Visualizing organization of RNAs and proteins in cells using advanced optical imaging methods

Abstract

The cytoplasm of a cell is crowded with different kinds of macromolecules, of which RNAs and proteins are two major forms. A cell can have more than 1 million protein molecules and hundreds of RNA molecules per μm3; despite the crowdedness, molecules inside of a cell are highly organized, and the proper organization and localization of these macromolecules is key to many biological functions. Protein imaging by immunofluorescence or protein tagging methods and RNA imaging by in situ hybridization, in situ sequencing, or RNA tagging methods offers the molecular specificity to visualize individual molecular species. However, traditionally, the study of intracellular organization of proteins and RNAs has been limited by both the optical resolution to resolve molecules due to the diffraction limit and the throughput to profile a large number of molecular species. In the first part of this thesis, we present recent advances in the multiplexed, error-robust fluorescence in situ hybridization (MERFISH) method for RNA detection to improve its imaging throughput, spatial resolution, and capacity for additional protein detection. With these improvements, we constructed a high-resolution spatial atlas of transcriptome (up to 4,209 genes measured) inside individual neurons. In the latter part of this thesis, we describe a super-resolution imaging study, by STORM, of roles of actin-spectrin-based Membrane-associated Periodic Skeleton (MPS), in retrograde signaling that is critical for trophic-deprivation-induced axon degeneration. Specifically, Chapter 1 introduces current needs for spatially resolved transcriptome detection. Specifically, it details the concepts of single-molecule fluorescence in situ hybridization (smFISH) and multiplexed error-robust fluorescence in situ hybridization (MERFISH), and the experimental implementations. Chapter 2 presents several technical advances for increasing the imaging throughput of MERFISH, in terms of the number of cells imaged per experiment, by two orders of magnitude. This is achieved through a combination of improvements, including using chemical cleavage instead of photobleaching to remove fluorescent signals between consecutive rounds of smFISH imaging, increasing the imaging field of view, and using multicolor imaging. With these improvements, we performed RNA profiling in more than 100,000 human cells, with as many as 40,000 cells measured in a single 18-h measurement. Chapter 3 describes an approach that combines MERFISH and expansion microscopy to overcome the molecular crowding problem and thus substantially increase the total density of RNAs that can be measured. Using this approach, we demonstrate accurate identification and counting of RNAs, with a near 100% detection efficiency, in a ~130-RNA library composed of many high-abundance RNAs, the total density of which is more than 10 fold higher than previously reported. This is a near 7-fold increase in detection efficiency compared to imaging the same set of genes with the same number of hybridizations without expansion. In parallel, we demonstrate the combination of MERFISH with immunofluorescence in expanded samples to visualize proteins of interest in the same cell. Chapter 4 details our efforts in constructing intracellular transcriptome atlas inside individual neurons. We performed super-resolution spatial profiling of RNAs inside individual neurons at the genome scale using multiplexed error-robust fluorescence in situ hybridization (MERFISH), and mapped the spatial organization of up to ~4,200 RNA species (genes) across multiple length scales, ranging from sub-micrometer to millimeters, in thousands of single intact neurons. Our data generated a quantitative intra-neuronal atlas of RNAs with distinct transcriptome compositions in somata, dendrites, and axons, and revealed diverse sub-dendritic distribution patterns of RNAs. Moreover, our spatial analysis identified distinct groups of genes exhibiting specific spatial clustering of transcripts at the sub-micrometer scale that were dependent on protein synthesis and differentially dependent on synaptic activity. Overall, these data provide a rich resource for characterizing the subcellular organization of the transcriptome in neurons with high spatial resolution. Along a different track, Chapter 5 focuses on the behavior of a protein structure, actin-spectrin-based Membrane-associated Periodic Skeleton (MPS), and effects of actin and spectrin manipulations in sensory axon degeneration. We show that trophic deprivation (TD) of mouse sensory neurons causes a rapid disassembly of the axonal MPS, which occurs prior to protein loss and independently of caspase activation. Actin destabilization initiates TD-related retrograde signaling needed for degeneration; actin stabilization prevents MPS disassembly and retrograde signaling during TD. Depletion of βII-spectrin, a key component of the MPS, suppresses retrograde signaling and protects axons against degeneration. These data demonstrate structural plasticity of the MPS and suggest its potential role in early steps of axon degeneration. Chapter 6 concludes the thesis by offering brief discussion about the challenges and opportunities for image-based study of RNAs and proteins.