TL;DR
This paper presents a detailed protocol for SUM-PAINT, a super-resolution microscopy technique that allows for the simultaneous imaging of dozens of proteins within a single neuron at the single-protein level. This enables the creation of detailed "neuronal atlases" that can be analyzed with machine learning to understand complex cellular structures and their heterogeneity.
What Problem Does This Paper Address?
Understanding the intricate organization of proteins within cells is crucial for deciphering biological processes. Traditional microscopy techniques struggle to resolve the sheer number of proteins involved in complex cellular structures like neurons, especially at the single-protein level. This limitation hinders our ability to map neuronal architecture and understand how variations in protein organization contribute to neuronal function and dysfunction. The authors aim to overcome this by providing a robust and detailed protocol for SUM-PAINT, a method that pushes the boundaries of multiplexed protein imaging.
Methods
The core of this protocol revolves around SUM-PAINT (Secondary label-based Unlimited Multiplexed DNA-PAINT). This technique builds upon DNA-PAINT, a super-resolution method that uses DNA hybridization for imaging. SUM-PAINT ingeniously employs a series of barcoding rounds, where different sets of antibodies are sequentially attached to target proteins. Each antibody is then labeled with a unique DNA strand (a "barcode"). These barcodes are then detected using complementary DNA strands ("imager strands") that are fluorescently labeled.
Key methodological aspects include:
- Neuronal Culture and Preparation: Detailed steps for preparing primary rat hippocampal neuron cultures, including options for "sandwich" and "mixed" cultures.
- Antibody Labeling and Barcoding: A multi-round approach where antibodies are first incubated with secondary nanobodies, which are then labeled with DNA barcodes. This allows for the sequential imaging of many targets.
- Signal Extinction: A crucial step where previously imaged DNA barcodes are "silenced" or blocked, preventing them from being detected in subsequent rounds and thus avoiding crosstalk. This is achieved through hybridization with complementary blocking strands.
- DNA-PAINT Imaging: The actual super-resolution imaging using the imager strands, which bind to the DNA barcodes. This is performed sequentially for each barcoding round.
- Image Reconstruction and Analysis: The protocol outlines the use of software like Picasso for fitting localization data, correcting for drift, and aligning images from different protein channels to reconstruct a comprehensive neuronal atlas.
- Synapse Segmentation and Feature Extraction: Advanced image analysis techniques, including DBSCAN clustering and UMAP dimensionality reduction, are used to segment synapses and extract morphometric features, enabling the classification of synaptic heterogeneity.
Datasets
Not discussed in the paper. The paper focuses on providing a protocol for generating data rather than analyzing pre-existing datasets.
Key Insights and Ideas
- Scalability of Multiplexing: The SUM-PAINT approach demonstrates a remarkable ability to scale up the number of simultaneously imaged proteins, moving beyond the limitations of traditional multiplexing techniques.
- Single-Protein Resolution for Complex Atlases: The protocol enables the generation of neuronal atlases with single-protein resolution, providing an unprecedented level of detail for studying cellular architecture.
- Machine Learning for Synaptic Heterogeneity: The integration of unsupervised machine learning (UMAP) for analyzing the extracted features is a powerful approach to uncover subtle differences and classifications within neuronal populations, particularly at the synapse level.
- Importance of Reagent Quality and Optimization: The protocol emphasizes the critical role of high-quality affinity reagents (antibodies and nanobodies) and the need for careful optimization of concentrations and incubation times to minimize crosstalk and ensure efficient labeling.
Key Results
The protocol enables the generation of neuronal atlases with up to 30 protein targets imaged simultaneously at single-protein resolution. The authors report a localization precision of approximately 6 nm. The analysis pipeline allows for the identification and classification of synaptic heterogeneity, revealing distinct clusters of synapses that were not anticipated based on traditional excitatory/inhibitory classifications. The protocol is designed to be robust, with detailed troubleshooting sections addressing common issues like image resolution, crosstalk, and alignment.
Limitations and Future Work
- Reagent Dependency: The success of SUM-PAINT is highly dependent on the quality and specificity of the affinity reagents. The protocol acknowledges that new reagents require thorough testing and optimization.
- Imaging Depth: The HILO illumination mode, while effective, has a limited imaging depth (approximately 5 µm from the coverslip) without significant resolution degradation.
- Imaging Time: Achieving high resolution requires sufficient sampling, which can lead to long imaging times, especially for high-plex experiments. The protocol suggests a balance between acquisition speed and resolution.
- Manual Intervention: While many steps are automated, some aspects like synapse selection and parameter optimization may require manual intervention and expertise.
- Future Work: The authors suggest that the protocol can be extended to image more than 30 targets by referring to specific sections. They also highlight the potential for further optimization of imaging conditions and the exploration of new dimensionality reduction algorithms for data analysis. The development of multiplexing blocker kits by companies like NanoTag Biotechnologies is also mentioned as a potential advancement.
Summary updated