bioRxiv preprint doi: https://doi.org/10.1101/2020.04.21.044982; this version posted April 23, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Determining the Depth Limit of Bioluminescent Sources in Scattering Media. Ankit Raghuram1,*, Fan Ye1, Jesse K. Adams1,2, Nathan Shaner3, Jacob T. Robinson1,2,4,5, Ashok Veeraraghavan1,2,6 1 Department of Electrical and Computer Engineering, Rice University, Houston, TX 77005, USA 2 Applied Physics Program, Rice University, Houston, TX 77005, USA 3 Department of Neurosciences, University of California San Diego, La Jolla, CA 92093, USA 4 Department of Neuroscience, Baylor College of Medicine, Houston, TX 77030, USA 5 Department of Bioengineering, Rice University, Houston, TX 77005, USA 6 Department of Computer Science, Rice University, Houston TX 77005, USA * [email protected] Abstract Bioluminescence has several potential advantages compared to fluorescence microscopy for in vivo biological imaging. Because bioluminescence does not require excitation light, imaging can be performed for extended periods of time without phototoxicity or photobleaching, and optical systems can be smaller, simpler, and lighter. Eliminating the need for excitation light may also affect how deeply one can image in scattering biological tissue, but the imaging depth limits for bioluminescence have yet to be reported. Here, we perform a theoretical study of the depth limits of bioluminescence microscopy and find that cellular resolution imaging should be possible at a depth of 5-10 mean free paths (MFPs). This limit is deeper than the depth limit for confocal microscopy and slightly lower than the imaging limit expected for two-photon microscopy under similar conditions. We also validate our predictions experimentally using tissue phantoms. Overall we show that with advancements in the brightness of bioluminescent indicators, it should be possible to achieve deep, long-term imaging in biological tissue with cellular resolution. Introduction 1 The imaging depth in biological tissue is typically limited by scattering; on average, a 2 photon scatters more than ten times within the first millimeter of most biological 3 tissues [19, 20]. This scattering results in blurring of the image and reduced image 4 contrast. Together these effects limit the maximum depth at which cellular-resolution 5 imaging is possible [19]. 6 To improve image quality within scattering tissue, researchers often perform in vivo 7 fluorescence imaging using confocal, two-photon (2P), or three-photon (3P) 8 microscopy [8, 23]. These techniques help to reduce out-of-plane fluorescence, which 9 improves image contrast, but they rely on scanning laser systems. These scanning laser 10 1/13 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.21.044982; this version posted April 23, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. systems are not only complex but subject the tissue to high intensities that lead to 11 photobleaching and phototoxicity, limiting the duration of an imaging 12 experiment [31, 32]. 13 Bioluminescence provides a promising alternative technique that eliminates the need 14 for excitation light, making it possible to perform imaging experiments for extended 15 periods of time with simplified, light-weight imaging systems [16,30]. Bioluminescence is 16 a form of chemiluminescence that generates light through a chemical reaction without 17 the need for excitation light. In this way, bioluminescent proteins act as sources of light 18 within the tissue, avoiding harmful side effects of excitation light like photobleaching 19 and phototoxicity [16, 30]. Additionally, light collection for bioluminescent imaging does 20 not require a light source, color filters, or dichroic mirrors allowing wide-field 21 microscopes [30, 32] or their lensless counterparts [24] to be simpler and smaller than 22 fluorescent systems. 23 A less-discussed advantage of bioluminescence is the fact it may be possible to image 24 more deeply into a scattering biological sample than is possible with 25 epifluorescence [30, 32]. Because bioluminescence does not require excitation light, 26 imaging contrast does not suffer from autofluorescence and excitation scattering, which 27 are the primary factors that limit deep fluorescent imaging in scattering media [20]. 28 While bioluminescence imaging still suffers from attenuation and blurring due to 29 emission scattering, with a bright enough source, one would expect to achieve a greater 30 imaging depth than epifluorescence [7,26, 34]. 31 In this paper, we quantitatively address the relationship between the 32 bioluminescence brightness and imaging depth in scattering media. Using a combination 33 of Monte Carlo simulation and experiments in tissue phantoms, we find the following: 34 (1) The depth limit for bioluminescent imaging ranges from 5-10 MFPs. (2) As the 35 brightness of bioluminescent reporters increases, imaging depth and spatial resolution 36 also increase. (3) With the available bright bioluminescent indicators like 37 eNanoLantern [26,34], we expect that a 2 nM concentration of reporters is needed to 38 achieve the necessary signal to image through > 5 MFPs of scattering media. 39 Fluorescence Imaging 40 Fluorophores emit light that is red-shifted compared to the light that they absorb, 41 allowing one to use color filters to isolate light that is emitted by the scene [5,23]. 42 Fluorescence can be used to image neural activity by monitoring membrane voltage or 43 2 2+ cellular Ca + concentration [13,25]. In recent years, genetically encoded Ca 44 indicators (GECIs) have shown promise as in vivo indicators of neuronal activity 45 2+ because they can respond to changes in cytosolic Ca brought about by opening and 46 2+ closing of Ca channels during an action potential and can be expressed in specific 47 neuronal populations [14]. To record neural activity from cells below the cortical 48 surface, one must image through several hundred microns into scattering brain tissue. 49 Due to excitation and emission scattering, signal originating from these deeper regions 50 profoundly diminishes. To sufficiently illuminate deeper portions of tissue in standard 51 epifluorescence microscopy, increased excitation energy is used to obtain a larger 52 emitted signal [5]. However, out-of-plane autofluorescence and excitation scattering also 53 increase and reduce contrast [30], limiting the depth limit to 3 MFPs for epifluorescence 54 imaging [20]. 55 Confocal and 2P microscopes were developed to combat these problems by reducing 56 the amount of detected background light [2, 28]. Confocal microscopy introduces a 57 pinhole at the illumination and imaging plane in order to image a single scene point at a 58 time, blocking out out-of-plane fluorescence (a combination of autofluorescence and 59 out-of-plane probe fluorescence). The scene point is scanned across the sample to create 60 the full image [28]. 2P microscopy relies on a nonlinear excitation process where 61 2/13 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.21.044982; this version posted April 23, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. fluorescence intensity scales as the square of intensity, making it possible to better 62 restrict fluorescence to the focus of the laser spot. Due to the extremely low probability 63 of excitation outside of the focal spot, 2P microscopy does not require a pinhole to 64 reject out-of-plane fluorescence [2]. However, these methods, along with conventional 65 fluorescence imaging, suffer from probe photobleaching and excitation light-induced 66 phototoxicity, making long term studies challenging [30,32]. Also, in the cases of 67 confocal and 2-photon microscopy, a high powered laser and complex optical systems 68 are needed to perform real-time imaging. These systems can be relatively large and are 69 particularly challenging to implement in a small form-factor design for in vivo imaging. 70 Bioluminescence: Potential and Pitfalls 71 Bioluminescence is a form of chemiluminescence that emits light from chemical reactions. 72 Bioluminescent proteins produce light via oxidation of a small-molecule luciferin by a 73 luciferase or photoprotein. Luciferases display constitutive activity, while photoproteins, 74 2+ like aequorin, produce light in a Ca -dependent manner [18]. Aequorin was first 75 isolated from jellyfish in 1962 [27], and since then, dozens of other bioluminescent 76 proteins have been cloned and engineered into more catalytically active and more 77 spectrally diverse bioluminescent sources [31]. Beyond improvements in the enzymes 78 themselves, the bioluminescence quantum yield of luciferases can be amplified through 79 F¨orsterresonance energy transfer (FRET) to a fluorescent protein. Among the first 80 demonstrations of this principle was the creation of "NanoLantern," in which Renilla 81 luciferase acts as a FRET donor to the yellow fluorescent protein Venus, resulting in a 82 large increase in emitted photons. Scientists are already using proteins like NanoLantern 83 (and variants) for tissue imaging. For example, the NanoLantern-derived calcium sensor 84 CalfluxVTN can be expressed via viral transduction, along with optogenetic actuators,