Fast, large-field fluorescence and second-harmonic generation imaging with a single-spinning disk two-photon microscope

. Confocal microscopes have been the workhorses of 3-D biological imaging, but they are slow, offer limited depth penetration and collect only ballistic photons. With their inefficient use of excitation photons they expose biological samples to an often intolerably high light burden. The speed limitation and photo-bleaching risk can be somewhat relaxed in a spinning-disk geometry, due to shorter pixel dwell times and rapid re-scans during image capture. Alternatively, light-sheet microscopes rapidly image large volumes of transparent or chemically cleared samples. Finally, with infrared excitation and efficient scattered-light collection, 2-photon microscopy allows deep-tissue imaging, but it remains slow. Here, we describe a new optical scheme that borrows the best from three different worlds: the speed and direct-view from a spinning-disk confocal, deep tissue-penetration and intrinsic optical sectioning from 2-photon excitation, and a large field of view and a low invasiveness of a selective-plane illumination microscope – all with a single objective lens. We validate the performance of our 2-photon spinning-disk microscope in various applications that have in common to simultaneously require a large depth penetration, high speed and larger fields of view. Beyond biological fluorescence, we demonstrate an application in material science, imaging coherent non-linear scattering from a 3-D nano-porous network.


Introduction
Confocal spinning disk microscopes excel in fast, directview, 3-D live-cell imaging but multi-photon variants of this multi-spot scanner have remained marginal.This is because: (i), the co-linear arrangement of microlenses and confocal pinholes on a tandem spinning disk is difficultly optimised for both infrared (IR) excitation and visible fluorescence so that spatial resolution must be traded against signal; (ii), for deep-tissue imaging, interference of IR excitation light emerging from multiple pinholes results in high-intensity nodes above and below the focal plane (Talbot effect), which invalides a fundamental assumption of 2-P imaging, that all generated fluorescence comes from a single focal plane; finally, (iii), in the classical Yokogawa design [1], the primary dichroic is located in the non-infinity space, between the two disks, inducing chromatic variations across the fieldview during scanning and prescribing the use of objectives having small pupil diameters, high magnification and low working distance.A variant in which the pinhole disk is omitted and only the microlens disk kept results in a loss of image contrast with depth, due to the increased collection of scattered fluorescence.Having these limitations in mind, we conceptualised, prototyped and characterised a new multi-photon multispot microscope geometry, in which the microlenses, pinholes and dielectric coating are all combined on a single, micro-machined disk [2].The expanded and flattened beam of a fs-pulsed IR laser beam is transmitted by the disk, focused by the microlenses and imaged onto the sample without pinhole filtering.2-P excited multi-spot fluorescence is collected by the same objective, acquires an optical path length difference compared to the excitation light on the way back to the detector, and it is focused to pass through the tiny pinholes etched in the dielectic coating on the rear side of the disk, Fig. 1.This optical layout permits the use low-mag, highnumerical aperture objectives [3], for example a Nikon ×25/1.1 water dipping lens, offering <400-nm xy-and <2-µm z resolution over a >200-µm field-of-view [2].We previously successfully used this microscope for fast 3-D imaging of brain organoids [2] as well as label-free second-harmony generation (SHG) and autofluorescence imaging across the entire thickness of the mouse intestinal wall, [4].Our multi-photon, multi-spot microscope permits a 3-to 4-fold deeper imaging than a 1-photon spinning disk confocal, it is 10 to 100 times faster than a point scanner, and it has a two-to 3-fold better axial sectioning than a typical light sheet microscope.

Imaging the dendritic tree of a Purkinje cell
Purkinje cells (PCs) are the principal output neurones of the cerebellar cortex.The have a large, flat and highly branched dendritic tree and a single long axon that forms an inhibitory projection to the deeper cerebellar nuclei.It has been difficult to record PC calcium (Ca 2+ ) activity and the summation of dendritic inputs throughout the entire arborisation of these large cells.In 200-µm thick sagittal cerebellar slices we filled a single PC with Oregon-Green-BAPTA488 fluorescent Ca 2+ indicator, using the whole-cell patch-clamp technique, and imaged its entire dendritic tree in less than 0.5 s.

Towards medical imaging of intrinsic signals
The enteric system is a quasi-autonomous nervous network of interconnected plexuses organized in a 3-D mesh, lining the gastro-intestinal tract.Our microscope allowed us the detailed label-free visualisation of enteric neurones and glia at much greater depths [4] and with better contrast than on a confocal microscope, illustra ting the interest for diagnostic imaging.

SHG hotspots in a 3-D metallic network
3-D nanoporous metallic networks [5] generate randomized hot spots at different frequencies and they are attractive substrates for photocatalysis, sensing or structured illumination microscopy.The OASIS microscope allowed us fast, large-field inspection of such 3-D structures by imaging the SHG signal generated in the nano holes and silver nano beads present in this volume.

Fig. 4 .
Fig. 4. Electron micrograph of a silver nanoporous network, left, and SHG hotspot generation, right, in a deep layer, imaged on the OASIS microscope, taken with a ×20 air lens.Image diagonal 400 µm, text = 480 ms.ex=850 nm, detection in the blue-green band.