Visualising Individual Molecular Features with Fluorescence Digital Molecular Imaging (DMI)

Visualising individual molecular features with fluorescence Digital Molecular

Imaging (DMI)

Mingjie Dai1,2, Ralf Jungmann1,3, Peng Yin*1,3

1. Wyss Institute, Harvard University, 2. Biophysics Program, Harvard University, 3. Department of Systems Biology, Harvard Medical School

* Corresponding author, email peng_yin@hms.harvard.edu
Abstract
Understanding individual molecular information from a large biomolecular complex, while maintaining its native environment, represents a key challenge in biology. Recent advances in fluorescence super-resolution microscopy have shown images of sub- cellular features and synthetic nanostructures down to ~15 nm in size, but direct optical observation of individual molecular-scale (~5 nm) features within a macromolecular complex (?Digital Molecular Imaging?, or DMI) has yet to be demonstrated. In this paper, we establish three technical requirements for meeting this challenge with localisation based microscopy, and demonstrate our ability to achieve DMI with DNA-PAINT (Point Accumulation for Imaging in Nanoscale Topography), a method which utilises programmable transient oligonucleotide hybridisation, on synthetic DNA nanostructures. In particular, we examined the effect of high photon count, high blinking statistics and appropriate blinking duty cycle on imaging quality, and reported fluorescence imaging of a densely packed triangular lattice pattern with 5 nm point-to-point distance. Using oligonucleotide conjugated small labelling agents, this imaging capability potentially provides a high-resolution, highly quantitative and single-molecule approach for studying structural heterogeneity, modifications and interaction networks towards a systems understanding of cellular biology with molecular-scale resolution.
Project summary
A key challenge for super-resolution fluorescence microscopy is to directly visualise individual molecular targets in situ, which are mostly around 5 nm in size (protein subunits, membrane receptors), and are in close proximity to each other. A method that addresses these challenges must:
(1) Provide ultra-high, ?molecular? resolution (~5 nm target-to-target distance)
(2) Visualise individual molecular targets (clear separation close targets)
(3) Operate from within compact native environment (dense cluster of targets)
Despite considerable recent development (PALM, STORM, STED, etc.), the limited imaging resolution of currently demonstrated methods (down to 15 nm) still failed to address this challenge and allow visualisation of closely packed and interacting molecular components.
We describe here a framework which enables stochastic localisation based super-resolution microscopy methods to achieve the above challenge, a scenario which we termed Digital Molecular Imaging (DMI), and present our result of the first super- resolution image achieving such quality on a synthetic DNA nanostructure template (5 nm point-to-point distance, each individual target is visualised, and within densely labelled grid pattern).
We propose a framework for achieving DMI of four technical requirements,

Summary Figure: Principle, requirements and demonstration of super- resolution Digital Molecular Imaging (DMI). (a) Concept of DMI. Top: schematics and point array description of a representative four- component biomolecular complex. Bottom: point array representation and schematics of scenarios (heterogeneity, modification, molecular interaction) that could be better studied with DMI. (b) Three technical requirements for achieving DMI, illustrated in 1D super-resolution. (1) Intensity profile, blue arrow indicates uncertainty of Gaussian fitted centre position. (2) localisation histogram, signal measures the peak-to-valley difference in 2-Gaussian fit, noise measures stochastic fluctuation of peak histogram count. (3) Localisation time trace, orange crosses indicate observed localisations, grey crosses indicate true molecular positions that are shadowed by false localisations. (c) DNA-PAINT demonstration of DMI with 5 nm lattice standard. (i) Schematic of 5 nm

4x6 triangular grid standard. (ii) A representative single-molecule 5 nm grid image with DMI quality. (iii) Automatic fitting of image in (ii). Overlaid crosses indicate Gaussian fitted centers (green) and regular grid fitted centers using the green crosses as targets (blue). Root mean square (RMS) deviation between the green and blue crosses is less than 0.7 nm in distance. (iv) More examples of 5 nm grid images. Scale bars: 10 nm.

and demonstrate their effects in silico on improving imaging quality. We show that well- controlled blinking kinetics supports high localisation precision (down to 1 nm), high image Signal-to-Noise Ratio (SNR), and low fraction of false localisations (<5%).
We show implementation of the above framework with DNA-PAINT method (Jungmann et al. Nature Methods, 2014), by exploiting the flexibility in blinking kinetics control offered by tuneable oligonucleotide hybridisation. We also develop a novel DNA nanostructure based drift correction algorithm that achieved <2 nm residual drift on a commercial microscope setup. We finally demonstrate the DMI imaging quality on a regular triangular pattern with 5 nm point-to-point distance constructed on a synthetic DNA nanostructure.

Understanding individual molecular information from a large biomolecular complex, while maintaining its native environment, represents a key challenge in biology. Recent advances in fluorescence super-resolution microscopy have shown images of sub-cellular features and synthetic nanostructures down to ~15 nm in size, but direct optical observation of individual molecular-scale (~5 nm) features within a macromolecular complex (“Digital Molecular Imaging”, or DMI) has yet to be demonstrated. In this paper, we establish three technical requirements for meeting this challenge with localisation based microscopy, and demonstrate our ability to achieve DMI with DNA-PAINT (Point Accumulation for Imaging in Nanoscale Topography), a method which utilises programmable transient oligonucleotide hybridisation, on synthetic DNA nanostructures. In particular, we examined the effect of high photon count, high blinking statistics and appropriate blinking duty cycle on imaging quality, and reported fluorescence imaging of a densely packed triangular lattice pattern with 5 nm point-to-point distance. Using oligonucleotide conjugated small labelling agents, this imaging capability potentially provides a high-resolution, highly quantitative and single-molecule approach for studying structural heterogeneity, modifications and interaction networks towards a systems understanding of cellular biology with molecular-scale resolution.

Project summary

A key challenge for super-resolution fluorescence microscopy is to directly visualise individual molecular targets in situ, which are mostly around 5 nm in size (protein subunits, membrane receptors), and are in close proximity to each other. A method that addresses these challenges must:

(1)  Provide ultra-high, “molecular” resolution (~5 nm target-to-target distance)

(2)  Visualise individual molecular targets (clear separation close targets)

(3)  Operate from within compact native environment (dense cluster of targets)

Despite considerable recent development (PALM, STORM, STED, etc.), the limited imaging resolution of currently demonstrated methods (down to 15 nm) still failed to address this challenge and allow visualisation of closely packed and interacting molecular components.

We describe here a framework which enables stochastic localisation based super-resolution microscopy methods to achieve the above challenge, a scenario which we termed Digital Molecular Imaging (DMI), and present our result of the first super-resolution image achieving such quality on a synthetic DNA nanostructure template (5 nm point-to-point distance, each individual target is visualised, and within densely labelled grid pattern).

We propose a framework for achieving DMI of four technical requirements, and demonstrate their effects in silico on improving imaging quality. We show that well-controlled blinking kinetics supports high localisation precision (down to 1 nm), high image Signal-to-Noise Ratio (SNR), and low fraction of false localisations (<5%).

We show implementation of the above framework with DNA-PAINT method (Jungmann et al. Nature Methods, 2014), by exploiting the flexibility in blinking kinetics control offered by tuneable oligonucleotide hybridisation. We also develop a novel DNA nanostructure based drift correction algorithm that achieved <2 nm residual drift on a commercial microscope setup. We finally demonstrate the DMI imaging quality on a regular triangular pattern with 5 nm point-to-point distance constructed on a synthetic DNA nanostructure.