GABA-A receptors are the site of action of many clinically important drugs. They are ligand gated chloride channels with pentameric structure assembled as from 19 members of the family. The non-selective benzodiazepine agonist diazepam potentiates the GABA-A receptor through all α1/α2/α3/α5βγ2 receptor subtypes resulting in a broad range of beneficial effects such as sedation, anxiolysis and anticonvulsion. Selective activation or suppression of GABA-A receptors containing different α subunits produces a distinct behavioral profile (α1 mediate the sedative and anticonvulsant effects of benzodiazepines, whereas the α2 and α3 mediate their anxiolytic effects). The α5 subunit-containing receptors are prominently expressed in the hippocampus and there is genetic and pharmacological evidence for a modulatory role in learning and memory. In support of a potential therapeutic use of GABA-A α5 negative allosteric modulators (NAM) to treat cognitive dysfunction e.g. in Down Syndrome, we have recently shown that repeated treatment with the selective GABA-A α5 NAM RO4938581 (Roche, Switzerland) rescues synaptic plasticity, improves spatial learning and reverses the pronounced hippocampal perfusion deficits observed by arterial-spin labeling MRI in the TS65Dn mouse model^1,2. To further explore the specificity of this finding and investigate the circuitry engaged by modulation of the GABA-A α5 subtype-containing receptors we performed pharmacological MRI (phMRI) studies in the rat using a selective GABA-A α5 NAM (RO4938581) and a GABA-A α5 positive allosteric modulator (PAM) and contrasted the neurofunctional responses to that of diazepam.phMRI studies were conducted in rats (male, 250-300g) sedated with a s.c. infusion of medetomidine (0.1mg/kg/h following a priming bolus of 0.2mg/kg). Each drug was tested over a range of doses and phMRI was performed at the time of expected maximum exposure (Table 1). MRI was carried out on a Bruker BioSpec 9.4T/20cm system equipped with a bird-cage resonator for excitation and a surface coil for reception. Perfusion MRI was performed according a protocol described earlier³. Briefly, continuous arterial spin labelling with centered-RARE readout (TR/TE = 3000ms/5.7ms, RARE factor = 32, FOV = 4cm x 4cm, 128 x 64 matrix, 8 slices). Images were registered to an anatomical template with an associated atlas defining 62 regions of interest (ROIs). Perfusion values for each ROI were normalized slice-wise to brain-mean perfusion to derive region-specific values independent of inter-individual differences of the animals’ global hemodynamic status. ROI-wise differences between drug-dosed groups and respective vehicle group were tested for significant effects using univariate statistics (ANOVA, post-hoc Welch’s t-test).All drugs tested led to distinct, region-specific changes in normalized perfusion compared to vehicle in a dose-dependent manner resulting in significantly modulated regions-of-interest at the higher doses (number of significant ROIs: GABA-A α5 NAM: 0, 0, 4 for 1, 3, 10 mg/kg; GABA-A α5 PAM: 2, 2, 6 for 10, 30, 100 mg/kg). Diazepam led to significant changes at both doses tested (9, 6 ROIs for 3, 10 mg/kg). Figure 1 shows the significantly modulated brain regions for all 3 drugs each at the dose with the highest number of affected ROIs. In line with the restricted expression of GABA-A α5 subunit-containing receptors the selective α5 NAM and PAM compounds elicited perfusion changes in fewer brain regions than diazepam but all compounds share perfusion changes in the hippocampus and the caudate putamen. Thalamus, VTA, amygdala and sensory-motor cortex are exclusively modulated by diazepam. Figure 2 shows the compound specific and dose-dependent response pattern in hippocampal subregions. The increased perfusion in the dorsal hippocampus upon acute dosing with the GABA-A α5 NAM (Fig. 1A, 2A) is in line with our previous data in the TS65Dn mouse model where chronic dosing reversed the hippocampal perfusion deficit and is likely reflecting a reduced inhibition (enhanced activity) upon negative allosteric modulation of local GABA-ergic circuits. Accordingly, both the GABA-A α5 PAM and diazepam decreased the perfusion in the hippocampus (significantly for PAM in dHpc and CA; for diazepam in CA region, Fig 2B, 2C). A different pattern emerged in the ventral hippocampus with an increased perfusion upon GABA-A α5 PAM and a trend for an opposite response following GABA-A α5 NAM and diazepam treatment.Our results demonstrate a differential phMRI signature upon acute dosing with either selective GABA-A α5 compounds or diazepam. The diazepam signature is characterized by a broader range of modulated brain regions, reflecting engagement of the neurocircuits (via α1/α2/α3/α5βγ2 receptor subtypes) underlying the effects of non-selective benzodiazepine agonists on sedation, anxiolysis, anticonvulsion and cognitive impairment. On the other hand, the α5-subunit selective compounds show a more restricted pattern reflecting primarily a selective receptor activation or suppression in regions with high α5 subunit-containing GABA-A receptor expression.