Skip to main content

Optical nanoscopy at 1 nm resolution: far-field fluorescence control at cryogenic temperatures

Periodic Reporting for period 4 - OptnanoATcryo (Optical nanoscopy at 1 nm resolution: far-field fluorescence control at cryogenic temperatures)

Reporting period: 2020-01-01 to 2020-06-30

Optical nanoscopy is a powerful technique used in biology to study subcellular structures and functions via specifically targeted fluorescent labels. Localization microscopy in particular offers a much better resolution (~10-50 nm) than conventional microscopy (~250 nm) while being relatively undemanding on the experimental setup and the subsequent image analysis. For these developments the Nobel Prize in Chemistry 2014 has been awarded.
The increase in resolution from traditional imaging to super-resolution has been tremendous but is still short of imaging larger biomolecules or structure of biomolecules assemblies. Cell biologists want to understand the function of proteins and sub-cellular machinery such as DNA at the molecular level. In order to do so, another increase of resolution with light microscopy is needed. Moreover, this improvement must be realized in 3D
In this project we propose such an increase, to be precise: The next revolution in imaging to 1 nm isotropic resolution in 3D. To go from the current state of the art towards this goal we must realize a big increase in the number of collected photons from single fluorescent emitters as well as in the labelling density. Only then can subcellular structures be imaged at the molecular level to study the molecular machinery of the cell. In a larger perspective, the outcome of this research will enable the combination of structural cryo-electron microscopy imaging at subnanometer resolutions with functional fluorescent imaging at the nanometer scale.
The overall objective is twofold: 1) to build and realize a localization based super-resolution microscope and image paradigm at cryogenic temperatures and 2) enable ultra-resolution close to 1 nm by data fusion algorithms on identical bio macromolecules.
We have completed the construction of a cryotank for 12h measurements with optical microscopy. It is based on the design from Joerg Enderlein in Goettingen, but with several improvements such as longer measurement times, possibility to move and position the sample with piezo motors and image from the side on the optical table for improved vibrational stability and axial drift. We have started designed a cryo-stage together with an industrial collaborator that could allow us a similar cooling capacity as the cryo tank filled with liquid nitrogen, but is much smaller/lighter and allows directly for dip/tilt xy-stage control of the sample without complicated constructions. In addition, imaging from both sides of the sample should be possible then
We already investigated several red dyes for imaging at cryogenic temperatures.
With regards to the polarization control have we completed the calibration setup for the excitation and STED light paths. We have summarized the technicalities of the setup and calibration protocol in a manuscript that is accepted. This is the basis for the polarization controlled induced sparsity. We have completed experiments at room temperature with fixed dipole emitters. Measurements at cryogenic temperature with fixed emitters have been performed. We introduced a STED beam for depletion with the perpendicular orientation than the excitation beam, however, we found that we do shelf too many molecules too quickly in a long-lived dark state to achieve the required sparsity at the moment.
We have developed phase plate (Vortex PSF) which has the unique capability of simultaneously estimating the 3D position of the emitter, as well as the polar and azimuthal of the molecular orientation, and the degree of orientational constraint, all from a single 2D image. A manuscript is nearly finished at the time of reporting.
We have developed an information optimal algorithm for data fusion in super-resolution light microscopy. From several talks at conferences about the topic we received very positive feedback. Several leading groups are interested in collaboration with us. We have published on our data fusion pipeline in a high impact journal and given several talk on international conferences on it. The resolution achieved by the data fusion approach is ~3-4 nm in 2D.
From this we have embarked on a journey to extend the method to full 3D imaging. We have a manuscript on bioarchive, that is under review at the time of reporting. Here Nuclear Pore Complexes from three different imaging systems and methods of the same macromolecular complex have been reconstructed by our method. We have been able to infer symmetry from the data itself and use it to promote the reconstruction with it.
We have further started to investigate if and how different class can be found from the data alone before averaging. To this end we introduced a suitable dissimilarity measure between particles and used unsupervised clustering after multi-dimensional scaling. This result is again only in the early phase and a manuscript is in preparation at the time of reporting.
For dissemination we provide code on all our methods on github and the version at the time of submission on our ftp site (and journal page). The data for our publication is stored on 4TU servers (an initiative of the Dutch research community) that allows data storage in a findable manner for years to come.
Over the course of the project we have had several findings beyond the state of the art. The data fusion algorithm presents a unique approach to data fusion that has been introduced to the field of fluorescent microscopy. We believe that his approach will be the basis for developments and biological findings. The current concept by just applying algorithms from the field of cryo-EM SPA will become redundant as we can show that a clear better performance can be achieved with our approach. We have presented a 2D approach in a high impact journal. The 3D approach is on available bioarchive and the software is available (GPU accelerated). We have shown 3D reconstructions of Nuclear Pore Complexes with unprecedented resolution in 3D. On the methodical side we made steps towards inferring rotational symmetry of a structure from the data – without prior knowledge. Here we can find the full symmetry group and the different rotation axis. Then we could use this obtained knowledge to start promoting the alignment of the particles with this symmetry restriction. This resulted in a clearly improved reconstruction. Inferring symmetry from the data and obtaining the full group and rotation axis is a major step which will be used in the future.
On the experimental side we have implemented a polarization control system that is significantly better than prior literature as it allows controlling the (linear) polarization on the sample for 2 different wavelengths. We have shown very good modulation depth for a polarization sweep and our superior setup and calibration is to our mind the key why we have this much improved modulation depth compared to the prior art. We have build a TIRF cryo imaging platform based on a waveguide. This waveguide technology has just 2020 been shown to improve structured illumination microscopy (SIM) at room temperatures for a very large field of view, important for high content imaging. We have developed the technology to do this at cryogenic temperatures. The design of a Vortex PSF will make 3D estimation of position for fixed dipole emitter much easier (and cheaper), which is a requirement of cryo imaging where the emitters are frozen and therefor fixed.
Image of result for publication Heydarian 2018 Nature Methods
Still of animation movie over light microscopy in Delft
inside cover of publication for Small Methods