Combinatorial Labeling And Spectral Imaging, (CLASI): A Method To Greatly Expand The Number of Distinguishable Fluorescent Labels in a Single Image

The number of fluorescent proteins, organic fluorophores, and inorganic fluorescent biomarkers is ever increasing. However, the ability to unambiguously distinguish more than a few different labels in a single fluorescence image is severely hampered by the excitation cross-talk and signal bleed-through of fluorophores with highly overlapping excitation and emission spectra. Here, we report the development of a fluorescence labeling, imaging, and analysis method to greatly expand the number of identifiable labels in a single image. The CLASI method involves labeling targets with specific combinations of fluorophore reporters. Commercially available microscopes with spectral detection capabilities are used to image the combinatorially-labeled specimens. Novel computational algorithms are used to analyze spectrally-recorded image data. We have developed a linear unmixing algorithm constrained to identify specific combinations of fluorophores. Our novel algorithms allow the concatenation of spectral data acquired with several different excitation wavelengths, either in parallel or sequentially. A goodness-of-fit is reported for each spectral combination, either in every pixel or for every particle identified in the image. We have applied the CLASI method to the study of the composition and spatial arrangement of complex microbial communities. Using fluorescence in situ hybridization with oligonucleotide probes specific for 16S rRNA sequences, we demonstrate that we can distinguish 120 differently labeled microbes in a mixture using binary combinations of 16 fluorophores.