition that is distinct from the phasic form mediated by synaptic GABAa receptors (orange). Tonic inhibitory currents in peri-infarct pyramidal neurons (red trace) are increased compared to control neurons (blue trace). Reducing tonic inhibition with a selective alpha5- GABAAR inverse agonist (L655,708) reverses the increase in tonic inhibition and improves behavioral recovery in forelimb motor control after stroke. 2 WWW.NATURE.COM/NATURE
SUPPLEMENTARY INFORMATION RESEARCH Supplementary Figure 2. Additional analyses of post-stroke tonic inhibition. (a) Cumulative probability plots of Itonic in all recorded control and post-stroke neurons. All post-stroke neurons were grouped together, as an ANOVA analysis indicated no differences among the 3d, 7d, and 14d post-stroke groups (F2, 42 = 0.10; P=0.90). Two distributions could be fitted (see Methods section) to each of the two cumulative probability curves: the components are as follows: control: R1 = 0.51, X1 =4.16 pa/pf, p1 = 2.24; R2 = 0.49, X2 =10.41 pa/pf, p2 = 9.0; post-stroke: R1 = 0.55, X1 =6.62 pa/pf, p1 = 5.49; R2 = 0.45, X2=15.45 pa/pf, p2 = 2.96. This indicates that the ratios of the two distributions were not much affected by the stroke, but the two means increased between 48-59%. The overall mean increase in Itonic was 47%, indicating that peri-infarct cortex is subject to highly elevated tonic inhibition. (b) To confirm our finding, we performed the tonic inhibitory current measurements by an alternative approach of using high-[cscl]-based internal pipette solution, along with blockade of excitatory currents by including 3mM kynurenic acid in ACSF, to record GABAergic currents at a holding potential of -70 mv. Left: Representative traces from control and post-stroke neurons. Right: Box-plots showing the significantly increased tonic inhibition in post-stroke neurons compared to controls (control: 3.66±1.21 pa/pf, n=7, vs. post-stroke: 9.05±2.15 pa/pf, n=6; *P<0.05, Mann-Whitney test; boxes: 25-75%, whiskers: 10-90%, lines: median). www.nature.com/nature 3
RESEARCH SUPPLEMENTARY INFORMATION A. B. C. D. E. F. Supplementary Figure 3. Properties of phasic excitation, inhibition, and E(GABA) after stroke. (a,b) Phasic excitation measured as Imean was unaltered at any time-point after stroke. The four raw traces (a) taken from control and 3-, 7-, and 14-days post-stroke neurons indicate similar frequencies and amplitudes of the sepscs recorded at -70mV. Their kinetic properties were measured individually and the lack of differences was confirmed by separate analyses (data not shown). (c,d) Phasic inhibition measured as Imean was largely unaltered, except for a transient decrease at 7-days-post-stroke. The four representative traces were recorded in the same - cells as the sepscs but at +10mV. (e,f) Resting membrane potential (Vrest) and GABA reversal potential (EGABA) measured using voltage ramps (-100 to +200mV) in the cell-attached configuration in control (e) and post-stroke neurons (f), showing that GABAergic transmission remains hyperpolarizing after stroke. Neither of these values were significantly different between control and post-stroke neurons (Control: Vrest = -58.4 ± 7.1mV, EGABA = -70.4 ± 10.5mV vs. post-stroke: Vrest = -60.4 ± 3.4mV, EGABA = -77.4 ± 8.7mV; P>0.05). This and all electrophysiological recordings were performed on layer-2/3 pyramidal neurons in motor cortex adjacent to the infarct. 4 WWW.NATURE.COM/NATURE
SUPPLEMENTARY INFORMATION RESEARCH Supplementary Figure 4. Additional analyses on the effects of GAT blockade and L655,708. (a) To demonstrate that GABAaRs are not saturated after GAT blockade, GABA concentration in ACSF was raised (+10 um) after blocking both GATs (by 10 um NO-711 + 40 um SNAP-5114). Graphs show the percent increase in Itonic (control: 25.7±15.0%, n=3; post-stroke: 30.6±2,72%, n=3), indicating lack of receptor saturation in either control or poststroke slices. (b) Bar graphs showing the effects of L655,708 (100 nm) on tonic and phasic inhibitory currents (Imean normalized to cell capacitance). Note that the effects on phasic inhibition are minimal, compared to the effects on tonic inhibition. (Tonic inhibition (pa/pf): control: 8.05±0.80, n=24; ctrl+l655: 5.97±1.50, n=4; poststroke: 13.60±1.41, n=45; stroke+l655: 10.17±1.74, n=13. Phasic inhibition (pa/pf): control: 0.71±0.09, n=26; ctrl+l655: 0.70±0.22, n=4; post-stroke: 0.89±0.09, n=44; stroke+l655: 1.31±0.84, n=13). www.nature.com/nature 5
RESEARCH SUPPLEMENTARY INFORMATION GAT-4 GAT-4/ Supplementary Figure 5. GAT protein levels after stroke. (a) Western blot of GAT-1 (top) and GAPDH (protein loading control, bottom). The blots are taken from two lanes from each condition. In all Western blot experiments for GAT-1 and GAT-3 (mouse GAT-4) samples were taken from five stroke animals and three controls. All Western blots were run in triplicate. (b) Western blot of GAT-3/GAT-4 (top) and GAPDH (bottom). The lanes are in the same configuration as (a). (c). Quantification of GAT-1 protein level in a ratio to GAPDH in control and stroke. There is no significant difference between stroke and control for GAT-1 (p=0.45). (d) Quantification of GAT-3/GAT-4 protein level in a ratio to GAPDH (1/GAPDH) in control and stroke. GAT-3/4 is significantly reduced in peri-infarct cortex (* = p< 0.01). Each Western blot was run in triplicate. 6 WWW.NATURE.COM/NATURE
SUPPLEMENTARY INFORMATION RESEARCH Supplementary Figure 6. Effect of two-week L655,708 treatment on functional recovery. Functional recovery was assessed behaviorally on both the gridwalking task for forelimb footfaults (a) and hindlimb footfaults (b), and on the cylinder task for forelimb asymmetry (c). Data are shown as mean ± s.e.m. for n=8 per group, *** = P<0.001 for stroke + vehicle vs Sham; ## = P<0.01, # = P<0.001 vs stroke + vehicle. www.nature.com/nature 7
RESEARCH SUPPLEMENTARY INFORMATION Supplementary Figure 7. Acute vs. Chronic Effect of L655,708 on behavioral recovery. L655,708 administered 30 minutes prior to behavioral testing on days 7 and 14 had minimal but significant effect on forelimb control in grid walking at day-7, and no significant effect on day-14. In percent difference, acute administration of L655,708 produced a 14% gain of function at 7-days compared to 45% following chronic administration at day 7 after stroke. At 14-days acute administration of L655,708 produces a non-significant 3% improvement in forelimb motor control vs 44% for chronic treatment. Data are shown as mean ± s.e.m. for n=8 per group. 8 WWW.NATURE.COM/NATURE
SUPPLEMENTARY INFORMATION RESEARCH Supplementary Figure 8. Infarct size and peri-infarct neuronal number. (a) Stereological counts of neuronal number in peri-infarct and contralateral cortex in wild type animals, and in peri-infarct cortex in Gabra5-/- and Gabrd-/-. (b) Infarct size in Gabra5-/- and Gabrd-/- at 7-days after stroke. Compare with Figure 4d. (c-j) Low-power photomicrographs of peri-infarct cortex stained with an antibody against NeuN (c,d, g-j) or GFAP (e,f ). Panels a-f are taken from wild-type mice in peri-infarct cortex (c,e) and contralateral cortex (d,f ). Note the presence of reactive astrocytes in peri-infarct cortex, at a site in which neurons were not lost. (g,h) are from peri-infarct and contralateral cortex of Gabra5-/- mice. (I,j) are from peri-infarct and contralateral cortex of Gabrd-/- mice. Note the lack of neuronal loss in peri-infarct cortex in either knockouts or wild-type, supporting the quantitative measurements of neuronal number in (a). In (c,e,g,i) the alphabet labels are located at the edge of the infarct, identified by the increase in fluorescent background in NeuN staining, or by the barrier of pallisading astrocytes in (e). w w w. n a t u r e. c o m / NATURE 9
RESEARCH SUPPLEMENTARY INFORMATION Supplementary Figure 9. Effects of picrotoxin treatment on post-stroke behavioral recovery alone and in concert with L655,708. Functional recovery was assessed behaviorally on both the gridwalking task for forelimb footfaults (a) and hindlimb footfaults (b), and on the cylinder task for forelimb asymmetry (c). Data are shown as mean ± s.e.m. for n=8 per group, *** = P<0.001 for stroke + vehicle vs Sham; # = P<0.001 vs stroke + vehicle; ^^ = P<0.01, ^ = P<0.001 vs stroke + L655,708. 10 WWW.NATURE.COM/NATURE