Ethics Statement: All procedures and animal maintenance were performed in accordance with Institutional guidelines, the German Animal Welfare Act, the European Council Directive 86/609/EEC regarding the protection of animals, and guidelines from local authorities (Berlin, T-0215/11)

1. Preparation of Acute-hippocampal Slices

1. Take a transgenic rat (17 to 24 day old), expressing the fluorescent Venus/YFP protein under the vGAT promoter, which labels the majority of cortical inhibitory interneurons¹³. Decapitate the rat. Rapidly dissect the brain (<40 sec) into semifrozen, carbogenated (95% O₂/5% CO₂) sucrose-based artificial cerebrospinal fluid (sucrose-ACSF, Figure 1A).
2. Assess the dissected rat brain for Venus/YFP fluorescence with a 505 nm LED lamp and 515 emission filter, mounted on a pair of goggles.
3. Remove the frontal third of the cortex and cerebellum; then separate the hemispheres, all with a scalpel. Remove the dorsal surface of the cortex to provide a flat surface to glue the brain down, as previously described^14.
4. Cut transverse slices (300 μm) of the hippocampal formation on a vibratome, the hemispheres should be surrounded with semifrozen, carbogenated sucrose-ACSF (Figure 1B)¹⁴. Remove additional regions of rostral cortex, midbrain and brainstem. Transfer each slice to a submerged holding chamber containing sucrose-ACSF, which is carbogenated and warmed to 35 ºC.
5. Leave the slices to recover at 35 ºC for 30 min from the time of the last slice entering the warmed ACSF. Do this in order to reactivate metabolic processes and facilitate the resealing of cut neuronal processes. Then transfer to room temperature for storage (Figure 1C).

2. Fabrication and Filling of Recording Pipettes

1. Pull patch pipettes from glass capillaries, so that a pipette resistance of 2-4 MΩ is achieved when filled with filtered (syringe filter, pore size: 0.2 μm) intracellular solution containing 0.1% biocytin (for intracellular labeling). Keep the intracellular solution chilled on ice to prevent degradation of its constituents.
2. Fill patch pipettes for identification of postsynaptic currents with a solution containing a physiologically relevant low Cl^– concentration (E_R(Cl-)= -61 mV; see solution list).
3. For paired recordings to identify the presynaptic receptor mediated responses, fill patch pipettes with intracellular solution with low Ca²⁺ buffer capacity to prevent interference with transmitter release presynaptically, as well as 4-fold higher Cl^– concentration (E_R(Cl-)= -20 mV) to improve signal-to-noise of observed IPSCs⁵ allowing accurate assessment of pharmacological responsiveness. Note that changing Cl^– concentration can alter IPSC kinetics¹⁵.

3. Whole Cell Patch-clamp Recording from FS-INs

1. Carbogenate the ACSF and feed through the perfusion system to the recording chamber, by means of a peristaltic pump (which also removes ACSF from the recording chamber through a suction line, Figure 2A). Turn on all equipment on the setup in preparation for recording.
2. Transfer a slice to the recording chamber and hold in place with a platinum ring strung with single fibers of silk. Position the slice so that the stratum (str.) pyramidale of CA1 runs vertically through the field of view, allowing access with 2 pipettes to both the str. radiatum and str. oriens simultaneously (Figures 2C and 4A).
3. Place the chamber into the setup and start perfusion of carbogenated and warmed (32-34 ºC) recording ACSF at a flow rate of 5-10 ml/min.
4. Assess slice quality under IR-DIC optics at 40X objective magnification, and visualize with a CCD camera viewed on a display. Assume good slice quality if a large number of round, moderately contrasted CA1 pyramidal cells (CA1 PC) can be seen in str. pyramidale at depths of 20-30 μm below a smooth and lightly cratered surface (Figure 2C). Poor quality slices contain large numbers of highly contrasted, shrunken or swollen cells, with an uneven slice surface.
5. Identify putative FS interneurons under epifluorescence illumination as those expressing Venus/YFP (Figure 2B), with large multipolar somata in or near the str. pyramidale. Select cells reasonably deep within the slice (50-100 μm, Figure 2C) in order to better preserve their morphological integrity.
6. Mount the recording electrode in the pipette holder on the headstage; then apply a low, positive pressure (20-30 mBar) through the tube line. Lower the pipette to the surface of the slice, slightly offset to the center of the selected neuron.
7. Obtain whole-cell recording configuration as described previously^14,16 and see also Figures 2D and 2E:
   1. Target a cell: Increase the pressure to 70-80 mBar and rapidly lower the pipette through the slice to just above the soma of the selected cell (Figure 2D, top).
   2. Approach the cell: Press the pipette against the cell membrane to produce a “dimple” on it (Figure 2D, top). Perform this step swiftly, in order to prevent biocytin labeling of neighboring cells.
   3. Create a giga-ohm seal: Release the pressure and simultaneously apply a 20 mV voltage command to the pipette. A giga-ohm seal (1-50 GΩ; Figure 2D, bottom and Figure 2E middle) typically develops rapidly. Once sealed, apply the expected resting membrane potential (typically between -70 and -60 mV) as a voltage command.
   4. Break through the patch: Once sealed, rupture the membrane patch with a short pulse of negative pressure; thereby achieving the whole-cell configuration (Figure 2E, bottom).
8. Compensate whole-cell capacitance and series resistance (R_S). R_s is normally 5-20 MΩ and stable for up to 120 min. Abandon cells if membrane potential (V_M) on break-through is more depolarized than -50 mV; R_S is initially greater than 30 MΩ; or R_S changes by more than 20% over the course of the recording.
9. Identify FS-INs by their response (in current-clamp mode) to a family of hyper- to depolarizing current pulses (-250 to +250 pA, Figure 2F, top). FS-INs have relatively depolarized V_M (typically -50 to -60 mV), short membrane time-constant (<20 msec) and respond to a 500 pA depolarizing current injection with a train of action potentials (APs) at frequencies >100 Hz ¹¹ (Figure 2F, bottom), which are markedly different from those in CA1 PCs (Figure 2F, middle).

4. Extracellular Electrical Stimulation to Evoke GABA_BR-mediated Responses

1. To observe synaptically evoked responses, position an extracellular stimulation electrode (a patch pipette filled with 2 M NaCl; Resistance: 0.1-0.3 MΩ) in the slice at the border of str. radiatum and str. lacunosum-moleculare. Position the electrode 200-300 μm lateral to the soma to prevent direct electrical stimulation of the cell and minimize stimulation artifacts (Figure 3A).
2. Once the stimulation electrode is positioned, obtain whole-cell recording of the chosen cell and assess the physiological phenotype in current-clamp mode as in section 3.9 (Figure 3B).
3. With the neuron recorded in voltage-clamp (V_M -65 mV), deliver electrical stimulation of presynaptic axons at 50 V (~500 μA effective stimulus) every 20 sec, using an isolated constant-voltage stimulator. Use single stimuli (100 μsec duration, Figure 3C, top) to observe GABA_BR mediated IPSCs, and interleave with trains of multiple stimuli (at 200 Hz) to produce greater transmitter release.
4. Bath apply ionotropic glutamate receptor antagonists (AMPA receptor: DNQX [10 μM]; NMDA receptor: d-AP5 [50 μM]) to reveal the isolated monosynaptic IPSC (Figure 3C, middle upper). Further isolate the GABA_BR-mediated IPSC with application of a GABA_AR blocker (gabazine [10 μM]; Figure 3C middle lower).
5. Confirm the resultant slow-outward current (Figure 3C lower, expanded) as being GABA_BR-mediated by the subsequent application of CGP-55,845 [5 μM] (Figure 3C lower, underlain in grey)

5. Paired Recordings of Synaptically Coupled FS-IN and CA1 PCs

1. Assess GABA_BR-mediated presynaptic control of inhibitory synaptic transmission with simultaneous recordings, performed between synaptically-coupled IN and PC pairs as described below.
2. First, establish a whole-cell recording of a presynaptic interneuron (as in section 3) and confirm the FS phenotype (Figure 4A).
3. Then patch a neighboring CA1 PC (20-100 μm distance, Figure 4A) and apply brief suprathreshold depolarizing current pulses (1 msec duration, 1-5 nA amplitude) to the presynaptic IN (held in current-clamp mode) to elicit APs. If a synaptic connection is present, APs in the IN result in IPSCs in the CA1 PC, held in voltage-clamp (compensate R_S to about 80%).
4. If necessary, fill a new recording electrode and record from further CA1 PCs until a connection is found.
5. Once a connection is established, elicit pairs of APs in the presynaptic FS-IN to assess both the unitary synaptic response and dynamic behavior. Use a typical paired-pulse protocol of 2 depolarizing stimuli with a 50 msec interval (Figure 4B).
6. Collect control traces in baseline conditions. Then, apply the selective GABA_BR agonist baclofen (10 μM) to the perfusing ACSF, thus activating GABA_BRs, followed by the antagonist CGP-55,845 (5 μM), to fully block the receptor mediated effects. Collect ~50 traces during steady state of each drug condition (Figure 4B and C).
7. Once the recording is complete, seal the somatic membrane by forming an outside-out patch: Slowly withdraw the pipette from the cell body in V-clamp and as the R_S increases, reduce the V_M to -40 mV. Do this to facilitate the formation of the outside-out patch; then remove the pipette from the bath.

6. Analysis of Electrophysiological Properties

1. NOTE: A multitude of different software packages are available for the acquisition of electrophysiological data. Here, WinWCP, a Windows program in the free Strathclyde Electrophysiology Software package is used, which allows recording of up to 16 analog input channels and output of 10 digital signals.
2. Low-pass filter all data at 5-10 kHz and sample at 20 kHz.
3. Analyze physiological data with an off-line analysis suite.
   NOTE: Stimfit, an open source software package which includes a Python shell, is used in this instance; however other alternatives can easily be used instead.
   1. Analyze passive membrane properties of recorded neurons, acquired in current clamp, from resting membrane potential.
   2. Measure the mean resting membrane potential from the baseline of recorded responses from the beginning of the recording.
   3. Calculate input resistance, using Ohm’s law, from the voltage response to the smallest hyperpolarizing current pulses (≤ -50 pA). To improve signal to noise ratio, average multiple traces. Note: our examples are typically averages of 10-50 individual sweeps.
4. Estimate the apparent membrane time constant by fitting a monoexponential curve to the decay of the responses to the smallest hyperpolarizing current pulses.
5. Analyze action potential waveform to determine threshold, amplitude (threshold to peak) and duration (width measured at half height) elicited by threshold level depolarizing current pulses.
6. Analyze GABA_BR mediated IPSCs from voltage clamp recordings. Filter traces off-line at 500 Hz (Gaussian filter) and assess the peak amplitude and latency of the GABA_BR mediated response (in averages of at least 10 traces).
7. Detect the effect of GABA_BRs on the inhibitory output of INs as a change in peak amplitude of the GABA_AR-mediated IPSCs measured between peak and preceding baseline. Calculate the mean amplitude from ≥50 traces for the control period and the steady-state of all pharmacological epochs.

7. Visualization and Immunocytochemistry of FS-Ins

1. Following the recordings, fix the slices by immersion in 4% paraformaldehyde with 0.1 M phosphate buffer (PB, pH=7.35) O/N at 4 °C.
2. If necessary slices can be transferred to PB and stored for up to ~1 week before processing.
3. Wash slices liberally in fresh PB and subsequently in 0.025 M PB with 0.9% NaCl (PBS, pH=7.35).
4. To reduce non-specific antibody binding, block the slices for 1 hr at RT in a solution containing 10% normal goat serum (NGS), 0.3% Triton-X100 (a detergent to permeabilize membranes) and 0.05% NaN₃, made up in PBS.
5. To label for PV expression, use an anti-PV monoclonal mouse antibody diluted in a solution containing 5% NGS, 0.3% Triton-X100, 0.05% NaN₃, in PBS. Incubate primary antibodies for 2-3 days at 4 °C¹². Rinse slices thoroughly in PBS.
6. Apply fluorescent anti-mouse secondary antibodies (e.g. Alexafluor-546) along with the biotin binding-protein streptavidin, conjugated to a fluorochrome (e.g., Alexafluor-647); and incubate in a solution containing 3% NGS, 0.1% Triton-X100, 0.05% NaN₃, diluted in PBS and incubate O/N at 4 °C.
7. Liberally rinse slices 2-3x with PBS followed by 2-3 rinses in PB. Mount the slices on glass slides. Use a 300 μm agar spacer to prevent the slice from collapsing. Cover-slip slices with a fluorescent mounting medium and seal with nail-varnish.

8. Imaging and Reconstruction of Visualized FS-Ins

1. Visualize the slices using a scanning confocal microscope, with the fluorochome reporter excited with the appropriate laser line (diode laser 635 nm for Alexafluor-647; Helium-Neon 543 nm for Alexafluor-546 labeling for PV and Argon 488 or 515 nm for Venus/YFP).
2. Take images at an appropriate Z-resolution (typically 0.5-1 μm steps, using a 20X objective) to produce a Z-stack of the whole cell. Multiple stacks are normally required to image the whole cell, which can be digitally stitched off-line using FIJI/ImageJ software (Figure 5A).
3. Reconstruct the cell from the stitched image stack using a semi-automatic tracing method (Simple Neurite Tracer plugin in FIJI/ImageJ software package¹⁷, Figure C).
4. Finally, assess the PV-immunoreactivity of the interneuron with a high numerical aperture objective lens (60X silicon-immersion, N.A.=1.3). Make images of the soma, proximal dendrites and proximal axon, or alternatively of axon terminals if somatic washout of PV is too strong. Cells are deemed immunoreactive for PV if immuno-labelling is seen to align with the biocytin-labeled structures (Figure 5B).