Project Summary Recent years have seen major breakthroughs in methodology for studying two complex yet fundamental aspects of brain structure: synaptic connectivity patterns and the heterogeneous distribution of molecules. Due to an ongoing technical barrier that has endured for decades, advances in circuit imaging and molecular imaging have progressed almost entirely in parallel, and there are still no routine methods for integrating molecular information into synaptic circuit maps. Imaging brain structure with enough resolution to visualize synapses requires electron microscopy (EM), and EM is not compatible with the standard methods used to identify molecules by light microscopy. It is clear from biochemical data that highly multiplexed labeling of proteins and RNA transcripts will be necessary to generate comprehensive maps of the brain’s molecular structure. To address this need, a number of approaches aiming to extend the spatial resolution and limits of multiplexed labeling of fluorescence microscopy have been developed. The resolution of EM is still orders of magnitude higher than any light-level technique, however, and EM remains the only modality that reveals structural details. EM also presents a unique opportunity for molecular labeling. EM image volumes are reconstructed from serial ultrathin sections, and by applying a different probe to each section a large number of molecules can be localized in a single structure – hundreds or more in the case of a neuron. In contrast to tissue specimens used in light microscopy, ultrathin EM sections are not readily amenable to simple immunohistochemistry (IHC) or in situ hybridization (ISH) protocols. A major reason for this is incompatibility between sample preparation practices: the strong fixatives and dense embedding resins used in EM damage or occlude molecular targets, while the harsh treatments used to facilitate molecular detection degrade fine tissue structure. The problem can be circumvented by the use of specially engineered transgenic reporters, but these do not solve the problem of detecting endogenous molecules in large numbers. In this project, we will draw on long-established methods from the EM and histology fields to develop an unconventional approach to labeling EM sections, and apply this approach to identify molecular cell and synapse types using three different workflows. Our strategy employs removable embedding media, which are standard in light microscopy and which, contrary to traditional assumptions, we have found to be perfectly compatible with EM imaging. To maximize efficiency and flexibility in imaging workflows, we will develop labeling protocols that prioritize resolution, sensitivity, and throughput to different degrees. If successful, this project will produce methods uniquely capable of combining EM-level structural imaging with multiplexed labeling of endogenous molecules, and will dramatically increase the depth of information obtained from EM volume recons...