Liquitopy^® – addressing a new paradigm in optical microscopy

Nowadays, despite the limitations due to physical laws like diffraction, the optical microscope allows to form images with molecular contrast and nanoscale spatial resolution. A curved piece of glass and visible light, like the one you can experience watching a rainbow, make this possible. It is not exaggerate to state that today the optical microscope is endowed of unlimited spatial resolution. The question posed by Giuliano Toraldo di Francia, late in the '40s, about the possibility of getting super-resolution finds its answer in the Nobel Prize in Chemistry 2014 that was awarded jointly to Eric Betzig, Stefan W. Hell and William E. Moerner “for the development of super-resolved fluorescence microscopy”. This has been done under the umbrella offered by the seminal work of Giuliano Toraldo di Francia without violating physical laws by simply adding additional information in the image formation process. “Two-point resolution is impossible unless the observer has a priori an infinite amount of information about the object” was the inspiring statement. Now, considering the current optical microscopy scenario it is self-evident that the approaches belong to two main categories, namely: optics-based approaches and probe-based approaches. Optics approaches are related to the development and implementation of microscopes that are able to improve spatial resolution conditions 7/8 times using different photon detection or illumination strategies or lens positioning and computational steps. Probes approaches introduce a deep knowledge about the probe being used, mainly fluorescent molecules, and take advantage of the possibility of controlling the photon emission conditions in space and time. This second case, probe approach, is the one able to push towards “unlimited” spatial resolution. In some cases, both approaches introduce non-linearities in the different steps related to the image formation and make capital of additional information that allows to circumvent the diffraction limit.

Figure 1 is the heart of LIQUITOPY^®, liquid tunable microscopy, that can be developed under the physical architecture of a “multi-messenger microscope”. A multi-messenger microscope can be associated with the modern idea of multi-modal microscopy and it can be more than this.

One of the major revolutions in optical microscopy is connected with the use of beam scanning systems to illuminate the specimen. The knowledge of the position of the beam is an additional information, with respect to the optical image formation process through the lens, that allows to get improved resolution. Confocal laser scanning microscopy, image scanning microscopy (ISM) and multi-photon excitation (MPE) microscopy, for example, allow to form images endowed with a super-resolved spatial resolution content. Illumination beam scanning introduces, point by point, the additional knowledge of the position of the probe in the image formation process. The use of fluorescence, under linear and nonlinear modality of excitation, allows to operate under favourable signal-to-noise ratio conditions –dark background and bright signal– coupled to a biochemical affinity of the fluorescent probe for the specific biological compartment under examination. This is the key point for making fluorescence microscopy an indispensable tool in cell biology: it is uniquely specific with regard to the objects to be mapped and visualized, it is largely non-invasive, it can probe the three-dimensional layers of the specimen at ambient conditions and enables spectroscopic diagnosis with high (bio)chemical sensitivity. Within the “optics approach” branch the improvement in terms of resolution is always hampered by the use of the lens and one can escape only introducing some saturation aspects that can push resolution unlimited and are related to the control of the light-matter interactions involved.

Moving to the “probes approach” side one immediately has different chances to explore the specimen with molecular contrast and at the nanoscale spatial resolution level. Everything happens within the Toraldo di Francia “umbrella” considering that there is nothing about any possible violation of the physical laws. The possible additional information that one can use “to further extend” the cut-off frequency of the microscope, i.e. spatial resolution, lies in the ability of controlling features related to the image formation process. In fact, super-resolved fluorescence methods exploit the knowledge about the emitters’ position that is incremented with the further ability of controlling fluorescent states. It is the ability to control the states of fluorescent molecules (for instance: dark-bright; different wavelengths emission) that produces an advantage comparable to the ones available by introducing known objects or filters or by controlling the position of an illumination beam. This allows circumventing the Abbe’s “wall” that, nowadays, is smashed by a large variety of “super-resolution” techniques coupled with an improved ability to perform 4D (x-y-z-t) imaging. This increases the role of fluorescence microscopy as an indispensable tool in cell biology because of its unique advantages: it is a largely non-invasive technique, it can probe the deeper layers of a specimen at ambient conditions and enables spectroscopic diagnosis with chemical sensitivity. Now, it is worth noting that scientists were pushed to consider other parameters about fluorescence than intensity and to increase the portfolio of fluorescent labels. Up to now, fluorescence lifetime, polarization, photobleaching rate, spectral changes and blinking are used, with other fluorescence properties, to form images in microscopy. A further keystone step was the advent of Green Fluorescent Proteins (GFP): a fluorescent label self-expressed by the biological system when and where needed. Its impact increased when photoactivatable and photoswitchable visible fluorescent proteins appeared both under conventional and two-photon photoactivation. In order to access dynamics in biological systems at molecular level one can use different biophysical probe-based approaches, top left of fig. 1, namely: FRET (Förster Resonance Energy Transfer) to disclose interactions at molecular distances, FCS (Fluorescence Correlation Spectroscopy) to evaluate molecular concentration changes, FRAP (Fluorescence Recovery After Photobleaching) to evaluate diffusion coefficents in living cells, fluorescence lifetime to detect environmental and structural changes of target molecules, fluorescence polarization anisotropy to define mobility behaviour at molecule-molecule interaction level.

The route to super-resolved fluorescence methods develops through stochastic and targeted read-out gates. Coordinate stochastic super-resolved fluorescence microscopy is also referred to as the single molecule localization method. Localization-based techniques, exploiting photoactivation, photoconversion or ground-state depletion of fluorescent molecules, allow super-resolution imaging of biological samples. Several variants belong to this family of super-resolution techniques. The knowledge of a single emitter as unique source of photons, apart from background, allows one to localize it with nanometric precision. This method can be extended to thick biological objects like cell aggregates or tumor spheroids.

The improved performances in terms of signal-to-noise ratio and imaging speed provided by the new generation of CCDs cameras, like CMOS and EMCCDs, have made these techniques more and more popular and allowed the application of super-resolution imaging techniques based on single-molecule localization to a wide range of biological cases. Expansion microscopy (ExM) is a novel method that, upon a specific sample treatment, allows super-resolved fluorescence imaging with conventional microscopes. It has been demonstrated that by synthesizing a swellable polymer network cross-linked with fluorescent labels within a specimen, it can be physically expanded including the linked fluorescent molecules, resulting in physical magnification that allows to distinguish objects originally closer than the “forbidden” diffraction-limited distance. Typically, sample preparation consists in soaking the biological cells in a polymer, inducing the polymerization to form a dense meshwork throughout the cell that cross-links the fluorophores. After digestion of cellular protein and rehydration of the sample the image formation process can take place. The effect of swelling of the polymer gel leads to an N-fold isotropic stretching of the sample. The separation between objects that otherwise could not be appreciated now becomes visible. The utilization of such an approach is growing in the scientific community. Coordinate-targeted super-resolved fluorescence microscopy belongs to the class of beam scanning image formation methods. The most known approach is stimulated emission depletion (STED) microscopy. It uses the fundamental process of stimulated emission to engineer a Point Spread Function (PSF) of arbitrarily small size and circumvent the diffraction barrier limitation. The two most relevant technical features for the STED beam are the following aspects: 1) the STED beam wavelength should be set in a spectral region having a low probability of excitation, as far as possible from the red tail of the absorbtion spectrum of the fluorescent molecule being used; 2) the intensity distribution of the STED beam, typical a doughnut shape, has to drop to zero at least in the center of the beam. So, since the doughnut shape has to feature a zero-intensity point in the center of the beam with an intensity profile driven by the diffraction-limited point spread function of the focusing lens coupled to the power of the STED beam, the STED beam depletes the molecular fluorescent state everywhere within the diffraction-limited excitation volume, by means of stimulated emission, except at the center and its neighbourhood where the probability is very poor due to the cross-section at the STED beam wavelengths. The use of the fundamental process of stimulated emission makes STED microscopy compatible, at least in principle, with all fluorescent probes. Figure 2 shows the excitation-emission-depletion applied simultaneously to two fluorescent molecules in a high density populated cell nucleus. It is a matter of fact that in STED microscopy the increase of spatial resolution is directly related to the extent of depletion, i.e. the fraction of excited fluorophores that undergo stimulated emission during the illumination scanning process. The key point that in STED implementation, in some cases, the power of the STED beam cannot be increased to the level required for getting a certain resolution pushed reserchers to introduce original solutions for overcoming the problem. Among them, it is worth mentioning MINFLUX as the latest and SPLIT (Separation of Photons by LIfetime Tuning).

MINFLUX allows nanometer-resolution imaging and tracking of fluorescent molecules with minimal photon fluxes implementing an original illumination and detection concept merging nanometer-level molecule localization and nanoscopy. The time of arrival of photons is the core element for the SPLIT-STED approach shown in fig. 3. Photons are classified in terms of arrival time and the related emitter distance from the center of the STED beam is used, along with uncorrelated original background signal, to get an improved result with respect to the application of a mere gated-STED solution. This can also be applied under two-photon excitation (2PE) regime, particularly suitable for thick specimens. Within the developments related to superresolved fluorescence microscopy, there was a growing interest in detectors’ technology and in exploiting different parameters related to the collected photons originated by light-matter interactions. Nowadays, “time” is the keyword and “encoding and decoding space and time to form images” is one of the more interesting advances in the field for the future in optical microscopy. Image scanning microscopy is the core for the design and implementation of original multi-modal modern optical microscopes. Figure 4 shows the intrinsic property of such an approach for improving spatial resolution with a brand new implementation based on a SPAD (Single Photon Avalanche Diode) array made of 25 elements that also offers the possibility of collecting spectroscopic information during the image formation process, frame rate independent. Among the key elements one has an improved signal-to-noise ratio that allows to obtain the image formation process while mapping spectroscopic information like fluorescence lifetime as shown in fig. 5. It is known that image scanning microscopy can improve the effective spatial resolution of confocal microscopy and at the very same time “classical” implementations are not appropriate for mapping fluorescence lifetime in an image. Here, because the photons emitted from each specimen location are spread across all the elements of the SPAD array, the resulting effective dynamic range is higher than in the common case of single-element detectors. Such a multi-element detector can become the standard for versatile and multiparameter ISM, capable of converting any conventional point-scanning microscope into a super-resolution microscope endowed with spectroscopic abilities. In some cases, for example the ones related to high density of molecules towards the understanding of the delicate relationship between structure and function or the simple perspective of removing labels exploiting the intrinsic property of the sample in the light-matter interactions context, one would like to take advantage of label-free methods. Intrinsic fluorescence, second- and third-harmonic generation, phase contrast are some of the eligible mechanisms of contrast. However, one approach deserves attention since it can provide a set of images resulting from the analysis of changes in polarized light while interacting with the specimen. Classification, by means of a 4-element vector, of illumination and detected light in terms of polarization makes it possible to describe the behaviour of the specimen, including the chain of optical elements, by means of a matrix, known as Mueller matrix. Recently there is a growing use of the 16 elements of this matrix to distinguish in a phenomelogical way, for example, between healthy and pathological tissues. At cellular level it was demonstrated that a specific element related to the differential scattering of circularly right-vs.-left polarized light, CIDS (circular intensity differential scattering), can provide information about condensation or decondensation of chromatin-DNA in solution. More recently, a CIDS microscope has been implemented forming images based on CIDS of the organization of DNA in the cell nucleus. Figure 6 shows a comparison with the information available using fluorescence labelling of DNA according to the perspective of using label-free images instead of fluorescence ones. This is a very first step towards the integration of fluorescence microscopy taken under all possible modalities and polarization-based emission in a label-free context. Now, considering again the optical microscopy picture reported in fig. 1, one can envision a new kind of optical microscope capable of collecting data simultaneously through different image formation modalities that can be “easily” combined, just like liquids. Within the current scenario related to optical microscopy, the ability of integrating different light-matter interactions to form images in optical microscopy is the starting point for the realization of LIQUITOPY^®. In the super-resolution era, there is today a general consesus from the optical microscopy community about the fact that super-resolution is not always needed and that scientists have the need and the capacity of choosing a microscopy technique tuned on their specific application. So far, LIQUITOPY^® is connected to the design and implementation of a "liquid tunable microscope". It represents a new paradigm in data collection and image formation linked to the fact that, considering the illumination-specimen-detection chain, we have in our hands a real multi-messenger microscope in analogy to multi-messenger astrophysics, started about 30 years ago with multi-messenger observations from the Sun and cosmic neutrinos emitted by Supernova SN1987A. From Galilei’s observations to Airy disks and the CCD (charge coupled device) sensors, for example, it appears evident that microscopy and astrophysics have always been technologically and scientifically correlated. Here, the tunability comes from the fact that we have at our disposal a number of mature methods allowing to “tune” the microscope across a large, almost unlimited, range of spatial and temporal resolution, and at the very same time we can collect messages due to different light-matter interactions and carried by photons generated by means of different mechanisms from fluorescence to second-harmonic generation, from polarized light scattering to optical phase contrast. The multi-messenger microscope is liquid because it aims to overlap in an efficient and optimized way different mechanisms of contrast and it is tunable because it offers a real-time tunability regarding spatial and temporal resolution like a radio tuned on the preferred radio station. It cannot escape being intelligent since it can dynamically adapt its architecture to the availability of specific light-matter interaction signals. The concept regards the fact that, following the physical realization of a multimodal –multi-messenger– microscope that integrates “all” the optical methods, a series of images collected by means of n mechanisms of contrast, I (k), k=1, n, is used to generate an n+1 image, the liquid L image, reporting for each pixel a combined information, L=ℓ {I ( k ) k ∈ 1, n}. There are two key aspects to assign a “liquid” value, pixel per pixel. One is related to the objective definition of the achieved resolution of each mechanism of contrast evaluated by means of Fourier Ring Correlation (FRC) and the second takes adavantage of the utilization in microscopy of a deep learning artificial intelligence approach. A benchmark for LIQUITOPY^® is related to a biological framework dealing with chromatin organization in the cell nucleus at the interphase. The biological system offers a dense distribution of biological macromolecules and DNA is packed in chromatin domains. For this reason, label-free approaches play a key role in the image formation process and offer the chance to escape from further increase of density in the nucleus. It is a matter of fact that the nanoscale the dimension of labels counts. In the case of chromatin organization the high density of packing can be perturbed by the presence of fluorescence labels ranging from 300 Da up to 30 kDa. A multi-messenger offers to attack the biological question related to DNA organization and its influence on key processes in the cellular lifetime like transcription and replication using different mechanisms of label-free contrast, namely: i) intrinsic fluorescence of biological macromolecules within the super-resolved fluorescence microscopy scenario considering interchangeable encoding and decoding of space and time at the nanoscale from lifetime to correlation spectroscopy from pump-probe to single molecule behaviour; ii) label-free unlimited super-resolved microscopy in transmission exploiting the saturation of the absorption states of the molecules under investigation within a pump-probe temporal window, non-linear interactions including multiphoton excitation and high-order harmonic generation; iii) label-free microscopy based on Mueller and Jones matrix signatures coming from angular scattering processes and exploiting differential polarisation interactions and refractive-index mismatches in the VIS-IR regions, as shown in fig. 6.

The multi-messenger scenario is young and the LIQUITOPY^® concept constitutes a bridge between advanced optical microscopy methods and artificial intelligence implementations within a state-of-the-art technology that allows to introduce relevant improvements in the illumination-specimen-detection chain.

Acknowledgements

I would like to thank SIF (Società Italiana di Fisica) and Prof. Luisa Cifarelli for the chance of bringing to the attention of qualified readers an ongoing trend in optical microscopy. I am indebted to my group at Istituto Italiano di Tecnologia and Departement of Physics at the University of Genoa for the work together. A special thank to Mrs. Marcella Missiroli for her patience and her professional contribute in editing this article.