To avoid bleed-through of stimulus light into the detection pathway of the microscope, the built-in red, green, and blue LEDs of the microprojector were externally supplied by a custom-built power source that was synchronized to the fly-back interval of the laser beam at the end of each line, during which no fluorescence data were acquired (fly-back width 0.33 ms, line frequency 864 Hz, 28.5% duty cycle). This generated a virtually flicker-free stimulus sequence. The animal was UMI-77 chemical structure positioned with the right eye facing the center of the projection area, which covered ∼90° of the visual field. The stimulus was a white bar on dark background, moving at a speed of 35°/s, in four or eight directions evenly spanning

360°. For whole-cell recordings in the tectum, the skin overlying the midbrain was cut with an etched tungsten needle and removed with fine forceps. The extracellular recording solution contained 134 mM NaCl, 2.9 mM KCl, 2.1 mM CaCl2, 1.2 mM MgCl2, 10 mM HEPES, and 10 mM glucose; pH 7.8/290 mOsm/kg. Patch pipettes were pulled from borosilicate glass (Hilgenberg, outer diameter 2 mm, inner diameter 1 mm) and filled with internal solution BMS-354825 chemical structure (125 mM K-gluconate, 10 mM HEPES, 10 mM EGTA, 2.5 mM MgCl2, 4 mM ATP-Na, and 0.3 mM GTP; pH 7.3/285 mOsm/kg). In some experiments, K-gluconate was replaced with Cs-gluconate to minimize voltage-gated and leak potassium conductances during voltage-clamp recordings. To analyze neuronal morphology, we

added sulforhodamine-B or Alexa Fluor 594 hydrazide (both 360 μM; Invitrogen) to the internal solution. Open tip resistance was 7–9 MΩ. Input resistance of tectal neurons was 2.8 ± 0.3 GΩ (n = all 19). Patch-clamp recordings were performed using a Multiclamp 700B amplifier (Molecular Devices). Signals were filtered at 3 kHz and recorded at 10–20 kHz using a PCIe-6251 board and custom-written LabVIEW data acquisition software (version 8.6, National Instruments). To isolate excitatory and inhibitory synaptic currents, we voltage clamped cells at −60mV (close to the reversal potential of GABA receptor channels) and 0mV (close to the reversal potential of glutamate receptor channels), respectively. Cell spiking

was recorded in the cell-attached mode or in the whole-cell current-clamp configuration. During current-clamp recordings, small hyperpolarizing current was injected in some cases to keep the cell at a resting potential near −60mV. In current clamp, action potentials were often small but could clearly be detected as spikes by taking the derivative of the voltage trace due to the fast depolarization of the membrane potential at spike onset. We thank W. Denk for generous support and helpful discussions. We thank B. Knerr, M. Glöck, and A. Wolf for their contribution in generating transgenic lines. We thank A. Borst, W. Denk, S. Preuss, and F. Svara for comments on an earlier version of the manuscript, J. Tritthardt and C. Kieser for expert help with electronic design, M. Lukat and N. Neef for mechanical design, M.