Three-Dimensional Arenas for the Assessment of Caenorhabditis elegans Behavior

Abstract

Caenorhabditis elegans nematode is a well-established model organism in numerous fields of experimental biology. In nature, C. elegans live in a rich three-dimensional (3D) environment. However, their behavior has been assessed almost exclusively on the open, flat surface of nematode growth medium (NGM) plates, the golden standard for C. elegans culture in the laboratory. We present two methods to build 3D behavioral arenas for C. elegans, by casting and by directly 3D-printing NGM hydrogel. The latter is achieved using a highly customized fused deposition modeling (FDM) 3D printer, modified to employ NGM hydrogel as ink. The result is the advancement of 3D complexity of behavioral assays. To demonstrate the potential of our method, we use the 3D-printed arenas to assess C. elegans physical barriers crossing. C. elegans decision to cross physical obstacles is affected by aging, physiological status (i.e., starvation), and prior experience. The 3D-printed structures can be used to spatially confine C. elegans behaviors, that is, egg laying. We consider these findings a decisive step toward characterizing C. elegans 3D behavior, an area long overlooked due to technical constrains. We envision our method of 3D-printing NGM arenas as a powerful tool in behavioral neurogenetics, neuroethology, and invertebrate model organisms’ neurobiology.

1. Introduction

Caenorhabditis elegans nematode is a well-established model organism in numerous fields of experimental biology, with prominent among them the biology of aging, behavioral neurogenetics, and neurobiology[1-4]. In nature, C. elegans live in a rich three-dimensional (3D) environment (e.g., rotten fruit and muddy soil)[5]. However, their behavior has been assessed almost exclusively on the open, flat surface of nematode growth medium (NGM) plates[6], which are the golden standard of C. elegans culture in the laboratory.

Recently, a new type of associative learning was reported[7], observed in T-shaped mazes. C. elegans learning to reach a target T-maze arm is related to the 3D nature of the arena[7,8], namely, walls, floor, and overall surfaces, which are perceived through multiple sensory modalities. In addition, C. elegans show a clear preference for richly patterned surfaces[9]. Combined together, these findings support the idea that 3D environments are required to witness the full behavioral expression of these nematodes, and a larger part of their nervous system’s capacity. C. elegans 3D locomotion is also gaining a lot of attention[10-13], revealing very interesting dynamics in both swimming and crawling nematodes.

To explore C. elegans spatial behavior, our group uses customized arenas. To this end, 3D-printed plastic molds are used to imprint mazes of various designs in NGM undergoing solidification[7,8]. This leads to reliable, highly repeatable, and importantly, nematode-friendly assays because the generated arenas are built entirely out of NGM. However, several constrains apply. Although the complexity in the x and y dimensions (plane parallel to the assay plate surface) can increase freely, complexity in z dimension (plane perpendicular to the assay plate surface) is essentially limited to varying depth of the arenas. Vertical elements, suspended features, multilevel structures, and other similar architectures are not allowed using molds.

To meet the need for alternative fabrication processes, we explored two methods. The first one includes the use of polyvinyl alcohol (PVA), a water-soluble synthetic polymer, to cast NGM structures (Figure 1). The ensuing parts are of high quality and provide valuable feedback regarding the NGM self-sustaining properties. However, this method’s limitations motivated us to seek another route, one that employs a 3D printer, which uses NGM as ink.

Figure 1.

NGM 3D structures made with the PVA casting method. (A) Two 3D-printed PVA casting molds, red arrows indicate the liquid NGM pouring input. (B) A four-legged NGM crossbridge, made using one of the casts shown in (A), placed on a NGM plate surface. Legs are slightly tilted outwards because of flipping and handling the structure. (C) A diving bell-like structure, consisting of a hemisphere (radius: 5 mm) and designed to have five cylindrical arms, diameter: 3 mm, length: 1 – 1.5 mm, each. Liquid NGM did not reach the entire length of the hollow space inside the PVA cast, resulting in much shorter arms than originally planned (4 mm). Note also the rough surface of the structure. (D) A four-legged crossbridge standing on a 2 mm thick base, raised 4 mm above the base’s surface. This is a much thinner structure than the one in (B), showcasing the self-supporting properties of NGM even in smaller arrangements. Note the missing right arm. (E) Close-up of the NGM diving bell-like structure, shown in C, side view. Yellow frame indicates the position of a C. elegans nematode. Note the bumpy surface and the incomplete beams. (F) Close-up of the four-legged NGM crossbridge, shown in (B), top view. Yellow frame indicates the position of a C. elegans nematode. Note the very rough surface, of which the protruding features (examples highlighted with green arrows) are similar to or even bigger than the worm’s body width. C. elegans worm is challenging to distinguish in both (E) and (F).

A growing number of researchers are employing 3D printing as a transformative tool for cell and tissue engineering[14,15]. This includes 3D scaffolds made of enriched hydrogel-based materials[16]. Hydrogels of 1 – 5% agar concentrations have been successfully explored for 3D bioprinting applications[17]. Most of the occurring structures are sturdily self-supported cubes or other non-hollow, no-overhang designs. Interestingly, the 3D-printing technology has not been used to produce behavioral arenas for the study of small invertebrate animal models, like C. elegans.

We present a highly customized prototype 3D printer, the Parnon Printer (Parnon: Mountain in South Greece, known for its many gorges), which can print 3D parts, suitable for C. elegans behavioral experiments, made of 2% agar-based NGM hydrogel. The resulting arenas are nematode-friendly, minimizing the stress that could have been induced when the animals are transferred from the culture plate into the arenas.

To demonstrate the suitability of the Parnon-printed parts, we used them to assess C. elegans physical barrier crossing ability, in the context of aging (young, middle-aged adults), feeding history (fully fed [FF], starved animals), and prior experience (have been or not in the presence of a 3D structure before). We also explored the usage of 3D-printed structures to spatially confine C. elegans egg laying behavior. C. elegans behavior in 3D environments is by definition not possible to be explored on standard flat NGM plates. Therefore, the findings reported here would likely not have been brought to light if the Parnon Printer had not been developed.

2. Methods

In this section, we describe the PVA casting and the Parnon printing methods, and C. elegans behavioral experiments process. Technical details on the printer’s customization, NGM rheological properties, and software communication between the printer’s parts can be found in the Supplementary File. A list of the major components used for the conversion of the commercial printer into Parnon is provided in the Supplementary File. Details on the Arduino code are provided in the Supplementary File.

The material used to build 3D behavioral arenas is NGM, which is an agar-based hydrogel used to culture C. elegans in the laboratory. NGM 2% in agar was prepared according to standard methods[6,56]. Once all the ingredients are diluted in water, NGM is in liquid form in temperatures higher than ~45°C (melting point), and it solidifies as it cools down. In room temperature, polymerized NGM is in solid state. This is a key property for both the PVA casting and Parnon 3D-printing methods.

2.1. PVA casting polyvinyl alcohol; PVOH)

PVA ([CH2CH(OH)]n) is known for its remarkable property to dissolve in water, because of its high sensitivity to moisture. This makes it particularly useful in 3D printing, when the part at hand needs to have support when printed, that is, in the case of overhung structures, which later must be removed, leaving the 3D-printed part intact[57,58]. This property, combined with the fact that PVA can be used with extrusion fused deposition modeling (FDM) printers, makes it suitable for use as mold material for casting NGM in our approach. To design the cast, SolidWorks (Dassault Systemes, France) was used. The cast was printed at an Ultimaker3 3D printer (Ultimaker BV, USA). The casts shown in Figure 1 were printed in ~5 h. Once the cast was ready, liquid NGM was injected in. NGM took the shape of the space that is left vacant inside the cast. Once the NGM solidified thoroughly, after ~30 min, the cast was immersed in deionized water and was sonicated for ~24 h in a sonicator bath (Branson Ultrasonics, USA). This accelerated dissolving of the PVA cast, which was entirely dissolved and was, therefore, not reusable. Since NGM is not water-soluble after polymerization, only the NGM structure was left intact at the end of this process.

2.2. FDM 3D printer frame

We used a commercially available FDM printer, the DOBOT MOOZ-2 machine (Shenzhen, China; Figure 3). It was preferred against other options due to (i) existing gCode interpretation firmware that controls the linear actuators in the Cartesian axis system, (ii) price efficiency, as a result of low build volume, and (iii) frame rigidity. According to the manufacturer, MOOZ-2 is made of aircraft grade aluminum alloy, which minimizes in situ vibrations and increases rigor.

Figure 3.

C. elegans physical barrier crossing with respect to age and feeding status. (A) Graph showing the % of four groups of day 1 adult nematodes that crossed the square barrier over 120 min; purple inverted triangles: Nematodes that have been FF, cross the barrier into a square baited with food (wF); blue diamonds: Nematodes that have been starved (S) for 24 h prior testing, cross the barrier into a square baited with food (wF); purple triangles: Nematodes that have been FF, cross the barrier into a square baited without food (woF); blue circles: Nematodes that have been starved (S) for 24 h prior testing, cross the barrier into a square without food (woF). (B) Graph showing the % of four groups of FF day 1 adult nematodes that crossed the square barrier over 120 min; purple inverted triangles: Nematodes that have been grown on regular NGM plates (R), cross the barrier into a square baited with food (wF); black diamonds: Nematodes that have experienced a 3D square for 24 h prior testing (3D), cross the barrier into a square baited with food (wF); purple triangles: Nematodes that have been grown on regular NGM plates (R), cross the barrier into a square baited without food (woF); gray circles: Nematodes that have experienced a 3D square for 24 h prior testing (3D), cross the barrier into a square baited without food (woF). Data of C. elegans grown on regular NGM plates (purple triangles and inverted triangles, R) are the same with FF nematodes data (FF) in panel A. Panels (A) and (B): Each data point corresponds to the percentage of worms scored inside the target square over a 120 min period; measurements were taken every 5 or 10 min; horizontal lines indicate the mean and error bars indicate the standard deviation. Comparisons were made using two-tailed, unpaired t-test. Results are significant when *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; only significant comparisons are shown. Shaded area: Confined area did not contain food (without food: woF). (C) Left: Schematic showing the relative size of the 5 × 5 mm 3D-printed square with respect to a 60 mm Petri dish, brown triangles indicate initial placement of nematodes. Right: Example of a 3D-printed square used in the above experiments, 5 × 5 mm, 3 layers, 0.5 mm thick each; scale bar: 1 mm.

All three axes are controlled with lead screw linear actuators which are more accurate, precise, repeatable, and user-friendly than the alternative. The MOOZ-2 is amenable to modifications (Figure 3, Figure S1 in Supplementary File ^{(1.3MB, pdf)} ) that allow Parnon’s custom print head and substrate to be attached.

2.3. Parnon custom substrate

Parnon’s substrate is designed to limit liquid spreading and promote build up along the Z-axis (Figure 2). The custom substrate consists of three major parts: (i) The plotting medium (Figure 2, left panel D [i] and [ii]), (ii) the cooling apparatus (Figure 2 left panel D [iii] and [iv]), and (iii) the leveling system (Figure 2 left panel D [v]).

Figure 2.

Overview of customization changes that convert a FDM printer to Parnon Printer (left panel), and Parnon-printed NGM three-dimensional structures (right panel). Left panel: (A) The FDM 3D Printer DOBOT MOOZ (image courtesy of the manufacturer). The three main parts are the print head, the substrate, and the tri-axial relative motion system. While the stock motion system was considered adequate, both the print head and substrate had to be modified. (B) The prototype Parnon Printer, featuring a customized print head and substrate, designed for printing NGM. Figure S1 in Supplementary File ^{(1.3MB, pdf)} . (C) Customized print head, front view schematic, connected to the X-axis of the triaxial motion system. (i) Bipolar stepper motor of the linear actuator. (ii) 3D-printed custom connector houses the stroke arm (iii) and connects the motor (i) to the heat sink (v). (iii) Stroke arm of the linear actuator. (iv) Heating element that heats the aluminum sink, keeping NGM in liquid state. (v) Custom aluminum heat sink. (vi) Custom connector that connects the stroke arm (iii) to the plunger (vii). (vii) Glass syringe plunger, adhered on the custom connector. (viii) Glass syringe that houses the liquid NGM. (ix) Metal luer-lock nozzle. (x) Copper wire heat induction system that prevents NGM solidification in the nozzle (ix). Details on the print head parts in Supplementary File. (D) Customized substrate, front view schematic, connected to the Y-axis of the triaxial motion system. (i) Glass Petri dish. (ii) Plotting medium. (iii) Peltier device, red arrows indicate direction of heat transfer. (iv) Heat sink to promote the thermal gradient created by the Peltier device (iii). (iv) Springs used as a 4-point bed leveling system (Figure S2 in Supplementary File ^{(1.3MB, pdf)} ). Right panel: The Parnon-printed 3D structures have visibly smoother surface compared to cast-made ones (Figure 1). (A) Perspective and (B) top view of a 7 × 7 mm square-shaped print, consisting of three layers, 500 µm 10 mm/s, actuation speed 31 µm/s, printing head specs: Nozzle ID 813 µm, linear actuator NEMA 14. Inset: A clearly distinguishable nematode, crawling on the printed NGM. (C) Top view of a 7 × 7 mm square print, consisting of three layers, 300 µm thick each. Printing conditions: Head speed 10 mm/s, actuation speed 38 µm/s, printing head specs: Nozzle ID 254 µm, Yellow frame indicates the smoothly deposed NGM line at the corner of the square (90° angle deposition). (D) Top view of a 20 × 20 mm square print, consisting of three layers, 300 mm thick each. Printing conditions: Head speed 10 mm/s, actuation speed 8 mm/s, printing head specs: Nozzle ID 254 mm, linear actuator NEMA 8. (E) Perspective and (F) top view of a C-shaped print, consisting of three layers, 300 mm thick each. Printing conditions: Head speed 10 mm/s, actuation speed 38 mm/s, printing head specs: Nozzle ID 254 mm, linear actuator NEMA 8. Although C-shaped parts were not used in any experiments, they are included here to demonstrate the layering details. Yellow bracket indicates the beveled alignment of layers. Parts in right panel (C), (D), (E), and (F) are printed with a higher resolution print head (combination of nozzle diameter and linear actuator properties), compared to the part in right panel (A) and (B). Prints from both print heads are shown for comparison. For reference, the squared-shaped prints shown in Figures 3 and 4 were generated using the higher resolution print head, as well (see specifications in Supplementary File section). Figure S7 in Supplementary File ^{(1.3MB, pdf)} .

2.3.1. Plotting medium

We used a plotting medium[39,59,60] to reduce liquid spreading, provide support, and promote the cooling process. The plotting medium (Figure 2 left panel D[ii]) is 1.4% glycerin (Sigma-Aldrich, USA) in water. It has the same density as liquid NGM (1.024 g/mL), so the extruded NGM can be effectively suspended during the print to assist in limiting liquid spreading. A 60 mm diameter glass Petri dish (Figure 2 left panel D [i]) holds ~20 mL of the plotting medium for each printing session.

2.3.2. Cooling apparatus

NGM was heated by the heating element when inside the printing head (Figure 2 left panel C, Figure S3A in Supplementary File ^{(1.3MB, pdf)} ). Moreover, there was Joule heating around the nozzle to keep the NGM liquid during extrusion (Figure S3 in Supplementary File ^{(1.3MB, pdf)} ). To cool down the plotting medium and facilitate NGM solidification, a Peltier device (Northbear Electronics) was used[31] (Figure 2 left panel D [iii]), which was operating in 16 V and 12 A. This device cooled the plotting medium according to the thermoelectric effect (Figure 2 left panel D, red arrows). An aluminum heat sink (Figure 2 left panel D [iv]) was implemented underneath the Peltier device to dissipate heat and ultimately increase its efficiency. The Parnon’s cooling mechanism operated at 86% efficiency, achieving a 2°C temperature decrease from the surface of the plotting medium to the floor after ~20 min (Figure S6 ^{(1.3MB, pdf)} in Supplementary File). Details on the efficiency of the Peltier device and the heat flux generated are provided in the Supplementary File.

2.3.3. Leveling system

The Parnon used four individually adjustable spring resistance screws (Figure 2 left panel D [iv] and [v]) to allow for substrate leveling. Additional circular levels with adjustable screws were placed on the Petri dish (Figure 2 left panel D [i], Figure S2 in Supplementary File ^{(1.3MB, pdf)} ).

2.4. Parnon custom print head

The custom print head consists of three major parts: (i) A mechanism to house and heat NGM (Figure 2 left panel C [iv], [v], [vii], [viii], [ix], and [x]), (ii) gCode and Arduino software communication, and (iii) a mechanism to provide actuation pressure (Figure 2 left panel C [i] and [iii]).

2.4.1. Housing and heating mechanism

The aluminum heat sink (Figure 2 left panel C [v]) was machined from 6020 aluminum. It connects all the components on the custom print head to Parnon’s X-axis. The heat sink was heated with a 3” 3/8” diameter, 200 W and 120 V heating element with an internal k-type thermocouple (Figure 2 left panel C [iv]), which was controlled by a programmable temperature controller. The temperature controller is set at 65°C because we expect G’ = ~5 Pa (Figure S6 in Supplementary File ^{(1.3MB, pdf)} ). This modulus was sufficiently low to allow easy extrusion and sufficiently high to limit liquid spreading. Copper wire with 5 V potential difference (Figure 2 left panel C [x], Figure S3 in Supplementary File ^{(1.3MB, pdf)} ) heated the nozzle (joule heating effect) so NGM did not solidify while in the nozzle.

The current version of Parnon used a 3 mL glass syringe (Figure 2 left panel C [viii], Figure S3 in Supplementary File ^{(1.3MB, pdf)} ) to house liquid NGM for extrusion. An early version used a 5 mL glass syringe. The glass plunger (Figure 2 left panel C [vii]) was connected to the linear actuator arm (Figure 2 left panel C [iii]) through a custom 3D-printed connector (Figure 2 left panel C [vi], Figure S3 in Supplementary File ^{(1.3MB, pdf)} , Formlabs Tough Resin), which was designed using Solidworks (Dassault Systemes, France).

NGM was extruded through a ½” stainless steel luer-lock nozzle. Various gauges of luer-lock nozzles were tested (higher gauge translates in lower inner diameter). The initial nozzle had an 813 µmID; after we improved it to a 404 µm ID nozzle, the current version features a 254 µm ID nozzle. Print resolution improves as the ID of the nozzle decreases (Figure 2 right panel, Figure S7 in Supplementary File ^{(1.3MB, pdf)} ).

2.4.2. gCode and software communication

Modifying the hardware of an existing 3D printer compromised the communication the printer had with the original print head. To resolve this issue, we introduced a limit switch, an Arduino card, and a stepper motor driver. More details are provided in the Supplementary File.

2.4.3. Actuation pressure

A linear actuator facilitates extrusion by compressing NGM in the syringe. A characteristic stress is required for the NGM to begin flowing through the nozzle. The time required to reach stress (t_eq.) is the target of our analysis of NGM under actuation pressure. Time t_eq. is used to guide the amount of time prior to the start of print extrusion dedicated to reaching the characteristic stress (strain). More details are provided in the Supplementary File.

2.5. NGM material characterization

NGM 2% agar was prepared according to standard methods[6,56]. The density of liquid NGM was experimentally evaluated to be ~1.024 g/mL (~10 mL of NGM weight ~10.24 g and the density of deionized water was considered 1.0 g/mL). The melting point of agarose (molecular biology grade, Sigma-Aldrich, USA) is ≤65°C and the transition temperature (gel point) is 36 ± 1.5°C (for 1.5% gel), according to the manufacturer. A 100 mL of blue food color (AmeriColor, CA, USA) were added to 100 mL of NGM (0.1%) for in situ visibility and observational purposes.

2.5.1 NGM rheology

Rheology experiments were performed on NGM to uncover the fastest solidification temperature using a TH DR2 Rheometer (TA Instruments, USA). The temperature dropped from 60°C to 25°C at a rate of 5°C/min. The optimal solidification temperature is determined by the maximum slope of the G’ versus T curve (Figure S6 in Supplementary File ^{(1.3MB, pdf)} ). The slope (dG’/dT) peaks at 35.9°C (Figure S6 in Supplementary File ^{(1.3MB, pdf)} ), indicating the fastest solidification temperature.

2.5.2. NGM compressive viscoelastic response

NGM presents a viscoelastic response to compressive stress. Compressive stress tests were run on NGM at 65°C in a 9.11 mm ID (inner diameter) and a 12 mm ID glass syringe. Time t_eq. is required to reach the inflection point of the extrusion equilibrium stress (σ_eq.) under actuation pressure, and time t_r is required to relax from it. The inflection point occurs at a stress value σ_eq., which varies with strain rate e. The time t_eq. required to reach it varies depending on syringe ID.

Details on NGM viscoelastic response characterization are provided in the Supplementary File.

2.6. C. elegans behavioral experiments

Snapshots of C. elegans actions during behavioral experiments were taken with a DP22 camera, mounted on a SZ61 dissection microscope, using CellSens Software (all by Olympus, Japan).

2.6.1. Barrier crossing

In the control experiment, where C. elegans were allowed to reach a food source not framed by a physical barrier, a droplet of OP50 was placed on a regular NGM plate and was framed by a figurative square, drawn with a marker on the bottom of the plate. Nematodes were placed around the targeted area (Figure 5) and were allowed to roam free.

Figure 5.

Assessment of C. elegans ability to cross a physical obstacle to reach a food source. (A) Schematic showing the relative size and position of the figurative square frame used in the control experiments (left) and of the 3D-printed square used in the physical barrier experiments (right), with respect to a 60 mm Petri dish. 3D-printed squares were ~20 × 20 mm, 3 layers, 0.5 mm thick each. Brown triangles indicate initial placement of nematodes. (B) Schematic of the assay. An NGM 3D-printed square (gray) is placed on an NGM plate, and E. coli OP50 food (yellow) is pipetted inside the square. A population of C. elegans nematodes (purple) is transferred onto the NGM plate, outside of the square, at least 3 mm away from it. At t₀, no nematode is inside the square (nematodes placed on assay plate). At t₁₂₀, a fraction of the worms’ population has crossed the square obstacle and is foraging on the food. For the control experiment, the 3D-printed square is absent, and a figurative square is marked at the bottom of the plate, to define the target area (6D, left). (C) Top panel: Snapshots of the square barrier assay when a population of Day 1 adult nematodes is tested, at t₀ (left) and t₁₂₀ (right). Bottom panel: Snapshots of the square barrier assay when a population of day 7 adult nematodes is tested, at t₀ (left) and t₁₂₀ (right). Red dots indicate location of nematodes; number of dots does not correspond to number of nematodes. Scattered dark spots in the substrate are noticeable due to localized crystallization of NGM components, which often happens if the plates are a few weeks old (Supplementary File); however, the overall suitability of the plate is not compromised. (D) Graph showing the % of two age groups of nematodes that crossed the square barrier over 120 min. Purple diamonds: Day 1 adult hermaphrodites (three independent experiments, n₁ = 18, n₂ = 12, and n₃ = 11); black circles: Day 7 adult hermaphrodites (three independent experiments, n₁ = 12, n₂ = 16, and n₃ = 16). As a control experiment, day 1 and day 7 adult nematodes were assayed on dishes with figurative square frame over 120 min; pink diamonds: Day 1 adult hermaphrodites (three independent experiments, n₁ = 21, n₂ = 17, and n₃ = 17); gray circles: Day 7 adult hermaphrodites (three independent experiments, n₁ = 6, n₂ = 8, and n₃ = 9). Comparisons made using multiple unpaired t-tests. Results are significant when *P > 0.05, **P < 0.01, and ***P < 0.001. Only significant comparisons between final (at t = 120 min) are shown; error bars indicate standard deviation, dashed line indicates 50% level. Table S1 ^{(1.3MB, pdf)} in Supplementary File.

In all experiments where Parnon-printed squares were used (Figures 3-5), the 3D structures were rinsed multiple times with deionized water and then were placed on a 60 mm NGM plate. Next, when applicable, a of Escherichia coli OP50 was pipetted gradually inside the square and was left to dry for ~10 min (10 µL for Figure 5 experiments, 5 µL for Figures 3 and 4 experiments). Several C. elegans nematodes were then transferred on the NGM plate, specifically outside of the square(s), at least 3 mm away, and at random locations.

Figure 4.

3D-printed structures can be used to confine C. elegans behavior. (A) 3D-printed square (5 × 5 mm, 3 layers, 0.5 mm thick each), seeded with OP50, on an NGM plate, at time t = 0. (B) The same square as in A, after 24 h (t = 24 h). Eggs have been laid only inside the confined area, and several nematodes can be seen on and inside the square. Note the absence of eggs in the proximity of the square frame. Nematodes initially placed as in Figure 7C. In panels (A) and (B), red dots indicate location of C. elegans. (C) Graph showing the % of eggs laid in the region of each 3D square tested, as shown in (A) and (B). Inside: Eggs laid inside the framed area; on: Eggs laid on the square frame; outside: Eggs laid in the premises of the squares, in 5 mm distance. Bars: Mean; error bars: Standard deviation; dots: Three individual experiments (worms used in each experiment: n₁ = 10, n₂ = 12, and n₃ = 11); above bars: P values of indicated comparisons conducted by Student’s paired t-test, with bold meaning P < 0.05 (significant difference).

To explore the effect of aging on crossing physical barriers (Figure 5), two age cohorts of adult hermaphrodites were tested, namely, young adults of day 1 (L4 + 1) and middle-aged adults of day 7 (L4 + 7). We used a ~20 × 20 mm 3D-printed square, made of 3 NGM layers, 0.5 mm thick each. We considered t₀ the time point at which all nematodes were placed on the plate. The worms were allowed 120 min to explore, and worms scored inside the square were counted every 5 or 10 min. Nematodes that were on the square were counted toward successful crossings. E. coli OP50 food used in all trials came from the same stock batch, and experiments were run on 3 different days.

To explore the effect of feeding history (Figure 3A and 3B), we tested young adults of day 1 (L4 + 1) and middle-aged animals of day 7 (L4 + 7) that had either been FF or starved for 24 h (S). To explore the effect of prior experience (Figure 3C), we tested young adults of day 1 (L4 + 1) that had been either grown on a regular NGM plate (R) or had been moved into an NGM plate containing a Parnon-printed square, similar to the one used in testing (3D). In all cases (Figure 3A-C), a baited (wF) square and a non-baited 3D square (woF) were tested. Moreover, in all cases (Figure 3A-C), we used ~5 × 5 mm squares consisting of three layers, which are 0.5 mm thick each.

NGM hydrogel has the tendency to shrink over time because of dehydration. However, the experiments of Figures 3 and 5 lasted for 2 h, during which the NGM structures maintained their size and shape.

2.6.2. Spatial control of egg laying behavior

For these experiments (Figure 4), ~5 × 5 mm 3D-printed NGM squares consisting of three layers, which are 0.5 mm thick each, were used. After printing, the squares were rinsed with deionized water and then were placed on a 60 mm NGM plate. Next, 5 mL of E. coli OP50 were gradually pipetted inside the square area and were left to dry for ~10 min. A population of day 1 (L4 + 1) adult C. elegans hermaphrodites was then transferred onto the NGM plate and was placed at least 3 mm away from the squares. The plate was checked for eggs after 24 h. For the experiment described in Figure 4, which lasted for 24 h, we sealed the Petri dish by stretching a strip of parafilm around the lid of the dish, a common practice to prevent or delay NGM dehydration.

2.7. C. elegans strains

N2 Bristol (wild type) C. elegans were used in all experiments. C. elegans strain used was initially acquired from Caenorhabditis Genetics Center (provided by C. elegans Reverse Genetics Core Facility at the University of British Columbia, which is part of the international C. elegans Gene Knockout Consortium) and maintained in the laboratory.

2.8. Statistical analyses

Statistical evaluations were made using t-tests (GraphPad Prism 9.0.0). Results are considered significant when P < 0.05. In Figure 4, multiple unpaired t-tests were performed, and in Figure 3, evaluations were made using two-tailed, unpaired t-tests. Additional information is provided in the figures’ captions.

3. Results

3.1. PVA-casted 3D structures

The first fabrication method explored was PVA casting (Figure 1). The produced structures show that it is possible to create 3D NGM structures in a systematic and predictable way.

Specifically, the crossbridge design showed that NGM can support itself, even at a height of 4 mm and an overhang length of 12 mm, with legs 2 × 2 mm in cross section (Figure 1B). Self-standing crossbridges are also feasible in smaller structures (Figure 1D), 2 mm high and 5 mm long overhang, where the bridge legs are thinner, 1.5 mm in cross-section. This means that the mechanical properties of NGM 2% in agar are such that allow for overhangs which stretch a few worm body lengths long, without the need of extra support. It is noted that during the solidification process, NGM enjoyed the support provided by the PVA cast itself (Figure 1A).

Cross-sectional dimensions of beams to as small as 1 × 1 mm have been attempted but were found too small to consistently allow NGM to enter all the way into the cast channels (Figure 1D, missing right arm). This was similarly the case when casting the diving bell design (Figures 1C and Figure 2B), where the cross-section of the arms was also 1 × 1 mm.

PVA casting produces parts with very rough surfaces (Figure 1E and F). This is a consequence of the cast 3D-printing process, during which the PVA is laid in a way that allows micropockets of air among the deposed PVA threads. These micropockets get filled later with NGM, thus creating tiny protrusions (Figure 1F, green arrows).

3.2. Parnon customization and Parnon-printed 3D structures

We extensively modified the commercially available fusion deposition modeling (FDM) printer of choice (Figure 2 left panel, Figure S1 in Supplementary File ^{(1.3MB, pdf)} ) and converted it in a highly customized hydrogel ink 3D printer, named Parnon. Customization included wide-ranging modifications of the print head and substrate (Figure 2 left panel, Figures S2 ^{(1.3MB, pdf)} and Figure S3A in Supplementary File> ^{(1.3MB, pdf)} ), and involved design, 3D printing (motor arm housing, syringe plunger connector), and machining (print head aluminum heat sink) of tailored parts (Figure S1 in Supplementary File ^{(1.3MB, pdf)} ). Effective synergy of all parts was imperative for the successful operation of the instrument.

Parnon printer can successfully use NGM as ink to print 3D structures (Figure 2 right panel and Figure 7 in Supplementary File). Although the examples presented here are of lower complexity compared to PLA and other plastic or resin 3D-printed objects, essential properties of 3D-printed structures are achieved. Hence, the Parnon-printed parts consist of multiple layers (up to 3), making this an effective way to increase vertical complexity of behavioral arenas in the dimension perpendicular to the surface of NGM plates. More layers are mechanically feasible, however, currently, Parnon does not allow for very precise stacking of deposed layers, which results in the top layers and bottom layers being misaligned (Figure 2 right panel).

Importantly, the Parnon-printed parts have smooth surfaces, especially when they are printed using a narrower nozzle and higher resolution printing head (Figure 2 right panel A, B, E, F). The extrusion is unhindered and continuous during a 90° angle turn (example: Figure 2 right panel E, framed area) and NGM lines can be laid successfully in 90° angles or smaller (Figure 7B in Supplementary File).

3.3. C. elegans ability to cross physical barriers and the effect of aging

To confirm that Parnon-printed structures can be used to investigate C. elegans behavior, we ran a series of experiments. In the first scheme, we assessed nematodes’ ability to cross physical barriers, to reach a food source (Figures 3 and 5). We also examined the role of aging (Figure 5), feeding history, and prior experience (Figure 3).

To establish that the 3D-printed squares constitute a physical barrier for C. elegans, we conducted a control experiment, in which day 1 adult nematodes are allowed to reach a food source not framed by a physical barrier (Figure 5). Animals reached the food-containing area in large numbers quickly, and almost all of them (96 – 100%) remained there for most of the 120 min assay (Figure 5D, gray circles).

Next, we challenged day 1 adult C. elegans with a food source, framed by a 20 × 20 mm 3D-printed NGM square (Figure 5). In this case, nematodes entered the framed area gradually and in lower rates, and at the end of the 120 min assay, 74% of animals had crossed the square barrier and had reached the food source (Figure 5D, black circles). Therefore, the NGM square frame presents a physical barrier for nematodes.

We hypothesized that decision-making related to physical challenges differs in young and old animals, as aging-driven behavioral changes have been broadly reported in C. elegans[18-20]. To test our hypothesis, we ran the square barrier experiment with day 7 adult nematodes. In contrast to young adults, only 41% of day 7 C. elegans have crossed the barrier after 120 min (Figure 5D, purple diamonds). When day 7 adult nematodes were tested without a physical barrier (control experiment), results were similar to the ones of day 1 adult nematodes (Figure 5D, pink diamonds). Since day 7 adults are beyond middle age, the above findings suggest an aging-related change in C. elegans physical barrier crossing behavior.

3.4. The effect of confined area size

Next, we asked whether the size of the confined area and its ratio over the total available plate area affects the dynamics of the assay, that is, whether worms travel in and out of the target area or stay inside the target square once they cross it.

To this end, we ran the same experiment, using a 5 × 5 mm 3D-printed NGM square instead of a 20 × 20 mm one. Indeed, when the 20 × 20 mm square was used (Figure 5), the nematodes that crossed the barrier stayed inside the framed area for the remainder of the assay. In case of the 5 × 5 mm square (Figure 5 vs. Figure 3C), nematodes did not remain inside the square once they enter; instead, they might enter, roam, exit, and even reenter. Due to this dynamic situation, the number of worms counted inside the square did not increase monotonically, as shown in Figure 5, instead, it fluctuated. For this reason, and to reflect this dynamic behavior, instead of a time course (Figure 5D), results are presented as scatter plots (Figure 3A and 4B), showing the mean and standard deviation of the percentage of worms scored inside the framed area at any given time point.

In addition, the 20 × 20 mm square (Figure 5) frames ~15% of the 60 mm plate surface area, whereas the 5 × 5 mm one (Figure 3) frames <1% of that area. This is probably why a smaller percentage of worms is scored inside the 5 × 5 mm square at any given moment, compared to what happens with the 20 × 20 mm one. Therefore, the relative size of the square with respect to the culture plate affects the dynamics of the assay and should be taken into consideration when data are interpreted.

3.5. The effect of feeding history

When two groups of day 1 adults were tested (Figure 3A), a FF one (FF; purple inverted triangles) and one that was starved for 24 h prior testing (S; blue diamonds), it was found that starved animals enter the food-baited square in higher numbers than FF ones. This suggests that starved animals might have a stronger motivation to explore their surrounding area and possibly to overcome physical obstacles as well.

When the experiment was performed with no OP50 inside the square (Figure 3A, woF – without food, shaded area, vs. wF – with food, non-shaded area), C. elegans behaved similarly, and starved animals (S woF, blue circles) entered the food-baited square in higher numbers than FF ones (FF woF, purple triangles). Therefore, the nematodes’ feeding history has a strong influence on their tendency to explore beyond a physical barrier even if there is no food beyond it.

3.6. The effect of prior experience

Laboratory populations of C. elegans are commonly grown either in liquid cultures or on flat NGM plates with practically two-dimensional surfaces. C. elegans used in the present work have been cultured for many generations on NGM plates. Hence, we asked whether a group of nematodes that have dwelled on a NGM plate featuring a 3D-printed square will have familiarized themselves with it, and thus will cross a similar 3D barrier in higher numbers. To this end, we placed L4 nematodes on a seeded NGM plate with a 5 × 5 mm Parnon-printed square for 24 h. Then, and on day 1 of their adult life, we tested them on a different plate equipped with a similar, food-baited 5 × 5 mm square.

Nematodes that have previously experienced an environment which includes a 3D square (Figures 3B and 5D) cross the barrier in higher rates than the ones that have experienced only the flat surface of a regular NGM plate (R). This is the case regardless of whether the 3D structure frames a food-baited area (wF) or a non-baited area (woF, shaded).

3.7. Spatial restriction of egg laying

We explored whether the Parnon-printed squares can be used to spatially control selected C. elegans behaviors, that is, egg laying. To this end, a 5 × 5 mm Parnon-printed square, baited with OP50, was used (Figure 4). We used the small size squares to apply a stricter constrain on the behavior we aim to control. Results showed that day 1 adults that were placed on the plate and were left there for 24 h laid eggs almost exclusively in the confined area or on the square barriers themselves, since almost no eggs were found at other plate locations (Figure 4A and 4B).

4. Discussion

Less than a handful of attempts has been made to date to establish worm-friendly and experimentally informative 3D arenas for C. elegans[21-23]. In two of these studies, the 3D platforms are made of porous materials, primarily intended for cultivating and imaging worms[21,23]. In the third case[22], nematodes swim in microfluidic devices that resemble their granulated natural environment, that is, soil. These substrates provide a significantly more realistic terrain than the NGM plate. However, each of them lacks one of the key traits (well-defined and structured arenas, compatibility with microscopy and imaging techniques, or easy and cheap fabrication) that would allow it to be widely adopted for 3D behavioral experiments. The methods presented here, especially the Parnon NGM printing, aspire to fill this gap.

4.1. PVA-casted 3D structures

PVA casting successfully results in creating diverse 3D NGM structures (Figure 1). However, there are important downsides. When the chambers of the PVA cast are small (~1 mm in diameter), NGM is not casted properly. This could be related to the fact that small spaces result in faster cooling of NGM and subsequent clogging. A possible way to resolve this would be to keep the cast warm throughout the process to prevent premature NGM solidification.

At the same time, some of the cast channels might be partially blocked due to trapped air bubbles or 3D-printing discrepancies, occasionally happening when such small empty spaces need to be achieved. A 3D printer with higher resolution than the Ultimaker3 could potentially help resolve this issue, increasing, however, the overall cost of production.

Finally, PVA casting results in parts with very rough surfaces (Figure 1). The increased roughness of the NGM surfaces may or may not be a desired feature, depending on the experiment to be conducted. Indeed, strong roughness could possibly impede or redirect nematode locomotion or interfere with optics and imaging. Note, for example, how challenging is to distinguish a nematode that is crawling on the NGM casted structures of Figure 1 (3A: Yellow arrow, 3B, 3C: Yellow boxes).

In addition, PVA casting process is time consuming, as it takes ~28 h per part to complete (see Methods section). It is also resource heavy, since PVA is significantly more expensive than most PLA or other standard plastic filaments. Moreover, the casts are not reusable, so a new cast must be printed and dissolved for each new NGM part. Furthermore, access to a high-resolution printer, like the Ultimaker3, is required. In addition, certain features cannot be obtained using casts, as for example, complex vertical elements, sharp angles, and very small-sized features. Consequently, PVA casts do not constitute a go to option. This outcome prompted us to explore the 3D-printing route discussed below.

4.2. Parnon customization and Parnon-printed 3D structures

The factor that controls polymerization of NGM is temperature. Our approach can be considered a type of material extrusion bioprinting, in the sense that it is mechanistically very similar to conventional FDM 3D printing[24]. Regarding other approaches, material jetting presents numerous advantages when compared to extrusion 3D printing, especially in applications where high precision in printing biomaterials (e.g., cells and biomolecules) is required[25,26]. In our case, although the application is bio-related, we do not print cells or biomolecules. In addition, harnessing the jetting technique for use with thermosensitive NGM would be challenging in a way that we consider disproportional to the benefits and not required for our application. Other techniques, like ultraviolet-assisted extrusion-based bioprinting[27], could not be applied in our case because NGM is not photocurable. For the same reason, vat polymerization, including stereolithography, digital light processing, and two-photon polymerization, is not he appropriate technique for our application[25,28].

The printability of hydrogels has been broadly demonstrated[15,29-31], including the extrusion of agarose-based hydrogels[32,33]. The 2% agar concentration in the NGM mix is among the lowest used in hydrogel 3D printing[17,31,34] without the use of enhancers[35]. Moreover, the Parnon produced structures presented here and used in C. elegans behavioral experiments feature three layers of NGM. This constitutes a promising achievement when compared to previously reported lattices and scaffolds, made with hydrogel ink of similar concentration and properties[15], mainly because it was possible without additional support[36]. In coarse trials where multilayer cylinders or walls were attempted, although the stability of the structures was satisfying, the overall printing quality was low and inconsistent (Figure S9 in Supplementary File ^{(1.3MB, pdf)} ).

NGM lines can be laid successfully in 90° angles or smaller (Figure S7B in Supplementary File ^{(1.3MB, pdf)} ). This allows for a diverse set of designs, although the printing quality is compromised (Figure S7B in Supplementary File ^{(1.3MB, pdf)} , yellow frame). Printing lines in sharp angles are in general a challenging task[31] and the performance of Parnon is currently considered satisfactory for the current experimental needs. Stricter control of the extrusion process would improve this aspect[31].

In general, the resolution of the FDM-printed structures is restricted by the nozzle diameter[37], which is limited by clogging and depends on the rheological properties of the extruded material. Hence, the typical resolution for FDM is ~100 µm[37]. When it comes to bioprinting hydrogel inks that contain biomolecules, the typical size of the printed features is ~500 µm[37]. According to other reports[38], the achieved resolution in non-hydrogel specific FDM printing is around 50/250 mm (z/xy), while older actuation pressure extrusion attempts using hydrogels[39] report a z/xy resolution of 500 mm. The resolution achieved by Parnon reaches 300 µm in z, and ~250 µm in xy. Therefore, Parnon’s performs well regarding its xy resolution, when compared to overall FDM printing, and scores better than other hydrogel ink printers. Parnon’s z resolution ranks close to its hydrogel ink peers but is considerably worse than generally achieved FDM z resolution. Better control of the actuation pressure and extruded NGM viscosity could help improve this property. Given the size of adult C. elegans body, that is, ~1 mm length and 70 – 100 mm diameter[40], a moderate upgrade would probably suffice for most applications. Nevertheless, the resolution achieved by the current version of Parnon serves well the experimental purpose of the printed parts.

Compared to the PVA-casted parts, Parnon-printed structures are smoother (Figure 1 vs. Figure 2), allowing for easier imaging of nematodes (Figure 2 right panel B inset). Apart from the initial purchasing cost of the commercial printer, 3D printing hydrogel arenas using Parnon are an affordable and fast way to create 3D assays, especially when compared to the PVA casting.

4.3. Parnon mechanics, extrusion, and software communication

At present, pausing and resuming printing is a challenging process. Unlike market FDM 3D printers, the extrusion of which can start and stop easily on demand, Parnon still struggles to do that quickly and precisely. The result is the undesired dripping of NGM, even when the actuation pauses or reverses. Because of this, the printing process needs to be continuous (Figure 7 in Supplementary File), thus affecting the printing paths of certain designs (e.g., cross shape, Figure 7C in Supplementary File). Deeper understanding of the compressive behavior and solidification kinetics of NGM, more efficient control of the nozzle heating, or interventions at the extrusion pathway could result in more precise extrusion control and ultimately allow for more tailored designs.

Parnon’s custom print head is not compatible with the basis printer firmware and thus requires its own commands. Market available gCode slicers do not have the capability to output relevant and processable linear actuator commands in the Parnon’s format. The development of the gCode and Arduino commands remains a rigorous manual process. We are working toward implementing appropriate gCode slicers in future iterations of the prototype.

4.4. C. elegans ability to cross physical barriers

To demonstrate the suitability of the Parnon-printed parts as C. elegans behavioral arenas, we showed that three-layered NGM squares can be successfully used to explore nematodes’ ability to cross physical barriers (Figure 5). This is the first attempt to explore C. elegans ability to overcome physical obstacles, which are made of the same nematode-friendly material on which worms are cultured in the laboratory.

The height of the square used (Figure 5) is estimated at ~1.5 mm (three layers, 0.5 mm thick each), which equals approximately to 1.5 times the adult C. elegans body length. This means that the hurdle is not extraordinarily high, compared to C. elegans, yet it is hard to establish a reference since similar experiments have not been conducted before. Having in mind the 3D character of C. elegans natural habitats[5,41], we speculate that an environment with 1.5 mm high features is not out of range for a field-dwelling nematode. Therefore, we assume that such encounters are not uncommon for C. elegans in the field. Hence, the finding that young adult worms can easily climb over them is not surprising. It is noted that in addition to nematodes which climb over the square barrier, ~5% of the population tested crawled between the square and the substrate on which it rested.

Basic motor skills and chemotaxis remain functional through later stages of adulthood in C. elegans nematodes[42]; however, some decline in locomotion speed[43] and spontaneous locomotion has been reported[19,44]. The observed difference between day 1 and day 7 adult nematodes (Figure 5) could be attributed to age-affected locomotion performance.

Results from dispersal assays have shown that young adult C. elegans move away from the original position and spread into a wide area on the assay plate, but older adults are more likely to remain close to their original location[19]. This agrees with our findings on day 7 adult nematodes, which do not spread over the square barrier as their day conspecifics 1 (Figure 5). Dispersal over physical obstacles has not been previously assessed. Notably, a non-negligible part of the day 7 population in study did cross the barrier, revealing a diversity in individual C. elegans behavior[45] and maybe a diverse impact of aging-related changes[46].

It is reported that learning and memory decline earlier in adulthood than other physiological operations[7,42,47] and show signs of deterioration on day 7[42] or even day 5 of C. elegans adult life[7]. It is possible that some aspects of decision making, for example, aspects related to overcoming physical challenges, are affected as well. Moreover, a decrease in food-seeking exploration in older nematodes is possible.

Our findings on starved C. elegans (Figure 3A) are in alignment with known starvation-induced modifications of nematode behavior that improves their food location likelihood. Indeed, after prolonged starvation, C. elegans change their strategy to long range dispersal[48-51]. Even after 10 – 20 min away from food, neuronal plasticity can change, leading to global, instead of local, search behavior in adult nematodes[52]. Moreover, it has been shown that food deprivation increases threat tolerance in C. elegans, which causes the worms to make “bolder” decisions[53]. Broadening the explored area, even if it includes some physical challenges, may be an additional manifestation of behavioral changes triggered by food deprivation. Feeding history seems to impact obstacle crossing more than the presence or absence of food inside the framed area (Figure 3B).

C. elegans nematodes used in the present work have been cultured for many generations on NGM plates, conforming with common practice in the field. Consequently, the 3D-printed squares are practically the first 3D structure they encounter, besides the unfriendly walls of the plastic Petri dish itself and the NGM chunk often used in culture maintenance, when transferring nematodes into a fresh culture plate. In case of the latter, C. elegans usually actively interact with the chunk, but they commonly just crawl away from it, since the chunk is flipped over the fresh NGM plate[6]. Our findings (Figure 3B) suggest that C. elegans behavior with regard to physical challenges is affected by their prior experience and familiarity with similar structures. We expect that this effect will be stronger if the animals are grown in a 3D environment since hatching, and more 3D structures exist in their culture plate.

4.5. Spatial confinement of C. elegans behavior

Given that in the above experiments (Figure 3), C. elegans get in and out of the square frames without much difficulty, and the localized egg laying behavior (Figure 4) can be attributed to the fact that the only food source available is located inside the 3D squares. This makes them a preferable location for egg laying since this way the newly hatched progeny has immediate access to food. Indeed, it is known that egg laying is regulated by a number of environmental conditions[54], leading to spatial control of the egg laying behavior. Moreover, the egg laying rate in the presence of abundant food is significantly higher than in the absence of food[55]. The use of the 3D squares works well with localized food availability, and it proves very effective when it comes to restricting egg laying and potentially the hatching of L1 larvae as well.

5. Conclusion

This is the first reported attempt to explore C. elegans ability to cross physical obstacles, made of the same nematode-friendly material on which worms are cultured in the laboratory. A method that enables fabrication of such structures, as is the prototype NGM extruding Parnon 3D printer, was the necessary condition to achieve this. We demonstrate that new tools and assays are not only the key to answering biological questions but, most importantly, can often shape the very questions posed. We envision the applications of 3D-printing NGM to enable artificial landscaping and development of diverse playgrounds for the exploration of C. elegans 3D behavior.

Appendix A

List of Parnon major parts and items used for customization

• DOBOT MOOZ-2 machine, Shenzhen, China

• 12V DC Power Supply, 5V Power Supply

• Arduino SD Cardboard, Arduino UNO

• Heating Cartridge: Insertion Heater with Internal Temperature Sensor, for 3/8” Hole, 120V AC, 3” Long Heating Element, 200W, McMaster-Carr, USA

• Heating Controller: Programmable Temperature Controller, for Type K Thermocouple, McMaster-Carr, USA

• Limit Switch: Monoprice Maker Select 3D Printer 13860 Endstop Limit Switch, Amazon, USA

• Linear Actuator: NEMA-8 captive, HeydonKerk, USA

• Microstep Driver: TB6600 4A 9-42V Stepper Motor Driver CNC Controller, Amazon, USA

• Motor Driver: EasyDriver V4.5, SparkFun, USA

• Peltier Devices: Northbear Thermoelectric Cooler Peltier Refrigeration Cooling TEC System Kits Double Fan DIY +Power Supply, NorthBear, USA

  • Thermal Compound: ARCTIC MX-4 - Thermal Paste for Coolers, Heat Sink Paste

Details on the Parnon customized print head

The customized print head includes the following parts (Fig. 2 left panel C): i) Bipolar stepper motor of the linear actuator, to provide the actuating force with a 40V input, controlled with Arduino. ii) 3D-printed custom connector (FormLabs Form2 3D printer, clear resin v4), used to house the stroke arm (iii) and to connect the motor (i) to the heat sink (v). iii) Stroke arm of the linear actuator with a 1.5” total actuation range and a 1:1 stroke of 3°μm. iv) Heating element, 3” length, 3/8” diameter, 200W, set to 65 °C, used to provide the temperature required to keep the NGM in a liquid state. v) Custom heat sink designed to transfer heat from the heating element (iv) to the syringe (viii). vi) Connector (FormLabs Form2 3D printer, tough resin v5), used to connect the stroke arm (iii) to the plunger (vii). vii) Glass syringe plunger which has the custom connector (vi) adhered on the inside. viii) Glass syringe that houses the liquid NGM. ix) Metal luer-lock nozzle. x) Copper wire heat induction system the prevents NGM solidification in the nozzle (ix).

Funding

This work was funded by the University of Michigan Office of Research (UMOR)-Faculty Grants and Awards Program (EG). EG is the recipient of a NIH-NIA K01 award. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Conflict of interest

The authors declare no conflicts of interest.

Author contributions

Conceptualization: Eleni Gourgou

Investigation: Steel Cardoza, Eleni Gourgou

Methodology: Steel Cardoza, Lai Yu Leo Tse, Kira Barton, Eleni Gourgou

Formal analysis: Steel Cardoza, Eleni Gourgou

Visualization: Steel Cardoza, Eleni Gourgou

Funding acquisition: Eleni Gourgou

Writing-review and editing: Lai Yu Leo Tse, Kira Barton, Eleni Gourgou

Writing-first draft: Steel Cardoza, Eleni Gourgou