Highly multiplexed immunofluorescence technologies provide a powerful tool to study biological complexity. Simultaneous detection of over ten biomarkers can reveal the molecular architecture and biomolecular interactions in normal cell physiology and disease states. However, achieving this level of multiplexing on a single specimen using fluorescence can be challenging. Spectral overlap between fluorophores, sample heterogeneity, and the underlying complexity, especially for clinical specimens, pose difficulties for current linear unmixing and blind source separation techniques. These techniques also require a large number of spectral measurements and reference spectra from all sub-regions of interest, making the process time-consuming, complicated, and sometimes impossible.
In a recent publication in Nature Communications, Junyoung Seo et al. describe a novel algorithm called PICASSO, Process of ultra-multiplexed Imaging of biomoleCules viA the unmixing of the Signals of Spectrally Overlapping fluorophores. The method relies on blind unmixing of spectrally overlapping fluorophores without reference spectra, and limits the number of images to only one per fluorophore. The algorithm unmixes images from overlapping fluorophores by progressively minimizing the mutual information (MI) between mixed images. The smaller number of images provides advantages for faster image acquisition, minimizing the number of required detection channels, and allowing a higher signal-to-noise ratio. PICASSO could reliably unmix overlapping fluorophores with only an 8-nm separation between their emission spectra within the same detection channel, as well as unmix the signals of spectrally overlapping fluorescent proteins and organic fluorophores. The technique could also be used to separate autofluorescence from a specific signal without a separate autofluorescence measurement. By performing PICASSO using preformed primary antibody–Fab complexes to circumvent the limited availability of different host primary antibodies, the authors demonstrated 15-color multiplexed imaging of the mouse hippocampus in a single round of staining and imaging. Spectrally overlapping fluorophores and dyes with large Stokes shifts, including CF®405S, CF®405M, CF®405L, CF®568, CF®633, CF®660R and CF®680R, were used to achieve higher multiplexing and imaging of spatially overlapping targets.
The authors demonstrated how PICASSO can improve multiplexing capabilities for a variety of applications including mRNA FISH, co-detection of proteins and mRNA, super-resolution imaging, cyclic immunofluorescence, and 3D imaging of thick tissue slices. Because the method relies not on special spectral detectors, but on an image analysis algorithm, it can also be used with simple emission filter-based microscopy without spectral imaging capability. PICASSO could be a useful tool for a broad range of applications providing deep spatial and molecular phenotyping with far-reaching implications in diverse fields including immunooncology, immunotherapy, cancer, and neuroscience research.
Seo, J., Sim, Y., Kim, J., Kim, H., Cho, I., Yoon, Y., & Chang, J.-B. (2022). PICASSO: Ultra-multiplexed fluorescence imaging of biomolecules through single-round imaging and blind source unmixing. Nat. Commun. https://doi.org/10.1038/s41467-022-30168-z