A Single-Objective Light-Sheet Microscope with 200 nm-Scale Resolution.

Etai Sapoznik, Bo-Jui Chang, Robert J. Ju, Erik S. Welf, David Broadbent, Alexandre F. Carisey, Samantha J. Stehbens, Kyung-min Lee, Arnaldo Marín, Ariella B. Hanker, Jens C. Schmidt, Carlos L. Arteaga, Bin Yang, Rory Kruithoff, Doug P. Shepherd, Alfred Millett-Sikking, Andrew G. York, Kevin M. Dean, Reto Paul Fiolka

Posted on: 27 May 2020

Preprint posted on 9 April 2020

Article now published in eLife at http://dx.doi.org/10.7554/eLife.57681

Configuring high-resolution light sheet imaging to use one objective brings fast volumetric imaging to any sample format.

Selected by Tim Fessenden

Categories: bioengineering, cell biology, immunology

Context:

The most significant recent advance in capturing cell and tissue behaviors in real time is undoubtedly light sheet fluorescence microscopy (LSFM). It offers substantially increased image acquisition speeds while delivering less excitation light, at minimal reduction in resolution compared with conventional confocal imaging. These improvements are made possible by forming the excitation light into a thin plane and moving it through the sample, instead of the confocal approach of illuminating the entire volume and excluding out of focus emission light. This is all fine in theory but it demands that two objectives are oriented at 90 degrees: the light sheet illuminates a plane that must be detected orthogonally to capture the resulting fluorescence en face. This microscope configuration makes enormous demands on sample size and preparation, so that it fits in a space of a few millimeters between two objectives. Thus LSFM is denied several decades of optimization in sample preparation and environmental controls developed for confocal microscopes.

However, just as many confocal setups use one objective to pass excitation light to the sample and detect the emitted fluorescence, so may a light sheet microscope. No laws of physics prevent this setup from working, and such instruments (dubbed oblique plane microscopes (OPM)) have been described previously¹. The challenge is this: because the lightsheet intersects the sample at a sharp angle, the resulting emission light passes back through the optics skewed and spread out in 3 dimensions (imagine looking at an image on your phone when it is held at a 45 angle from your eyes). The emitted light must be mapped back into a 2D flat format for detection. This unavoidable step makes demands of conventional microscope parts that almost always degrade the resulting image. The solution described in this work preserves image quality using a cleverly designed chunk of glass.

Data:

The central advance defining this instrument is a new objective designed to be used in a single objective LSFM². Named Snouty after its odd tip, it sits downstream from the sample and is the 3^rd objective encountered by emission light from the sample. It faces the second objective at an angle to correct incoming light (skewed in 3D) into a 2D image for detection by a camera. Previous examples of single-objective light sheet instruments accomplished this step by aligning two conventional objectives pointing at each other³. However, the tips of the opposing objectives physically hindered a configuration that would pass all incoming light from one to the next. Thus, significant light was lost at this step. Enter Snouty, atop of which sits a beveled piece of glass. The glass tip captures much more incoming light from the opposite objective thanks to its shape and refractive index. Snouty opens the door for adjustments to all other components of the instrument to suit the needs of various biological specimens. The present work reports observations of living cells and fixed tissues made possible by Snouty in conjunction with components for fast, high-resolution imaging. Of the several imaging trials described, a few that stood out to this reader are described here.

The authors show time-lapse imaging of a Natural Killer (NK) cell engaging a K562 tumor cell, each expressing a different membrane marker to distinguish them. The NK cell makes contact with its target, and expands its contact area as waves of actin-rich ruffles, labeled here by Lifeact, flow backwards from the contact surface. In doing so the NK cell draws small patches of the tumor cell’s membrane back, resulting in long tethers of the tumor cell membrane stretching over the body of the engaging NK cell. These tethers suggest that retrograde flow of NK adhesive receptors are coupled to patches of the target cell over several microns, and/or that low cortical tension of the target cell permits tether elongation.

tumor cell (cyan) engaged by a natural killer cell (red). Scale bar, 5 microns.

A significant capability of this instrument is photoactivation using a genetically encoded photoactivatable PI3K. By diverting beam path for the 488 nm laser through a point scanner and an additional galvanometer, this light can be conformed to arbitrary 2D shapes that permit photoactivation with simultaneous detection via light sheet using the remaining laser lines. Using this configuration, the authors report shape changes of the cell membrane in response to local PI3K activation with exquisite time and spatial resolution. The membrane’s response is predominantly ventral protrusion and even separation from the coverslip surface. This membrane activity extends several microns from the zone of photoactivation along the cell’s leading edge, implying the extent and timing of the self-propagating wave of PI3K activity initiated by photoactivation.

In addition to these the authors report several separate imaging experiments that demonstrate the robust capabilities of their instrument. They focus on imaging setups not compatible (or extremely difficult) with dual-objective LSFM, such as micropatterned structures that control cell geometry to capture cell responses to confinement and nuclear deformation.

Implications:

The microscope configurations to which biologists are accustomed represent only a fraction of the potential engineering solutions to collect and reconstruct fluorescent light from a biological specimen. The instrument described here should encourage more nonexpert microscopy users to question the parameters for good imaging that we have inherited from our forebears. Why, for instance, should the entirety of a sample be flooded with laser light only to block almost all of it? Why should translocating the image plane require moving the entire microscope stage? This work presents one very specific engineering solution, but it encourages a broader shift in considering what microscope components do for the user and what they might do better.

Imaging instrumentation has a tendency to steer biology towards those questions that make full use of its capabilities. Witness, for example, the benefits of LSFM advances that accrue largely to developmental biology using zebrafish embryos. To take an extreme counter example of engineering around a biological question, consider the “gravity-scope” built in the Prakash lab for imaging vertically migrating plankton⁴. Hopefully the present work will encourage the push for new and different imaging tools that foster deeper questioning across subfields of biology.

Special thanks to Jeffrey Kuhn of MIT for help with this post.

References

Kim, J. et al. Oblique-plane single-molecule localization microscopy for tissues and small intact animals. Nat. Methods 16, 853–857 (2019).
Millett-Sikking, A. & York, A. AndrewGYork/high_na_single_objective_lightsheet: Work-in-progress. (2019) doi:10.5281/ZENODO.3244420.
Yang, B. et al. Epi-illumination SPIM for volumetric imaging with high spatial-temporal resolution. Nat. Methods 16, 501–504 (2019).
Krishnamurthy, D. et al. Scale-free Vertical Tracking Microscopy: Towards Bridging Scales in Biological Oceanography. http://biorxiv.org/lookup/doi/10.1101/610246 (2019) doi:10.1101/610246.

doi: https://doi.org/10.1242/prelights.21204

Read preprint

(2 votes)

Author's response

Kevin Dean, Reto Fiolka shared

Since this system obviates the particular drawback of tightly constrained sample mounting in standard LSFM, it may be a particularly exciting system for users whose samples are necessarily larger. Such users might also value imaging depth. How would you further modify the system for improving imaging depth?

Dean:

The primary objective can be swapped out so long as the remainder of the optical train has been designed to accommodate it. For example, Andrew York and Alfred Millett-Sikking have documented a variety of configurations, based upon your selection of the primary objective:

https://andrewgyork.github.io/high_na_single_objective_lightsheet/appendix.html#Optical_configs

Unfortunately, you cannot just switch the primary objective like it were a traditional confocal or widefield microscope at this point. The positioning of the primary objective relative to the downstream optics is more sensitive than most microscopes.

In my experience, it is incredibly hard to keep samples sterile in traditional light-sheet microscopes that use water dipping objectives. As a consequence, anytime you image a sample, it is effectively ruined. Now we can keep everything sterile in the long-term, and happy and in focus in the short-term because we now have the proper environmental control and autofocusing mechanisms. And yes, we can now image samples that otherwise cannot fit between two orthogonally mounted objectives, and this opens up new opportunities. For example, we could build an upright microscope designed to perform oblique plane microscopy superficially within a head-mounted mouse.

Fiolka:

This microscope was really intended to push the resolution, so we chose one of the highest NA objectives. The primary lens performs well, but you lose some of its performance in the OPM optical train and with Snouty. I think it was a worthy attempt to try to squeeze as much out of the OPM concept as possible. I am happy with the ~280nm in one lateral dimension, but the 320 nm in the scan direction could be better. We still do not know if that is a fundamental limit, or if we did something sub-optimal with the alignment.

Which brings me to optical depth. This is limited by several factors, but the least by the actual working distance of the primary objective. It is more limited by the remote focusing, which only remains aberration free over a small axial distance for this high NA. In a straight configuration of Snouty, it would cover about 40 microns. However in the tilted configuration, it shrinks. I think practically, you can reach maybe 20 microns in z into the sample. Thus we really stick with adherent cells. There would be the possibility of axial tiling (focusing on a higher plane) with the primary objective, but we have not done this.