(568l) Spectral Imaging and Analysis of Lung Tissue Using Fluorescence and Confocal Microscopy

Tracking critical proteins that control the frequency response and spatial spread of intracellular signaling molecules in the pulmonary vasculature is a critical step in assessing the role these agents play in pathologies such as pulmonary edema and acute respiratory distress syndrome (ARDS). Fluorescent proteins (green fluorescent protein, red fluorescent protein) are ideal tools for tracking cell signaling these proteins in cell cultures and living tissues. However, attempts to quantitatively detect fluorescent protein expression in the lung using traditional fluorescence and confocal microscopy have been unsuccessful, due to the very high background autofluorescence of the pulmonary vasculature and surrounding lung tissues. The inability to accurately identify and quantify fluorescent protein expression has, in turn, limited the in vivo study of intracellular signaling molecules in the pulmonary vasculature. Hence, a method is needed to allow quantification of fluorescent protein expression at the microscopic level in intact lung tissues.

To address the high autofluorescence of pulmonary vasculature, we have used an approach combining fluorescence microscopy with spectral imaging techniques and image analysis tools. In specific, we have used an inverted fluorescence microscope (TE2000, Nikon) equipped with an acousto-optic tunable filter (HSi Hyperspectral Imaging Platform, Gooch and Housego) and a high-sensitivity camera (Cascade 512B, Photometrics) to acquire spectral image sets. MicroManager and ImageJ (open source) software were used to synchronize spectral image acquisition, while post-acquisition spectral analysis was performed using ENVI (ITT Visual Information Systems), MATLAB (Mathworks), and Excel (Microsoft). Both unfixed (cryotome) and fixed (microtome) lung samples were investigated. Distinct spectral signatures within each sample were identified, isolated, and compiled into a spectral library. This spectral library was then used to identify the accumulations (relative intensities) of these signatures within lung tissue samples using classification and linear unmixing methods.

We will present the results from our initial studies of unfixed and fixed lung samples, including methods for acquiring spectral image data, identifying autofluorescence spectra of the pulmonary vasculature, and filtering the fluorescent protein spectrum from the surrounding autofluorescence. Successful results in quantifying fluorescent protein expression will hold great clinical importance, as these methods will allow assessment of cell-signaling agent levels in animal models for pulmonary edema and ARDS. These results are also important for extending approaches that work in in vitro cell culture lines, such as protein-based biosensors, to studies in intact tissues and in vivo assays.