Hydrogel encapsulation of living organisms for long-term microscopy

Imaging living organisms at high spatial resolution requires effective and innocuous immobilization. Long-term imaging, across development or behavioral states, places further demands on sample mounting with minimal perturbation of the organism. Here we present a simple and inexpensive method for rapid encapsulation of small animals of any developmental stage within a photocrosslinked polyethylene glycol (PEG) hydrogel, gently restricting movement within their confined spaces. Immobilized animals maintained a normal, uncompressed morphology in a hydrated environment and could be exposed to different aqueous chemicals. We focus in particular on the nematode C. elegans, an organism that is typically viewed with paralyzing reagents, nanobeads, adhesives, or microfluidic traps. The hydrogel is optically clear, non-autofluorescent, and nearly index-matched with water for use with light-sheet microscopy. We captured volumetric images of optogenetically-stimulated responses in multiple sensory neurons over 14 hours using a diSPIM light-sheet microscope, and immobilized worms were recoverable and viable after 24 hours encapsulation. We further imaged living pygmy squid hatchlings to demonstrate size scalability, characterized immobilization quality for various crosslinking parameters and identified paralytic-free conditions suitable for high-resolution single cell imaging. PEG hydrogel encapsulation enables continuous observation for hours of small living organisms, from yeast to zebrafish, and is compatible with multiple microscope mounting geometries.

movement within their confined spaces. Immobilized animals maintained a normal, 23 uncompressed morphology in a hydrated environment and could be exposed to different 24 aqueous chemicals. We focus in particular on the nematode C. elegans, an organism that is 25 typically viewed with paralyzing reagents, nanobeads, adhesives, or microfluidic traps. The 26 hydrogel is optically clear, non-autofluorescent, and nearly index-matched with water for use 27 with light-sheet microscopy. We captured volumetric images of optogenetically-stimulated 28 responses in multiple sensory neurons over 14 hours using a diSPIM light-sheet microscope, 29 and immobilized worms were recoverable and viable after 24 hours encapsulation. We further 30 imaged living pygmy squid hatchlings to demonstrate size scalability, characterized 31 immobilization quality for various crosslinking parameters and identified paralytic-free 32 conditions suitable for high-resolution single cell imaging. PEG hydrogel encapsulation enables 33 continuous observation for hours of small living organisms, from yeast to zebrafish, and is 34 compatible with multiple microscope mounting geometries. Fluorescence microscopy has had a profound impact on biomedical research, providing spatial and 3 temporal information about gene expression, molecular dynamics, morphology of labeled structures 4 (Chalfie et al., 1994) and functions such as genetically encoded calcium indicators (GECIs) that 5 indicate activity of electrically-excitable cells (Knöpfel, 2012). While many imaging experiments last 6 only minutes per sample, long-term time-lapse microscopy can indicate changes in more gradual 7 processes lasting hours or days. These longitudinal studies require reliable immobilization of the 8 sample during imaging periods while simultaneously preserving the organism's health and 9 maintaining the function of fluorescent markers. Thus, methods of immobilization must be 10 compatible with required environmental conditions such as hydration, temperature, and nutrition, 11 and with imaging parameters for maximal signal with minimal phototoxicity and sample 12 perturbation (Laissue et al., 2017). 13 14 Small model organisms are widely used for in vivo studies of basic physiology and systemic 15 responses. The nematode Caenorhabditis elegans is a particularly useful model organism due to 16 its <1 mm size, optical transparency, short life cycle, and ease of genetic manipulation. Standard 17 methods for immobilization of C. elegans include treating the organism with paralyzing reagents 18 such as the mitochondrial inhibitor sodium azide or the acetylcholine agonists tetramisole or 19 levamisole (Shaham, 2006). However, chemical paralytics have disadvantages: azide causes gradual 20 fluorophore bleaching, whereas tetramisole contracts the body, altering some physical 21 structures. Additionally, these anesthetic reagents have toxic effects on the worm when applied 22 during long-term studies. Another method mechanically immobilizes worms with nanobeads (Kim 23 et al., 2013) and is advantageous for studies in which recovery and long-term animal health post- 24 imaging are needed, such as after laser axotomy or laser cell ablation (Fang-Yen et al., 2012). In 25 both chemical and nanobead immobilization, the animal remains in a closed environment 26 sandwiched between a cover slip and an agar pad throughout the duration of the experiment, 27 preventing the application of external probes (such as a microinjection needle) or chemical stimuli 28 for neurosensory or physiological studies. Further, the closed environment limits gas exchange 29 and leads to hypoxic conditions within minutes (Jang et al., 2016). Microfluidic traps can 30 immobilize animals in small confined geometries for neural recordings (Chronis et al., 2007), parallel 31 animal imaging (Hulme et al., 2007), and worm sorting applications (Aubry et al., 2015), and some 32 have the ability to present chemicals quickly and precisely (Chronis et al., 2007;Larsch et al., 33 2013). For worm immobilization in physically accessible and open environments, animals have been 34 glued for electrophysiology (Goodman et al., 2012) or mounted under paraffin oil for microinjection 35 (Evans (ed.), 2006). In both cases, animal health can deteriorate over time and the immobilization 36 methods are not compatible with long-term studies. 37 38 An alternative approach to immobilization encapsulates the sample in a three-dimensional hydrogel. 39 Low-melting-point agarose is used to immobilize larger samples such as zebrafish (Renaud et al.,40 2011), but its relative softness allows smaller organisms to move and burrow and image quality is 41 affected by light scattering and weak autofluorescence. A thermoreversible Pluronic hydrogel can 42 allow periodic cycles of C. elegans immobilization and release, gelling at 25 C and solubilizing upon 1 cooling to 22 C (Hwang et al., 2014). These hydrogels are also soft, gelation is slow, and precise 2 temperature control is required, making them challenging to use for continuous long-term, high-3 resolution imaging. Alternatively, covalently-crosslinked hydrogels, such as those based on 4 poly(ethylene-glycol) (PEG), can maintain permanent stiffness and form a gel quickly at any 5 temperature. PEG hydrogels have been studied extensively for cell culture and other 6 applications (Durst et al., 2011;Hou et al., 2010;Lin and Anseth, 2011) and can embed cells for 7 long periods up to several weeks (Albrecht et al., 2006;Skaalure et al., 2015). PEG hydrogels are 8 particularly attractive biomaterials for their tunable mechanical, diffusive, and optical 9 properties that can be varied easily by monomer chain length and concentration (Bryant and Here, we explore the use of PEG hydrogels as an embedding medium for continuous long-term 13 imaging. In particular, we have developed a rapid, convenient method for mounting and 14 immobilizing C. elegans that gently encapsulates the worm in a non-toxic, photosensitive PEG 15 hydrogel, trapping it within a small confined volume. The covalently-crosslinked hydrogel 16 immobilizes worms within seconds of light exposure, holds them permanently for long-term studies, 17 yet can be easily broken to recover animals. The encapsulation process uses standard lab equipment 18 and readily-available materials, works with any size organisms, including all larval and adult stages, 19 and can occur at any desired temperature, unlike thermally-gelling hydrogels that require heating or 20 cooling. The method is versatile, compatible with a wide range of hydrogel size, stiffness and 21 diffusivity, and the degree of animal constraint can be controlled. Here, we characterize the speed 22 and quality of immobilization for high-resolution imaging under varying conditions including light 23 sources, substrates, polymer concentrations, and buffer conditions. The flexibility of worms allows 24 some movement within their confined spaces, and small-scale movements can be further limited 25 by temporary introduction of paralysis reagents, or by crosslinking under hyper-osmotic 26 conditions without any paralytic chemicals.

28
Additionally, we show here that PEG hydrogel encapsulation is ideally suited for light sheet 29 fluorescence microscopy (LSFM), an attractive imaging modality for continuous long-term 3D 30 imaging because it reduces photobleaching by more than an order of magnitude compared hydrogels that are nearly index-matched with water. 38 39 Overall, PEG hydrogel encapsulation is a rapid, gentle, versatile, and inexpensive alternative 40 mounting method useful for continuous long-term imaging, in an open format that may benefit 41 other C. elegans techniques such as laser ablation and microinjection, and scalable to other 42 model organisms such as Drosophila and zebrafish.

RESULTS
2 Immobilization of C. elegans within PEG hydrogels 3 4 Encapsulation of animals in a PEG hydrogel comprises the following steps: preparing a glass 5 slide base and cover, placing spacers that determine hydrogel geometry onto the glass slide 6 base, pipetting the hydrogel precursor solution, picking animals into the precursor droplet, 7 covering the droplet with a cover slip, and crosslinking the hydrogel by brief exposure to light 8 (Fig. 1A). During light exposure, the hydrogel increases viscosity until gelation up to the surface 9 of the embedded worm (Supplementary Video 1). After crosslinking, the animals were fully 10 encapsulated in the hydrogel, preventing large movements beyond their encapsulated space 11 ( Fig. 1B). Encapsulation occurred within seconds and could last for days, as animals could not 12 escape. The flexibility of animals and the hydrogel enabled some movement within this 13 confined space during contraction of body wall muscles, mostly anterior/posterior translation 14 along the body axis and less frequently axial rotation (Supplementary Video 2). Gravid adults 15 could also lay eggs into the encapsulation space. Differential interference contrast (DIC) images 16 at 100x magnification showed minimal change in cellular morphology in animals embedded in a 17 10% PEG hydrogel. 18 By comparison, conventional worm mounting requires melting agarose into a thin pad, 19 picking animals onto the pad, adding a chemical or mechanical immobilizer, and covering with a 20 cover slip (Fig. 1A). Animals exposed to 25 mM sodium azide, which inhibits mitochondrial 21 function and thereby relaxes muscle tone, can take an hour or more to fully immobilize ( Fig.   22 1C). Further, sodium azide-paralyzed animals displayed characteristics of necrotic cell death 23 after 6 hours (Crook et al., 2013). Animals became immobilized with 100 nm polystyrene 24 nanobeads more quickly than with azide, but they could still move gradually over hours ( Fig.   25 1D). For effective nanobead immobilization, animals were compressed causing an apparent 26 increase the width of the worm . 27 Animals of all stages, from eggs to larval stages to adults, could be mounted in the same 28 hydrogel (Fig. 1E). Encapsulated animals were recoverable by breaking the hydrogel with gentle 29 pressure from a worm pick or fine-point forceps (Fig. 1F). After 24 hours of hydrogel 30 encapsulation, young adult C. elegans remained mostly viable; of 241 worms, 207 (86%) 31 crawled away upon release.  Parameters affecting PEG hydrogel crosslinking 14 15 The mechanical, diffusive, and optical properties of PEG hydrogels can be tuned via other photocrosslinking parameters, we imaged animal movement during crosslinking and 19 measured the exposure time at which motion ceased. Using several UV light sources, hydrogels 20 of different size, monomer concentration, and photoinitiator concentration gelled in less than 21 one minute with an irradiance dose of 15 -220 mJ/cm 2 ( Fig. 2A). Higher concentrations of PEG-1 DA reduced the required time of UV light exposure, with a 20% hydrogel gelling in about one-2 half the exposure time required for a 10% hydrogel. 3 The intensity and wavelength of light exposure affects crosslinking rate. The absorbance of 4 the I2959 photoinitiator drops rapidly above 300nm (Fig. 2B) such that shorter wavelength UV 5 sources crosslink the hydrogel more efficiently. For example, a 308 nm medium-wave UV-B 6 transilluminator box gelled the polymer in 6-10s at 2.6 mW/cm 2 (16-26 mJ/cm 2 dose) for 20%-7 10% hydrogel concentration, due to strong overlap between light emission and photoinitiator 8 absorbance. Similarly, a 312 nm handheld medium-wave UV-B light required 12-25s exposure 9 at 2.2 mW/cm 2 (30-55 mJ/cm 2 dose). A 365 nm long-wave UV-A source required a longer 10 exposure time of 30-70 s at 3.1 mW/cm 2 (95-220 mJ/cm 2 dose) for 20%-10% polymer 11 concentration due to weak photoinitiator absorbance at this wavelength. Glass slides and cover 12 slips that absorb strongly at UV-B wavelengths (Supp. Fig. 1), such as those made from soda- 13 lime glass and many plastics, require extended exposure times. 14 Hydrogel geometry had a minor effect on crosslinking rates. Thinner hydrogels (100 µm vs. 15 500 µm) and smaller hydrogels (1 µL vs. 10 µL) gelled slightly faster by less than 20% ( Supp. Fig.   16 2). Photoinitiator concentration did not affect crosslinking rate.  29 30 Because encapsulated animals could push against the hydrogel and move or rotate slightly, we 31 explored modifications that would reduce micron-scale movement for long-term high- 32 resolution microscopy (Fig. 3). Micron-scale movement was quantified by tracking nuclei over 3 33 mins (Supp. Fig. 3) and by a Movement Index (M.I.) sensitive to changes in position, rotation, 1 focus, and photobleaching (calculated as the relative pixel intensity difference across frames,  7 Photobleaching observed when using sodium azide contributed a majority of the M.I. compared 8 with tetramisole, as apparent in individual worm M.I. traces (Supp. Fig. 5). 9 10 Toward paralytic-free tight immobilization, we explored the ability for cooling temperatures or 11 osmotic changes to reduce movement. Cooling to about 4 °C on ice temporarily immobilizes 12 animals (Chung et al., 2008). However, while cooling stopped thrashing before crosslinking, 13 movement after embedding remained similar to uncooled animals. Cooling also allowed 14 animals to settle within the hydrogel droplet, positioning them parallel to the glass substrate 15 for improved imaging. 16 17 Buffer osmolarity changes body size, shrinking animals in hyper-osmotic solutions over the 18 course of minutes through the loss of water (Lamitina et al., 2004). We reasoned that animals 19 placed in a hyper-osmotic solution before or during gelation would shrink, thereby reducing the 20 encapsulation space and tightening their confinement upon return to normal osmolarity. 21 Conversely, swelling animals before crosslinking in hypo-osmotic solutions could expand the 22 hydrogel space, thereby providing more space for movement. Crosslinking the hydrogel in 23 water (0 mOsm), then imaging in S-Basal buffer (280 mOsm), did not significantly increase

13
Imaging neural activity in encapsulated adult worms with light sheet microscopy 14 15 Hydrogels are nearly index-matched with water, therefore suitable for light sheet microscopy. 16 We used PEG hydrogel immobilization to image young adult C. elegans for up to 14 hours 17 without chemical paralytics. Worms expressing the Chrimson channel and GCaMP calcium 18 reporter in the AWA neuron pair were imaged for three 1 hour trials over 14 hours, stimulated 19 with 10-s pulses of red light every minute. Responses were reliably observed in the soma and 20 neurites of both AWAL and AWAR neurons after embedding, and continued strongly in AWAR 21 after 6 h and 13.5 h at room temperature (Fig. 4). Responses in AWAL declined over the first 22 hour, and while present at the beginning of 6 h and 13.5 h trials, declined further over time. 23 AWAR response magnitude declined about 2-fold over 14 hrs in the cell body and 5-fold in 24 neurites.  Encapsulation and imaging of squid hatchlings 11 Soft mounting methods are especially important for flexible specimens, such as marine 12 organisms. We encapsulated three-day old pygmy squid hatchlings in 1.2 mm thick, 4 mm 13 diameter 16% PEG hydrogel disks under anaesthesia and transferred them to sea water for 14 recovery (Fig. 5). After several minutes, hearts were beating and chromatophores actively 15 opened and closed, indicating recovery of muscular behavior in internal structures. External 16 structures in contact with the hydrogel, including arms, fins, and mantle, were immobilized. 17 Prior to encapsulation, squid were stained with 1 uM BODIPY C3 succinimidyl ester dye to 18 generally label cell boundaries throughout the organism (Fig. 5C). A volumetric stack was 19 obtained using the diSPIM light sheet microscope through the tip of one arm of the squid (  11 We present here a method for gentle, effective immobilization of living organisms for long 12 periods of time in covalently photo-crosslinked PEG hydrogels. Mounting larval and adult C. 13 elegans by hydrogel encapsulation was as simple as, or potentially easier than, standard 14 agarose preparations, and offers several advantages. The hydrogel material is inexpensive 15 ($10/mL or 1¢/µL) and photocrosslinking equipment is already present in most biological labs. 16 Animals were immobilized faster than chemical paralytics, more completely than with 17 nanobeads, and without compression under glass that can alter morphology. The overall 18 encapsulation process takes only minutes, with gelation occurring within seconds, trapping the 19 animal within a confined space. The amount of organism movement within the hydrogel cavity 20 could be tuned by applying different solutions that rapidly diffused into the hydrogel. For 21 example, transient exposure to standard paralytic chemicals restricted micron-scale 22 movements within minutes. Alternatively, we identified a paralytic-free immobilization method 23 that temporarily reduced worm body size in hyperosmotic solutions before crosslinking, 1 thereby tightening the hydrogel and immobilizing the animal as effectively as chemical 2 paralytics. High-resolution image quality was maintained over 9 hours of recording in PEG 3 hydrogels, and most animals were recoverable after 24h by physically breaking the hydrogels 4 and allowing animals to escape from the confined space. 5 PEG hydrogels can be crosslinked in various sizes, making this approach suitable for embedding 6 biological samples from <1 µm to mm or larger, a size range spanning from bacteria, yeast, and 7 mammalian cells to small organisms including nematodes, marine organisms, Drosphilia, and 8 zebrafish. We demonstrated encapsulation of living pygmy squid hatchlings, ~1.5 mm long, 9 whose hearts beat internally while soft motile external structures such as arms, fins, and 10 mantle remained immobilized for light sheet or confocal microscopy. 11 The hydrogel mounting method is versatile and compatible with various polymer 12 concentrations that tune mechanical stiffness and diffusive properties. Mechanical stiffness 13 increases 10-fold from 30 to 300 kPa between 10%-20% (w/v) concentrations of PEG-diacrylate 14 (PEG-DA, 3 kDa), whereas mesh size decreases from 4 nm to 3 nm over this range (Nguyen et 15 al., 2012). In these hydrogels, small molecules can diffuse freely, but larger proteins cannot. 16 Longer PEG-DA monomer chains can extend pore size over 10 nm, enabling diffusion of small 17 proteins. Bacterial food would not permeate the hydrogel and embedded animals could not be 18 fed by applying a becterial suspension, although they could be nourished by a chemically 19 defined medium such as C. elegans Maintenance Medium (CeMM) (Szewczyk et al., 2003). 20 Higher PEG-DA concentration reduced crosslinking time, as reported previously (Hockaday et   21 al., 2012), and stiffer hydrogel disks were easier to handle than weaker ones. However, animal 22 immobilization was as good in 12% gels as in stiffer 20% gels (Supp. Fig. 6). This suggests that a 23 concentration of 12% may be optimal for worm immobilization with higher diffusivity when gels 24 needn't be moved, while higher concentrations may be preferred if gels need to be transferred 25 manually. 26 PEG-DA gelation involves activation of the photoinitiator and covalent joining of acrylate 27 polymer chain end-groups via radical chemistry. Oxygen quenches the crosslinking reaction by 28 scavenging the initiator radicals, slowing or preventing gelation. Thus, PEG-DA solutions 29 exposed to air do not gel, and hydrogel disks formed between glass slides retain a ~100 µm 30 border of ungelled polymer where exposed to air (Fig. 1E). This effect may explain subtle 31 increases in exposure requirements for taller and larger hydrogel disks, which have a larger 32 surface exposed to air. Purging the surrounding space with an inert gas, such as nitrogen, 33 effectively reduces this ungelled border, although this step is generally unnecessary unless a 34 precise hydrogel geometry is needed.  (Fairbanks et al., 2009). UV exposure has the potential to cause DNA damage, 2 especially at short wavelengths. However, the UV-B irradiance at 2-3 mW/cm 2 is comparable to 3 sunlight; in fact, PEG hydrogels will crosslink with sun exposure alone. Thus, minimal DNA 4 damage is expected during the ~15s exposure. Wild-type animals can also detect blue and UV 5 light via the LITE-1 channel and respond with transient avoidance behaviors (Edwards et al., 6 2008). 7 8 Brief exposure to photoinitiator free radicals could also cause oxidative damage. While assays 9 for oxidative stress were not performed here, animals remained mostly viable even after 24h 10 encapsulation, including UV and radical exposure during crosslinking, suggesting minimal 11 physiological perturbation. In cell culture, pre-incubation with antioxidants such as ascorbate potentially be mitigated by similar pre-exposure to antioxidants prior to crosslinking. 15 16 PEG hydrogels could be crosslinked by a variety of UV-A and UV-B lamps, many already 17 available in biological labs for DNA gel documentation. Gelation time is dependent on the 18 amount of light absorbed by the photoinitiator: shorter-wave UV-B sources (308, 312 nm) 19 crosslinked faster as they better match the photoinitiator absorbance compared with UV-A 20 sources (365 nm). A narrow-range LED flashlight (365 nm) could also crosslink the hydrogel, but 21 required over 6 mins exposure due to minimal spectral overlap (data not shown). Shorter-wave 22 UV can be blocked by different glass materials (Supp. Fig. 1). For example, UV-C lamps (254 23 nm), should overlap well with I2959 absorbance, but did not cause gelation as most glass and 24 cover slips absorb nearly all UV light in this band. With UV-B sources, we found substrates 25 composed of soda-lime glass required about twice the exposure time as borosilicate glass. 26 27 For time-lapse microscopy, embedded animals were easily identifiable across time points, as 28 they could not escape. However, they could still move within their encapsulation volume more 29 than might be desired for high-resolution microscopy. Paralytics could diffuse into the hydrogel, 30 limiting movement to <8 m over 3 min. Altering osmolarity before crosslinking could 31 immobilize a worm to the same level without paralytics. While long-term exposure to high  19 Venkatachalam et al., 2016), due to its requirement for far greater excitation intensity than 20 LSFM. SDCM is also difficult to use with simultaneous optogenetic activation and optical 21 readout of calcium activity, due to partial spectral overlap between light-sensitive channel and 22 calcium sensor excitation wavelengths (at least with current sensor/channel pairs such as 23 GCaMP/Chrimson and RCaMP/ChR2). Previously, no methods could restrain hatched C. elegans 24 under the open environmental conditions required by the diSPIM, whereas embryos could be 25 directly attached to the substrate (Christensen et al., 2015). Here, by physically preventing 26 thrashing in larval and adult worms, the hydrogel encapsulation method represents the first 27 opportunity to study post-embryonic processes in a light sheet system, and the only current 3D 28 neural imaging method compatible with long-term optogenetic stimulation. 29 Overall, this method provides researchers with a gentle, rapid, inexpensive way to immobilize 30 C. elegans and other organisms for continuous long-term experiments up to several hours, as a 31 complement to existing sample mounting methods. It may not be suitable for experiments in 32 which feeding is required, or for experiments requiring full movement between imaging 33 periods. Nonetheless, by providing open fluidic access to the hydrogel and the potential for 34 paralytic-free imaging, this method benefits long-term studies of dynamic processes in C. 35 elegans and other small model organisms, such as Drosophilia and zebrafish, and may further 36 improve non-imaging techniques such as laser ablation of cells (Bargmann and Avery, 1995) and 37 microinjection of DNA (Stinchcomb et al., 1985) in which recovery of healthy organisms post- 38 immobilization is essential. I.paradoxus pygmy squid adults were collected from sea grass beds in Nagoya, Japan and 13 shipped to the Marine Biological Laboratory (Woods Hole, United States), where they were 14 maintained in aquaria for several months before dying of natural causes. Mature animals 15 readily mated and laid egg masses, with embryos hatching after one week to produce actively 16 swimming and hunting squid larvae. While invertebrate care is not regulated under the US 17 Animal Welfare Act, care and use of I.paradoxus in this work followed its tenets, and adhered to 18 EU regulations and guidelines on the care and use of cephalopods in research. 19 20 Preparation of materials for hydrogel embedding 21 22 PEG hydrogel solutions were prepared by combining 10% -20% w/v poly(ethylene glycol) 23 diacrylate (PEG-DA, 3350 MW, ESI BIO) with 0.05% -0.10% w/v Irgacure 2959 photoinitiator  hydroxy-4'-(2-hydroxyethoxy)-2-metylpropiophenone, I2959, BASF) in deionized water (diH 2 O) 25 or 1x S-basal buffer (100 mM NaCl, 50 mM KPO 4 buffer pH 6.0). A clean 1" x 3" glass slide (VWR 26 Micro Slides) was rendered permanently hydrophobic by exposure to vapors of (tridecafluoro-27 1,1,2,2-tetrahydrooctyl) trichlorosilane (Gelest) under vacuum for 1 hour, or temporarily 28 hydrophobic by wiping with Rain-X Glass Water Repellent. Glass slides were cleaned with 29 ethanol and water, then dried by air gun. For covalent attachment of hydrogels to glass, #1.5 30 cover slips (Thermo Scientific) were silanized by coating with 3-(trimethoxysilyl)propyl 31 methacrylate (Sigma-Aldrich) (21mM in ethanol) for 3 min, followed by ethanol wash, water 32 rinse, and air dry. Treated glass slides can be prepared months in advance. Spacers were 33 prepared by casting polydimethylsiloxane (PDMS, Sylgard 184, Ellsworth Adhesives) in a 1:10 34 (curing agent:base) ratio to thicknesses of 100, 200, and 500 µm. 35 36 37 Embedding live animals in PEG hydrogel 38 39 1 hydrophobic glass slide flanked by two PDMS spacers whose thickness matched the desired 2 hydrogel thickness. Animals were transferred into the hydrogel solution by worm pick and 3 optionally cooled on ice or in a freezer to slow animal movement. A cover slip, untreated or 4 silanized, was placed over the hydrogel droplet and supported by the spacers. The glass 5 slide/cover slip sandwich was then placed over a UV light source and illuminated until gelation, 6 5 -100 sec (depending on lamp power and hydrogel concentration). The sample was either 7 observed immediately or the hydrogel disk was exposed by lifting the cover slip and adding a 8 drop of aqueous solution over the disk. Hydrogel disks could be transferred to wet agar dishes 9 to keep embedded animals hydrated. 10 11 12 Mounting C. elegans for observation with differential interference contrast (DIC) 13 14 For hydrogel encapsulation, 5 µL of a 10% PEG hydrogel with 0.1% I2959 containing late L1/L2 15 stage animals was placed on a fluorinated glass slide with 50 µm thick tape spacers, 16 polymerized with 312 nm UV light, and imaged on an upright Zeiss Axioskop microscope with a 17 100x/1.4 NA objective using DIC optics. Images were captured every hour for 12 hours without 18 movement of the sample. For comparison, animals were mounted onto conventional agarose 19 pads with chemical or physical restraint. In azide paralysis conditions, 1 -5 µL drop of S-Basal 20 buffer with 25 mM sodium azide was placed on a 1% agarose pad that also contained azide. 21 Animals were picked into this droplet and a cover slip was placed on top before imaging. 22 Alternatively, a 0.25 -0.5 µL drop of polystyrene nanobeads (100nm, Polysciences, Inc.) was 23 placed on a 10% agarose pad prior to adding animals, a cover slip, and imaging (Kim et al., 24 2013). 25

27
Recovery of C. elegans from PEG hydrogels 28 29 Young adult worms were embedded in 5 µL 20% PEG hydrogels with 0.1% I2959 by crosslinking 30 for 15 s at 312 nm on an unsilanized glass slide with 100 µm tape spacers. Hydrogels were 31 transferred to an agar plate to keep hydrated and were stored at room temperature (22-23°C 32 degrees). After 24 hours, hydrogels were separated with tweezers and animals were scored for 33 viability (movement and pharyngeal pumping). Some animals were transferred to cover slips 34 with 1% agarose pads for imaging. 35

37
Characterization of light sources and crosslinking conditions 38 39 Several ultra-violet (UV) light sources were used for crosslinking the hydrogel: a gel box 40 transilluminator at 308 nm (Hoefer Scientific Instruments, model UVTM-25) and two hand-held 41 compact UV transilluminators, at 312 nm (International Biotechnologies, Inc, model UVH, 12W) 1 and at 365 nm (UVP, model UVGL-15, 4W). Light power was measured with a power meter 2 (Thorlabs PM100) and 200-1100 nm sensor (Thorlabs S120UV) placed directly on the light 3 source or on a glass slide or cover slip. Power values were converted to irradiance by dividing 4 by the area of the 9.5 mm diameter sensor. Illumination spectra were obtained using a 5 spectrometer (mut GmbH TRISTAN) with a fiber light guide (200-1100 nm range, 400 µm 6 diameter). The photoinitiator absorbance was obtained with a spectrophotometer (Thermo 7 Scientific Multiskan Spectrum). 8 9 The hydrogel crosslinking time was determined optically by monitoring movement of young 10 adult worms during crosslinking. Gels of varying PEG-DA concentration, photoinitiator 11 concentration, and geometry (spacer thickness and volume) were compared for each light 12 source. Videos were captured at 1 frame/s using a Leica S6D stereoscope, converted to 13 reflectance illumination by replacing one eyepiece with a white LED lamp, and recording via the 14 opposite eye path with a UniBrain Fire-I 580c camera. Animal movement was analyzed by 15 comparing frame-to-frame image differences in ImageJ. Crosslinking time was determined as 16 the time until image difference measurements reduced to within 1 standard deviation of noise, 17 and verified visually during observation of videos. 18 19 20 Quantification of movement of hydrogel-embedded animals 21 22 Young adult QW1217 worms with pan-neuronal expression of nuclear-localized mCherry were 23 embedded in 10% and 20% PEG hydrogels (3 µL with 100 µm spacer, 0.1% I2959, 20 s exposure 24 using a 312 nm UV source). Worms received either no pretreatment, exposure to hypoosmotic 25 (diH 2 O, 0 mOsm) or hyperosmotic buffers (0.5 M glycerol in diH 2 O, 500 mOsm or 1.5x S-Basal, 26 420 mOsm) for 10 min, or cooling in a -20 o C freezer or in contact with ice for 1 -3 min prior to 27 crosslinking. PEG hydrogel solutions were prepared in diH 2 O or S-Basal buffer, and some 28 contained the paralytic reagents 25 mM sodium azide (Massie et al., 2003) or 1 mM tetramisole 29 hydrochloride (Sigma) (Larsch et al., 2013). Other hydrogel solutions were prepared in 500 mM 30 glycerol or 1.5x S-Basal hyperosmotic buffers. After crosslinking, all hydrogels were submerged 31 in an aqueous solution of either S-Basal, diH 2 0, sodium azide (25 mM) or tetramisole (1mM) for 32 imaging. Videos were captured with a Hamamatsu Orca-Flash 4.0 camera at 1 frame/sec for 3 33 min on a Zeiss AxioObserver inverted epifluorescence microscope with a 20x/0.5 NA objective. 34 35 Animal movement was analyzed by comparing frame-to-frame image differences (Supp. Fig. 4). 36 First, the absolute value differences between each pair of consecutive frame averaged pixel 37 intensities were calculated. Next, the difference image stack was smoothed (average of its 3 × 3 38 neighborhood). To reduce the contribution of pixel noise, a value of 10 (corresponding to 39 average pixel noise) was subtracted uniformly from each frame and negative values were set to Individual neurons were tracked using NeuroTracker (Larsch et al., 2013) and centroid positions 5 were used to determine the range of axial movement by the animal. 6 7 Long-term Imaging of Optogenetically-Induced Calcium Transients Using 3D Light Sheet Young adult animals co-expressing Chrimson and GCaMP2.2b in the AWA chemosensory 10 neurons were embedded in a PEG hydrogel bonded to a 24 x 50 mm methacrylate-silane- 11 treated cover slip. Animals were picked into a 2.4 µL drop of 13.3% PEG-DA solution with 12 0.067% I2959 in 500 mM glycerol in diH 2 O. After 5 min, the sample was cooled on ice for 30 s 13 and exposed to UV light for 30 s with a 308 nm handheld lamp. Cover slips were mounted into a 14 light sheet chamber (ASI, I-3078-2450) and filled with ~5mL diH 2 O. The dual-inverted selective 15 plane illumination microscope (diSPIM) recorded the calcium response of AWA with a 488 nm 16 excitation laser (Vortran Stradus VersaLase) at 1 mW power setting and a 525/50 nm emission 17 filter. Single-view volumetric stacks (40 slices with 1 µm spacing, 166 x 166 x 40 µm 3 ), were 18 obtained at 1 volume/s for three 60 min recording sessions beginning at t = 0 h, 6 h, and 13.5 h, 19 for a total of 432,000 image frames. A red LED light (617 nm, Mightex, with 620/30 nm filter) 20 was mounted either above the stage, illuminating the animals at a 45 degree angle, or from 21 below with a 600 nm shortpass dichroic through a 4x objective, and controlled via MATLAB and 22 an Arduino controller. Red light pulses, 10 s in duration, were repeated each minute during 23 recordings. For each time point, the volume stack was compressed into a single maximum 24 projection plane, and intensity values were integrated across ROIs surrounding each neuron or 25 neuronal process. After subtracting background intensity from a nearby region, fluorescence 26 intensity (F) integrated across the neuron or neurite ROI was normalized to the initial intensity 27 averaged over 1 s (F 0 ). No interference was observed in background or neural ROIs during red 28 light exposure. 29

30
Imaging of Living Pygmy Squid Using 3D Light Sheet Microscopy 31 Pygmy squid (Idiosepius paradoxus) at three days post-hatching were treated with 1 µM 32 BODIPY 564/570 C 3 succinimidyl ester vital dye (ThermoFisher D2222) in filtered seawater for 1 33 hr to label cell boundaries and generally visualize morphology, modifying methods for the 34 marine worm Platynereis dumerelli (Steinmetz et al., 2007). Squid were washed five times in 35 filtered sea water, then paralyzed by exposure to 3.75% MgCl 2 in sea water for 15 minutes and 36 transferred to a hydrophobic glass slide. Excess sea water was aspirated by pipette such that 37 approximately 4 µL liquid remained. To this, 16 µL of 20% PEG-DA in sea water with 38 photoinitiator was added and gently mixed. A methacrylate silane-treated 24 x 50 mm 2 cover 1 slip was placed 1.2 mm over the droplet using glass slides as spacers. The hydrogel was 2 crosslinked by exposure to 365 nm UV light for 1 min. The coverslip was mounted into the light 3 sheet chamber and filled with sea water. Dual-view volumetric stacks (30 slices with 1 µm 4 spacing, 332 x 332 x 30 µm 3 ) were obtained using a 561 nm laser (4 mW power setting). 5 6 Statistical Analysis 7 Statistical comparisons were made by two-way ANOVA with significance level set at = 0.05, 8 followed by Bonferroni's multiple comparison tests. Data are presented as mean ± standard 9 deviation, unless otherwise noted. 10