Bringing super-resolution microscopy to the masses
Image: The Howard Research Group, University of Notre-Dame, US, creates open source tools to produce super-resolution, 3D images of cells and tissue in living organisms.
Earlier this year, researchers from the Howard Research Group, University of Notre-Dame, US, unveiled an open source tool called 'DeSOS' to produce super-resolution, 3D images of cells and tissue in living organisms.
Breathtakingly detailed videos of actin filament remodelling during spinal cord development in zebrafish demonstrate how the software augments traditional confocal and two-photon fluorescence images.
As Professor Scott Howard highlights: “The open-source application eliminates the expense of buying a super-resolution microscope, and is faster and has more functionality than such a microscope when evaluating living organisms.”
Images of zebrafish cells before and after DeSOS. [Howard Group]
DeSOS combines a blind deconvolution image processing method (De) with super-resolution, stepwise optical saturation microscopy (SOS), pioneered by Howard and colleagues.
The method identifies the differences between two fluorescence images to produce a single image with greater clarity, and unlike conventional super-resolution imaging, generates long-term time-lapse videos with relative ease.
“Many of our results have been coming out only in the past year or so,” says Howard. “Every day we're going into the lab and pulling out these amazing videos that I would have thought impossible to achieve even a year ago.”
DeSOS is the most recent of several super resolution imaging methods developed in the Howard Research Group to affordably stretch the limits of fluorescence microscopy.
Howard set up his lab at Notre Dame University in 2011, with a view to improving the resolution, sensitivity and speed of advanced imaging and microscopy techniques, especially those that could be used in vivo.
“We wanted to see the molecular micro-environment down to the cellular and sub-cellular level as so many unanswered biomedical questions exist here,” he says. “But many tools used here are limited by the physics of traditional microscopy, are sensitive to scattering effects or are very expensive.”
Given this, Howard and colleagues started to look at fluorescence lifetime microscopy (FLIM) and how exactly the decay rate of fluorophores depended on these molecules' surrounding micro-environment.
“There is so much rich information in this micro-environment,” points out Howard. “So we really went back to first principles and asked how do non-linear microscopes, such as multi-photon instruments, interact with fluorophores that are being quenched.”
A top-down FLIM microscope image of a 3D volume of mouse brain, imaged through the skull. The left is intensity only, the centre is a false colour image with lifetime mapped to hue. The green are microglial cells, the yellow is the plasma of the blood inside of blood vessels. The black streaks in the plasma of the blood are very fast-moving red blood cells in the blood vessels. The fact that a FLIM microscope can image these cells and resolve the fluorescence lifetime of the micro-environment in real-time and in 3D demonstrates the power of the system.
FLIM is particularly powerful when combined with multi-photon microscopy, providing 3D resolution, deep penetration and minimal phototoxicity to samples.
So, as their work continued, the researchers started to analyse harmonic distortions in multi-photon microscopy and also explore what happens to fluorophores during saturation with a view to using saturation to further localize fluorescence signals.
In 2016, they went onto develop an analytical model to describe the signal-to-noise ratio in a multi-photon fluorescence lifetime imaging system and developed super-sensitive, multiphoton frequency-domain fluorescence lifetime imaging microscopy (MPM-FD-FLIM), to image living tissue at twice the speed of a conventional instrument.
Crucially, as Howard points out, results were achieved through simple modifications to data analysis in the conventional MPM-FD-FLIM microscope.
From here, the researchers homed in on the issue of saturation in FLIM, designing algorithms to compensate for fluorophore saturation that could be easily applied to existing set-ups.
Come 2018, they also delivered saturation-based super-resolution fluorescence microscopy, based on Stepwise Optical Saturation (SOS). Here, raw fluorescence images are linearly combined to extend resolution beyond the diffraction limit.
“This opened up so many doors for our research as well as our collaborators, and through all our theoretical approaches, we could come up with a really deep understanding of the physical process behind the imaging,” says Howard.
The researchers soon extended their SOS concept to deliver Generalised Stepwise Optical Saturation for super-resolution FLIM and developed a novel two-photon frequency-domain FLIM system.
One of the most advanced fluorescent microscopes in the world, the system uses phase multiplexing to boost imaging speed and also includes adaptive optics to generate 3D lifetime images in the deep scattering tissues of living animals.
“We have the only microscope in the world that can image fluorescence lifetime multi-photon in-vivo samples in real-time,” highlights Howard. “And to our knowledge, we have achieved the first implementation of super-resolution imaging in frequency-domain fluorescence lifetime imaging microscopy.”
According to the researcher, the system is entirely custom-built and includes software written by himself and colleagues.
“This gives us the flexibility to, for example, switch objectives or add optics for dispersion compensation so we can carry out deep imaging through bone,” he says.
But as Howard emphasises, he and colleagues also work hard to ensure that their developments are translatable to other users, and as such, work on commercial Nikon and Olympus platforms, on campus.
“We have been developing algorithms and computer programs, and testing these on commercial instruments so researchers don't have to build their own platform,” he says. “We really have designed all of this so it can be manufactured as a low-cost add-on module to a commercial instrument so users also do not have to make a million-dollar investment.”
Right now, Howard and colleagues are busy collaborating with biologists, who are using their software and tools for their research.
"I'm trying to get as many researchers as possible to use our technology." Scott Howard, University of Notre-Dame.
As well as shedding light on actin dynamics in zebrafish, Howard's collaborators have also been looking at the molecular mechanisms that govern cell fate and matrix production in bone.
“Our collaborators are really helping us to make sure that our technology is useful,” he says. “We have distributed an open source app for DeSOS and are developing several apps and plugins, including a plugin for ImageJ as well as an app for really fast de-noising.”
“I'm just trying to get as many researchers as possible to use [our technology] and once they get results then this becomes a real partnership,” he adds. “I've been learning as much about neuroscience problems as I can, and collaborators are learning about the limits of their microscopes, and together we can then come up with new physics to solve the problems.”
So where next for the Howard Research Group? Howard and colleagues will continue to make sure their super-resolution fluorescence lifetime imaging is as accessible as possible and are also keen to use lower light levels for gentle, long-term animal imaging.
Comparison between Conventional and DeSOS Microscopy Images.
What's more, Howard is convinced that artificial intelligence will prove instrumental to future microscopes.
“Neural networks and machine learning are real hot topics and can be useful to help researchers understand and interpret noisy data,” he says. “We will work on this as well as continue to push our understanding of what's happening at the microscopic level so we can really squeeze as much information as possible from live animals and tissue.”