RainBar: Optical Barcoding for Pooled Live-Cell Imaging with Single-Cell Resolution

This is the authors' abstract. We don't add key points for this paper.

High-throughput pooled screening has advanced functional genomics, but most existing methods rely on endpoint sequencing and are blind to dynamic, time-resolved phenotypes. We developed RainBar (Rainbow Barcodes), an optical barcoding system that supports pooled live-cell imaging with single-cell resolution. RainBar uses lentiviral co-delivery of spectrally distinct nuclear and cytoplasmic fluorescent proteins to encode up to 64 unique perturbations per well. To mitigate barcode recombination and improve decoding accuracy, we employed single-template viral production, codon diversification, and a ratio-based spectral unmixing pipeline tailored to overlapping fluorophores. An inverted cytoplasmic segmentation approach and multilayer perceptron classifier enabled accurate barcode identification in both arrayed and pooled formats. As a proof of concept, we applied RainBar to dissect NF-κB signaling dynamics in epithelial cells. Live imaging of RelA translocation uncovered stimulus-specific kinetics: IL-1β triggered rapid recovery, while TNF induced prolonged nuclear localization. In pooled CRISPRi screens, RainBar recovered known NF-κB regulators (e.g., IL1R1, MYD88, TNFRSF1A) and highlighted additional modulators, including the Ino80 chromatin remodeling complex subunits and KAT2A acetyltransferase. Together, these results position RainBar as a flexible platform for multiplexed, image-based functional genomics, with potential to reveal dynamic signaling architectures across diverse cellular contexts in live cells.