Chronic two-photon imaging through a microprism combines the optical access of ex vivo brain slice preparations with in vivo behavioral context. This procedure involves insertion of a microprism attached to a cranial window (Figure 1A and Figure S1 available online). The hypotenuse of the microprism is coated with aluminum and thus serves as a right-angled mirror or “microperiscope,” with a vertical field-of-view parallel to the prism face. In different experiments, we implanted a microprism into Decitabine supplier either mouse somatosensory barrel cortex or visual cortex. As described in detail below (see Experimental Procedures and Figures S1A–S1D), a microprism (barrel cortex: 1.5 × 1.5 mm2 imaging face; visual

cortex: 1 × 1 mm2 imaging face) was glued to a coverslip. A craniotomy and durotomy were performed under sterile conditions, a small incision was made orthogonal to the cortical surface, and the microprism assembly was carefully inserted into Panobinostat in vivo cortex. Wide-field epifluorescence and two-photon images parallel to the cortical surface showed a vertical field-of-view across cortical layers 2–6 through the prism, revealing radial blood vessels and the expected laminar pattern of GCaMP3 expression (Figures 1B and 1C) or YFP expression (Figure 1D). The procedure for microprism insertion in the primary

visual cortex (V1) (Figures 1B and 1C) involved an ∼20% vertical compression of cortex (to ∼675 μm in area V1) to decrease brain motion and prevent dural regrowth at the cortical surface, as in previous studies (Andermann et al., 2011 and Dombeck et al., 2007). We first used microprisms for chronic two-photon structural imaging of genetically labeled cortical neurons across the depth of cortex. Somata and dendrites of layer 5 neurons in barrel cortex of anesthetized Thy1-YFP-H mice were imaged immediately following and for up to 2 months after prism insertion (n = 5; Figure 1D). Large field-of-view imaging with a 4× objective immediately

following prism insertion revealed labeled neurons in layers 2/3 and 5 (Figure 1D, left panel). Consistent with our earlier studies (Chia and Levene, 2009b), images included dendrites next of hundreds of neurons up to depths ∼900 μm below the pial surface. Imaging with a 40× objective, 29 and 68 days following prism insertion, yielded progressively clearer images, allowing visualization of fine structural details in proximal portions of layer 5 pyramidal neuron basal dendrites (Figure 1D, middle and right panels; Movie S1). The population of labeled neurons was stable over time, as demonstrated by tracking of over 40 neurons in one field-of-view across imaging sessions spaced 13 days apart (Figures S1E and S1F). We found that when the surface of the cortex around the prism was unobstructed, the fluorescence collection efficiency was improved and “shadow” effects of radial vessels located between the image plane and the prism face were reduced.