LITE microscopy: a technique for high numerical aperture, low photobleaching fluorescence imaging

Fluorescence microscopy is a powerful approach for studying sub-cellular dynamics at high spatiotemporal resolution; however, conventional fluorescence microscopy techniques are light-intensive and introduce unnecessary photodamage. Light sheet fluorescence microscopy (LSFM) mitigates these problems by selectively illuminating the focal plane of the detection objective using orthogonal excitation. Orthogonal excitation requires geometries that physically limit the detection objective numerical aperture (NA), thereby limiting both light-gathering efficiency (brightness) and native spatial resolution. We present a novel LSFM method: Lateral Interference Tilted Excitation (LITE), in which a tilted light sheet illuminates the detection objective focal plane without a sterically-limiting illumination scheme. LITE is thus compatible with any detection objective, including oil immersion, without an upper NA limit. LITE combines the low photodamage of LSFM with high resolution, high brightness, coverslip-based objectives. We demonstrate the utility of LITE for imaging animal, fungal, and plant model organisms over many hours at high spatiotemporal resolution.


Introduction
To properly visualize and measure cellular and subcellular dynamics, cell biologists demand imaging at high spatial and temporal resolution. The fluorescence microscope is a popular modern tool used to address these demands and solve cellular dynamics problems. However, conventional fluorescence microscope modalities require high intensity light to illuminate the sample through the objective lens, exciting all fluorophores in the path of the collimated excitation light. The fluorophores emit light that Fadero et al., 2017 2 is collected by the objective lens and transmitted to the detector. A disadvantage of the traditional "epi-illumination" geometry is that light is emitted from fluorophores outside the focal plane and contributes to the image, which confounds the focal information. Confocal microscopy mitigated this problem by selectively collecting light from the focal plane through the use of conjugate pinholes 1 . However, the reduction of out-of-focus fluorescence by confocal microscopy does not overcome the need for high-intensity illumination light that generates out-of-focus excitation events ( Fig. 1A; blue box). High intensity illumination transmits intense energy to the sample, damaging fluorophores that release reactive oxygen species upon photobleaching. Consequently, these reactive oxygen species chemically damage living samples through phototoxicity 2 .

Light Sheet Fluorescence Microscopy (LSFM, or Selective Plane Illumination
Microscopy SPIM) minimizes excitation-based photodamage by only partially illuminating the sample 3 . In the 15-year existence of modern LSFM, various implementations have arisen, most of which use two traditional objective lens elements arranged orthogonally in order to 1) illuminate the sample with a sheet of light and 2) align the detection focal plane with the illuminating sheet [3][4][5][6] . LSFM reduces or eliminates out-of-focus excitation, increasing the signal-to-background ratio (SBR) for fluorophores in the focal plane ( Fig.   1A; red box). This higher SBR allows detection of image features with lower excitation energy, thus reducing the photodamage incurred with conventional optical configurations.
These features allow the acquisition of a significantly larger number of exposures of a sample than any other mode of fluorescence microscopy. However, the orthogonal orientation of the illumination light sheet with respect to the detection objective generally requires that the sample be mounted at a minimum of one millimeter from the detection Fadero et al., 2017 3 objective, forcing use of low numerical aperture (NA, below 1.1) detection objective lenses. Therefore, the use of highly efficient, high-resolution oil-immersion objectives is incompatible with current LSFM regimes.
The detection of subcellular structures that drive cell biological processes including mitosis, endocytosis, and cytokinesis require high-NA detection objectives, due to their increased resolution and detection efficiency. Because 1.1 was the highest feasible NA detection objective used with traditional geometries 5 (Fig. 1B; green box), use of LSFM to study these sub-cellular structures with the traditional resolution or efficiency was not possible. Multi-view SPIM geometries have been able to accommodate a 1.2 NA waterimmersion objective to increase the resolution and detection efficiency of LSFM 6 ( Fig. 1B; yellow box); however, in order to approach the native resolution of oil-immersion objectives ( Fig. 1B; red box) traditionally used in cell biology, post-acquisition deconvolution was required. This data processing has high requirements for time, user expertise, specialized software, and data storage, which are currently inaccessible to the average cell biology laboratory. Accordingly, there existed a need to build upon the currently available designs for LSFM by combining selective illumination with conventional microscope stands and objective lenses that enable detection and resolution of subcellular structures and dynamics.
Here we present Lateral Interference Tilted Excitation (LITE) microscopy, which we developed in order to use high-NA, oil-immersion objective lenses to image samples illuminated by a light sheet (Fig. 1A, magenta box). We achieved this goal by using a tilted sheet that can access the working distance of high-NA oil-and water-immersion objective lenses, including a 60X 1.49 NA oil-immersion objective that accepts 88% more emitted Fadero et al., 2017 4 fluorescence and offers a 26% increase in native lateral resolution compared to a 25X 1.1 NA water-dipping objective 5 . LITE is compatible with traditional coverslip-based mounting conditions, meaning that LITE can be used with water-and oil-immersion objectives. The LITE method can also be implemented unobtrusively on most existing upright or inverted microscope systems, meaning high-resolution differential interference contrast (DIC) or other microscopic modalities can be used simultaneously (or in rapid succession) with LITE imaging. LITE images do not require computational reconstruction to view; the native images received from the camera are the data. In sum, LITE microscopy combines the low photodamage of LSFM with the high-NA objective lenses to allow high spatiotemporal imaging.

Methods
LITE is a novel method for introducing a light sheet within the working distance of high-NA objective lenses (Fig. 1A). Briefly, these goals were accomplished by first directing a collimated beam of excitation light through a photomask and cylindrical lens.
The cylindrical lens focused the excitation light to form a roughly "wedge-shaped" beam of light. The beam converged to its minimal thickness and formed the light sheet at the focal plane of the cylindrical lens, approximately three centimeters away from the cylindrical lens. The photomask was used to pattern the focusing beam so that the light sheet was lengthened 7 . To access the working distance of high-NA lenses, the excitation light was tilted such that the bottom of the converging "wedge" was parallel to the detection objective focal plane. Thus, the light sheet was formed at the focal volume of the detection objective, in which the fluorescent sample was mounted. LITE allows Fadero et al., 2017 5 mounting samples on coverslips, provided the chambers also have an optically clear opening to allow access by the converging illumination light. We have engineered several suitable chambers and present imaging data from a diverse range of model organisms.
We generated such a beam using a collimator illuminated by a laser combiner (Agilent Monolithic Laser Combiner 400, MLC 400) with an FC/APC fiber-coupled laser output of four wavelengths (405, 488, 561, and 650 nm). The four laser sources were solid state and pre-aligned to deliver a radially symmetric, coherent beam (Fig. S1). The maximum power outputs, after the fiber, of the four lasers in order of increasing wavelength were 18, 52, 55, and 37 mW, although only a fraction of each beam is used to generate the light sheet. The choice of illuminator should be based on specific application, fluorescent proteins in vivo in this case. An internal acousto-optical tunable filter (AOTF), analogcontrollable via DAQ Board interface, was used for modulating wavelength intensities.
For brevity, we mainly describe our setup as monochromatic illumination at 488 nm excitation (for EGFP).

Beam Conditioning
LITE illumination involves conditioning from the laser source such that the diameter of the radially symmetric beam is magnified to a value that is equal to or greater than the full aperture of the slits of a customized photomask (see below, Methods Part 3). The beam should remain collimated after conditioning. Here, collimation and beam expansion Fadero et al., 2017 6 were combined by an FC/APC-coupled (fiber connector/angled physical contact) TIRF (total internal reflection fluorescence, Nikon Instruments) microscopy collimator that achromatically collimated the lasers to a beam diameter of 22 mm (Fig. S1).

Photomask/Cylindrical Lens System
We used a cylindrical lens to focus a radially symmetric, collimated beam along one axis in order to approximate a non-diffracting "sheet" of light at the focus of the cylindrical lens. The sheet itself (in the focus of the cylindrical lens) can be approximately defined as a rectangular prism with three dimensions: the thinnest, diffraction-limited vertical width (w) that the converging laser reached at the cylindrical lens focal plane, the axial length (L) over which the laser remained at its diffraction-limited width before diverging, and the unfocused horizontal breadth (b) of the laser. The full width at half maximum intensity (FWHM) of the sheet (hereafter referred to as w) is defined by: In equation (1), n is the refractive index of the medium in which the laser was focused to a sheet (typically ~1.33 for aqueously media, although this value varies based on the temperature and chemical composition of the media, and the wavelength of the excitation light), λex is the wavelength of the excitation laser (in μm), and NAeff is the effective numerical aperture of the cylindrical lens. Note that NAeff can be smaller than the reported NA of the cylindrical lens, as NAeff depends on the percentage of the cylindrical lens NA that is used (i.e. the vertical height of the collimated excitation light Fadero et al., 2017 7 incident on the cylindrical lens back aperture). Thus, w is inversely proportional to the diameter of the collimated beam incident to the cylindrical lens, assuming the beam diameter is less than the full cylindrical lens back aperture. The thinnest sheet possible is traditionally preferable in LSFM, for two reasons: (1) to minimize out-of-focus excitation/emission in the fluorescent sample and (2) to prevent photodamage in out-offocus planes. However, the choice of sheet thickness in LITE was complicated by the mathematical interdependence of w and L, in equation (2): As shown in equation (2), it is evident that L increases with the square of w.
Practically, this meant that the thinnest sheet possible (minimal w) was not necessarily the best sheet for LITE, as the distance over which the sheet remains diffraction-limited In order to maximize the L for a given w, we placed a quadruple-slit photomask (FrontRange Photomask) in the principle plane of the cylindrical lens, before the beam enters the lens (Fig. S1). The theoretical and practical design of these slits were first described and implemented by Golub et al. in 2015. Briefly, this method increased L of a cylindrical lens-based light sheet beyond what equation (2) predicts by creating an Fadero et al., 2017 8 interference pattern at the cylindrical lens focal plane between two harmonic cosine waves 7 . Golub et al. (2015) presented the equation for the depth of field of the elongated light sheet below in equation (3): In equation (3), L' is the elongated sheet length, and R1 is the radius of the inner photomask slits 7 . In order to put equation (3) in terms of w, we equated R1 to NAeff using equation (1) and substituted the equivalence into the equation (3) denominator to arrive at equation (4): In LITE as described here, the thickness and spacing of the photomask slits were scaled from the values for a 152-mm focal length cylindrical lens 7 to the scale of our selected 40-mm focal length, aspheric, cylindrical lens (ThorLabs; AYL5040-A). The optical trade-off of this interference strategy was the generation of side lobes and loss of illumination intensity. Side lobes should theoretically manifest as coplanar light sheets above and below the bright center peak of the main light sheet. However, more than 80% of the total laser energy should in the center sheet 7 . Side lobe minimization is important to reduce the probability of excitation and emission outside the detection objective focal plane.

Optimization of Sheet Dimensions and Parameters
Fadero et al., 2017 9 Creating a non-diffracting light sheet of a width within an order of magnitude of the wavelength of light requires that the light be focused. Accordingly, previous light sheet fluorescence microscopes have used standard (or custom) objective lenses to focus a beam to create a light sheet of a minimal width in the sample 3-6 . This orthogonal, twoobjective method sterically limits the choice of detection objectives to those with a long enough working distance (greater than one millimeter) to focus on the sheet, since the illumination and imaging objectives cannot touch. Here, we present a novel solution for using virtually any existing microscope objective, including those with high NA, for imaging fluorescence signal from a light sheet (Fig. 1A). This represents a significant advance in LSFM, as biologists are no longer limited in their choice of objectives ( Fig.   1B). A detailed, a step-by-step method for selecting the ideal setup of a LITE microscope illuminator based upon the desired objective is presented below.
For effective imaging with LITE, it is necessary to illuminate an objective's volumeof-view (VOV) while minimizing illumination outside the VOV. An objective's VOV can be defined by the product of the two-dimensional field-of-view (FOV) and the onedimensional depth-of-field (DOF). The DOF of an objective, otherwise known as axial resolution, is a set parameter that varies based on the NA and wavelength of the emitted fluorescence (λem) that is collected by the objective (see equation (6) below).
The relationship between the light sheet FWHM w and the objective DOF was derived from the necessity to form the light sheet at the coverglass surface so that it is within the working distance of high-NA objectives. Confined by this geometry, it is impossible to form a light sheet that is completely orthogonal to the focal plane of a high NA objective within its standard working distance (typically < 300 μm) while also projecting Constants in equation (7) (c1 -c5) are unchanging factors that result from the explicit derivation of w. Thus, we arrived at a function of two variables such that w = f(FOV, DOF). In sum, the ideal light sheet width for any given objective could be calculated. In order to illustrate the general trend of how w varied as a function of objective parameters, we obtained the FOV, M, NA, and DOF of 90 commercially-available Fadero et al., 2017 12 detection objectives and plotted the optimal light sheet thickness (w) for each objective as a function of its NA (Fig. 1B).
Once the width of the light sheet was known, we then calculated the length L' over which a light sheet of that width remains non-diffracting from equation (4) 7 . Finally, we also calculated the half-angle of the converging laser, θ, that forms the light sheet, through substituting the general equation for lens numerical aperture, equation (8), into equation (1) and solving for θ to yield equation (9): The resultant angle from (9) is the maximum angle at which the focused sheet should be tilted relative to the VOV inside the sample chamber. Our selected cylindrical illuminating lens was a dry lens that focuses the laser into air (n ≈ 1), so the laser must first refract into the sample chamber (n ≥ 1.33) before reaching the sample (see Methods Part 5 below). Equation (9) is plotted in Figure 1C in order to visually illustrate that θ decreases exponentially as w increases. Since θ and w vary with respect to the different excitation and emission wavelengths among fluorescent proteins, five traces are shown in Figure 1C that correspond to five commonly used biological fluorophores (BFP, CFP, GFP, YFP, and mCherry) and their respective maximal excitation wavelengths (383, 433, 488, 513, and 587 nm).
If the tilting is kept to the minimum θ necessary to completely illuminate the FOV of interest, then out-of-focus excitation was still dramatically reduced (compared to Fadero et al., 2017 13 conventional illumination) in the case of all objectives over a wide range of numerical apertures, magnifications, and depths of field ( Fig. 1B,C). A byproduct of this scheme was that w was always wider than the DOF, a feature that (in principle) lead to increased outof-focus excitation compared to conventional light sheet illumination. However, in part due to the optical sectioning ability of high-NA lenses and the Gaussian nature of light sheet intensity, this effect was not observed in practice (see discussion).

Sample Chambers
To be compatible with LITE, sample chambers must meet two main criteria: (1) have a glass coverslip as the bottom surface for use with high-NA objectives, and (2) have a flat, optically clear, and homogenous side in order to allow the laser to focus inside of the chamber at the coverslip surface. Images presented in this paper were acquired using one of two types of chambers that meet these criteria.
The first type of chamber ( Templates for microfabrication were made by spin coating 1002-F negative photoresist 8 onto clean 50 x 75 mm glass slides at various speeds (1500-3000 rpm) for various thicknesses of photoresist (7-50 μm). Slides were exposed to 400-600 mJ of UV radiation under the patterned photomask. Unpolymerized photoresist was removed chemically, leaving behind hardened microfeatures on the template that act as a negative for imprinting PDMS. The template was placed inside a sealed metal casting chamber with two polished, flat metal sides, tilted at θ relative to the normal of the template's surface.
Pre-mixed liquid PDMS (1:10 w/w ratio of crosslinker to base) was cast over the template before vacuum de-gassing. PDMS was heated to 40 ºC and left to polymerize for 24 hours before separation from the template. We used a 1.0 mm biopsy punch to create inlet and outlet channels through the PDMS into the microchamber for flowing in media/samples.

Image processing
Images were acquired using NIS-Elements (Nikon Instruments). Unless otherwise specified, all images presented in this paper are raw acquisition data (after camera offset subtraction). No post-acquisition deconvolution or stitching is required to view LITE Fadero et al., 2017 19 images, although for some multi-plane movies have been maximum intensity projections (max IP) were generated in the z dimension or deconvolved (specified in the figure legends). Fluorescence intensity measurements, kymographs, maximum intensity projections, image scaling, false-coloring, and movie annotations were performed using Fiji. Richardson-Lucy deconvolution images and three-dimensional supplementary movies were made using NIS-Elements (Nikon Instruments).

LITE illuminates a thin slice of fluorescent samples
The feature shared by all SPIM/LSFM technologies is the spatial restriction of the illumination light to a volume on the order of magnitude of the detection objective's focal plane, so that fluorophores outside of the focal plane do not experience unnecessary illumination. We used LITE microscopy to produce a sheet of light with constant thickness over the desired objective's field of view (150 μm). We theorized we could accomplish this thin illumination scheme using established cylindrical lens-based cosine wave optics 7 .
We therefore calculated the theoretical side view of the light sheet to visualize the predicted sheet width and length (width = 4.3 µm, length = 300 µm; Fig. 2A). In order to verify that our experimental light sheet recapitulates what our calculations predict, we visualized the experimental sheet from the side at 1X magnification through a dilute solution of fluorescein (Fig. 2B, upper). We acquired a 40X magnified image of our experimental light sheet in order to quantify the width ( Figure 2B lower, red box). When compared to the theoretical intensity profile 7 ( Fig. 2A) predicted by the theoretical electric field amplitude at the focal plane of the masked cylindrical lens, our experimentally Fadero et al., 2017 20 observed central peak has nearly identical sheet dimensions (width = 4.3 µm; length = 296 µm; Fig. 2D). Practically, we observed that excitation intensity was too low in these side lobes to generate signal in low density fluorophore regimes, such as those of live cell imaging (data not shown).

LITE operates at native, diffraction-limited spatial resolution
The main goal of LITE microscopy is to combine the use of high-NA objectives to maximize resolution and detection efficiency with LSFM. We thus tested if LITE could be used with high-NA, oil-immersion objectives and provide the high resolution expected from those objectives. The spatial resolution of LITE images should depend solely on the objective NA and the wavelength of emitted fluorescent light. Therefore, spatial resolution in LITE images should be identical to spatial resolution in epi-illumination images, when the objective and samples are the same. In order to quantitatively test whether the spatial resolution is the same, we suspended sub-diffraction (100 nm diameter) fluorescent beads in 2% agarose and acquired images from the same field of beads using LITE (Fig.   3A) and epi-illumination (Fig. 3B) with a high-NA detection objective (60X 1.49 NA oil immersion). We then measured the point spread function (PSF) of each bead in three dimensions by fitting a Gaussian trace to pixel intensity and interpolating the FWHM in each dimension (Fig. 3C). The Gaussian FWHMs of beads visualized with LITE (blue) are identical to those visualized with epi-illumination (orange) (Fig. 3D-F). Spherical aberration artifacts in the z-resolution of the objective were identical for LITE and epiillumination (Fig. 3F), supporting the conclusion that LITE operates at the expected resolution for the chosen objective. histone H2B (LP148 strain) 11 . To measure the true rate of GFP photobleaching without any confounding biological variables, such as new protein translation, proteolysis, and active transport of the fluorescent signal in the z-dimension, we needed a method to inhibit these biological processes. Accordingly, we immobilized the embryos by dissection into M9 nematode media + 2 mM NaN3. This treatment inhibits adenosine triphosphate (ATP) synthesis, thereby indirectly inhibiting ATP-dependent processes such as protein translation, cytoskeleton motor protein activity, and proteolysis. Thus, any decrease in the measured fluorescent signal should be due to excitation-induced photobleaching.
Fluorescent embryos were imaged under identical growth and mounting conditions using either epi-illumination or laser-illumination via LITE. The intensities of the epiillumination field and the LITE laser were specified to generate images with similar initial starting characteristics: namely, signal-to-background ratio (SBR, qualitatively referred to as contrast) and raw integrated fluorescence density. We found that LITE preserves SBR over the course of imaging (Fig. 4A,B). Epi-illumination (orange) starts at a lower SBR and approaches the lower limit of 1.0 (a level precluding analysis) more quickly than LITE (blue; Fig. 4B). Fadero et al., 2017 22 In addition to preserving SBR, LITE also decreases the rate at which the fluorescent signal photobleaches. At equivalent frame numbers, the nucleus visualized with LITE is brighter than that visualized with epi-illumination (Fig. 4C). We quantified the fluorescence intensities of nuclei over time and found that the detected fluorescence decreased more rapidly in the epi nucleus (orange) than in the LITE nucleus (blue; Fig.   4C). In order to more thoroughly illustrate the photobleaching improvement from epiillumination to LITE, we measured the number of frames we could acquire from nuclei before the samples bleached to 90, 80, 70, 60, or 50% (Fig. 4D, S5). On average, LITE significantly increases the number of frames that can be acquired before the nuclei have bleached to a given percent starting intensity. In sum, compared with epi-illumination, LITE preserves SBR and reduces photobleaching. Movie S9). Collectively, these data reveal that LITE can be used to visualize these organisms at high native resolution with constant illumination (i.e. no laser shuttering) for over two hours without any observable phototoxic effects (Fig. 5B,D).

Long-term imaging with LITE enables nuclear lineage analysis
We next set out to demonstrate the power of combining long-term timelapse imaging with low photodamage and high spatiotemporal resolution. The filamentous fungus Ashbya gossypii has emerged as a powerful system in which to study syncytial cell biology 12 . Despite existing in a common cytoplasm, Ashbya nuclei proceed through Fadero et al., 2017 24 the cell cycle out-of-sync with each other. Previous statistical analyses investigating the source of nuclear asynchrony in Ashbya have been based only on single pairs of sister nuclei born of a single mitotic event 13,14 , limiting robust statistical analysis of division patterns across multiple generations. However, long nuclear cycles (between 40 and 200 minutes) 14 and highly oscillatory nuclear motions 15 in Ashbya necessitate high spatiotemporal resolution, 4D imaging for at least two iterations of the average nuclear cycle (~three hours) to trace lineages across multiple nuclear generations. To date, tracking nuclei for this duration at high spatiotemporal resolution has been confounded by photobleaching and phototoxicity. To overcome these limitations, we used LITE to image Ashbya and track nuclear motion and mitotic asynchrony continuously for over seven hours. We expressed a fluorescent histone (H4-EGFP) in Ashbya to detect nuclei for measuring motion and division ( Fig. 6A; Movie S11). Nuclear divisions in a hypha (Fig.   6A, purple box) were readily identified and tracked for five generations (colored arrows).
After 437 minutes of imaging, the Ashbya cell was alive and not detectably photobleached ( Fig. 6A,B; Movie S11). Kymograph analysis enables us to create a temporally scaled pedigree of the nuclear generations (Fig. 6B,C). In sum, LITE is a powerful approach for long-term, high spatiotemporal resolution live imaging.

Discussion
Traditionally, LSFM has been used to reduce photodamage to fluorescent samples by reducing the illumination to only the focal volume of the detection objective, but its geometry has limited use of high-NA objective lenses. LITE is the first SPIM/LSFM modality that allows the use of any objective, allowing researchers to take full advantage Fadero et al., 2017 25 of the efficiency of high-NA objectives. If, for example, LITE is used with a 1.49 NA oilimmersion objective, this setup accepts 88% more emitted fluorescence and offers a 26% increase in native lateral resolution (Fig. 3) than the 1.1 NA water-dipping objective currently used with the Lattice Light Sheet 5 . Collecting more light affords LITE the ability to generate brighter images, which in turn allows the user to illuminate the sample with proportionally less intense laser power to collect the same number of photons as with other SPIM/LSFM modalities, which in turn lowers the photobleaching rate (Fig. 4). The high native spatial resolution of LITE (Fig. 3) will allow cell biologists to obtain images with the spatial resolution to which they are accustomed without sacrificing (and, likely improving) temporal resolution, since LITE does not require deconvolution of multiple structured views as does structured illumination microscopy (SIM).
LITE is also compatible with several other common aspects of modern microscopy.
LITE can be installed non-obtrusively on any upright or inverted stand, allowing for the use of standard equipment, such as eyepieces, objective turrets, and trans-illumination ( Fig. S3). In addition, since the native point-spread function of LITE is identical to that of epi-illumination (Fig. 3), standard post-acquisition deconvolution algorithms can be used on LITE images just as with epi-illumination. For example, the Richardson-Lucy deconvolution algorithm was used for our Ashbya images to increase the contrast between the nuclei and the cytoplasm.
LITE is less photodamaging than epi-illumination, both in the rate at which fluorophores photobleach (Fig. 4C) and the preservation of the image contrast over the acquisition time (Fig. 4B). By selectively illuminating a thin slice of the sample (Fig. 2), LITE reduces the background (a combination of out-of-focus signal and out-of-focus Fadero et al., 2017 26 autofluorescence) relative to the in-focus signal, thus increasing the overall image SBR.
High SBR provides high contrast of the structure of interest from the confounding out-offocus background fluorescence, as well as from sample autofluorescence. The higher variability in the LITE photobleaching rates (Fig. 4D) could be attributed to variability in sheet alignment, chamber construction (Fig. S2-A), or biological noise. Although the LITE sheet measured 4.3 µm thick FWHM (Fig. 2D), in actual cellular imaging conditions, it behaved as if it were thinner. Due to the complexities of cells, this phenomenon is difficult to measure and is best illustrated by the observation that focusing the detection objective (without moving the sheet) by ~1 µm resulted in being outside the excitation volume.
While we have no experimental evidence for this observation, it is conceivable that, given the sheet has a Gaussian intensity profile, only the very peak of the focal volume contains a photon density adequate for fluorophore excitation. Regardless of the variability in sheet alignment or its functional thickness in living samples, our work demonstrates that LITE can be used to image fluorescent samples for longer periods of time than with epiillumination ( Fig. 4D; Movie S4).
As has been observed with current LSFM designs 3-6 , we found that LITE decreases the fluorophore bleaching rate in comparison to epi-illumination (Fig. 4).
Theoretically, this decrease could allow users to reach an equilibrium between photobleaching and turnover at a higher signal and higher SBR with LITE than with epiillumination. Furthermore, we observed an intriguing phenomenon in several of our model organisms in which fluorescence intensity does not detectably decrease over the course of the timelapse (Fig. 5B, 5D, 6; Movies S8, S10, S11). To explain this phenomenon, we suggest that addition of new fluorophores in live organisms could compensate for loss via Fadero et al., 2017 27 photobleaching. If the translation, maturation, and loading of unbleached biological fluorophores collectively result in a simple linear increase in fluorescence, fluorophore turnover could compensate for most photobleaching in live-cell fluorescence microscopy, provided the photobleaching rate is low enough. Understanding this phenomenon will require further study, as it requires characterization of protein abundance and turnover rates to accurately calculate the photobleaching rate in living, developing samples.
We are confident that the decreased rate of photobleaching that LITE offers will allow cell biologists to observe intracellular dynamics at higher native spatiotemporal resolution and for significantly longer periods of time than previously possible using other modes of fluorescence microscopy. We have demonstrated one application of LITE in tracing nuclear lineages (Fig. 6). Lineage tracing has powerful implications, as asymmetric and symmetric inheritance of factors that determine cellular behavior is integral in determining how cells born of a single ancestor can differentiate to different fates.
In the past, we have used the model fungal system Ashbya gossypii where divide asynchronously in a common cytoplasm 13 . Previous work has found that individual nuclear cycles in a single Ashbya cell can vary significantly in their timing 13 , suggesting that there exists nuclear-intrinsic and/or -extrinsic factors that influence nuclear timing. A limitation of past nuclear tracking experiments 14,15 was that photobleaching and phototoxicity prevented long-term imaging that would allow collection of nuclear lineage data over multiple generations, limiting the ability to robustly test for lineage-dependent similarities in nuclear timing. Fadero et al., 2017 28 With LITE microscopy, we are now able to image nuclei for over seven hours to visualize multiple rounds of nuclear division with no noticeable photodamaging effects ( Fig. 6; Movie S11). These data will allow us to study the heritability of division timing over several generations and further our understanding of how heritable nuclear-intrinsic signals contribute to division asynchrony in Ashbya. These sorts of extended image series and statistical analyses are relevant to establishing lineages and division patterns in any cell type, from stem cells to tissues.
Beyond tracking nuclei in Ashbya, we demonstrate that LITE can be effectively     Representative images show P1 nucleus. All images were taken using the same 60X 1.4 NA oil-immersion objective with a frame exposure time of 100 ms, a z-step size of 0.5 μm, a z-range of 20 μm, and no delay between timepoints. Images shown are z maximum intensity projections. Epi and LITE images are outlined in orange and blue, respectively.
Lengths of time the nuclei were exposed to the laser (LITE) or arc lamp (epi) are denoted  of each representative image. Images presented in 5A, 5B, and 5D are taken from the full movies available in Movies S7, S8, and S10, respectively. Images in 5C and 5E are static images taken from three-dimensional z-stacks, which are presented fully in Movies S9 and S11, respectively. Insets in 5B and 5D show images taken from later timepoints (identically scaled) to show low photobleaching. All two-dimensional images presented in