Generation of 3-D Collagen-based Hydrogels to Analyze Axonal Growth and Behavior During Nervous System Development

This protocol uses natural type I collagen to generate three-dimensional (3-D) hydrogel for monitoring and analyzing the axonal growth. The protocol is centered on culturing small pieces of embryonic or early postnatal rodent brains inside a 3-D hydrogel formed by the rat tail tendonderived type I collagen with specific porosity. Tissue pieces are cultured inside the hydrogel and confronted to specific brain fragments or genetically-modified cell aggregates to produce and secrete molecules suitable for creating a gradient inside the porous matrix. The steps of this protocol are simple and reproducible but include critical steps to be considered carefully during its development. Moreover, the behavior of growing axons can be monitored and analyzed directly using a phase-contrast microscope or mono/multiphoton fluorescence microscope after fixation by immunocytochemical methods.


Introduction
Neuronal axons, ending in axonal growth cones, migrate long distances through the extracellular matrix (ECM) of the embryo over specific pathways to reach their appropriate targets. The growth cone is the distal portion of the axon and it is specialized to sense the physical and molecular environment of the cell 1,2 . From a molecular point of view, growth cones are guided by at least four different molecular mechanisms: contact attraction, chemoattraction, contact repulsion, and chemorepulsion triggered by different axonal guidance cues 3,4,5,6 . Contact-mediated processes can be partially monitored in two-dimensional (2D) cultures on micro-patterned substrates (e.g., with stripes 7,8 or spots 9 containing the molecules). However, axons can navigate to their target in a non-diffusive manner by sensing several attractive and repulsive molecules from guidepost cells in the environment 4,5,10 . Here, we describe an easy method of 3-D culture to check whether a secreted molecule induces chemorepulsive or chemoattractive effects on developing axons.
The earliest studies aimed to determine the effects of axon guidance cues used explant cultures in three-dimensional (3-D) matrices to generate gradients simulating in vivo conditions 11,12 . This approach, together with in vivo experiments, allowed for the identification of four major families of guidance cues: Netrins, Slits, Semaphorins, and Ephrins 4,5,6 . These molecular cues and other factors 13 are integrated by the growing axons, triggering the dynamics of adhesion complexes and transducing mechanical forces via the cytoskeleton 14,15,16 . To generate molecular gradients in 3-D cultures for axonal navigation, pioneering researchers used plasma clot substrates 17 , which was also used for organotypic slice preparations 18 . However, in 1958, a new protocol to generate 3-D collagen hydrogels was reported for studying with Maximow´s devices 19 , a culture platform, used in several studies suitable for microscopic observations 20 . Another pioneer study reported collagen gel as a tool to embed human fibroblasts for studying the differentiation of fibroblasts into myofibroblasts in wound healing processes 21 . In parallel, Lumsden and Davies applied collagen from the bovine dermis to analyze the putative effect of nerve growth factor (NGF) on the guiding of sensory nerve fibers 22 .
With the development of new culture platforms (e.g., multi-well plates) by different companies and laboratories, collagen cultures were adapted to these new devices 6,23,24,25,26 . In parallel, an extract of ECM material derived from the Engelbreth-Holm-Swarm tumor cell line was made commercially available to expand these studies 27 .
Recently, several protocols have been developed to generate molecular gradients with putative roles in axon guidance using 3-D hydrogels (e.g., collagen, fibrin, etc.) 28 . Alternatively, the candidate molecule can be immobilized at different concentration in a porous matrix (e.g., NGF 29 ) or generated by culturing in a small region of the 3-D hydrogel cell aggregates secreting the molecule to generate a radial gradient 1. Plate 2 x 10 6 COS1 cells into a 35 mm Petri dishes and incubate with complete culture medium composed of 100 mL of D-MEM containing 10% (vol/vol) heat-inactivated fetal bovine serum, 0.5% (wt/vol) glutamine and 1% (wt/vol) Pen/Strep in a cell culture incubator, in order to reach 70-80% confluency overnight. Prepare one Petri dish for each transfection procedure. 2. The following day transfect COS1 cells with the DNA encoding the candidate molecule (Netrin-1 or Sema3E) using liposome-based transfection method following the manufacturer's instructions.
1. To do this, mix 250 µL of serum-free medium and DNA (1-2 μg per condition) to a 1.5 mL centrifuge tube (DNA tube) and mix. Incubate at room temperature (RT) for 5 min. Prepare a second tube (liposomal tube) by adding 240 µL of the serum-free medium and 10 µL of the liposomal transfection reagent. Incubate at RT for 5 min.
2. After incubation, add the content of the DNA-tube to the liposomal tube and mix gently. Now incubate at RT for 15 min. Replace the medium on the cultured cells with 1.5 mL of the serum-free medium and add the DNA-liposomes mixture to the Petri dish slowly dropwise. Incubate for 3 h in the CO 2 cell culture incubator.
3. After 3 h of transfection incubation, replace the medium with the complete culture medium and incubate overnight in the incubator. 4. Next day, rinse the cells with 0.1 M Dulbecco's phosphate buffered saline (D-PBS), treat cultures with Trypsin-EDTA (800 µL per each dish for 5-15 min in the CO 2 incubator) and collect detached cells with 15 mL of complete culture medium. 5. Centrifuge the cells at 4 °C at 130 x g for 5 min. After centrifugation, remove media and preserve the pellet containing COS1 cells on ice. 6. Repeat steps 1.12.3-1.12.6 to prepare the collagen working mixture. 7. Add 100-150 μL of the collagen mixture to the pellet of the transfected cells and mix gently by pipetting up and down and spread 45-50 µL of this mixture onto a Petri dish (35 mm diameter) to form a uniform band of collagen-cells of around 1-1.5 cm in length. Place the dish in the incubator at 37 °C (5% CO 2 ) until the gelation (± 15-20 min) is observed. 8. Prepare a second strip containing control cells (mock transfection) in a second culture dish and plate it in the incubator (± 15-20 min) and when gelation is completed add 3-4 mL of warmed (37 °C) COS1 complete culture medium to each dish containing the gelled collagen-cells strips and keep them in the CO 2 cell culture incubator. Thereafter, cut the collagen-cells strips to generate small square pieces (400 to 500 µm in length) using a fine scalpel or a tissue chopper. 9. Transfer all the sections from the same transfection condition to a Petri dish containing 3-3.5 mL of neuronal culture media (NCM) and check under a dissection microscope the quality of the pieces. Again, keep them in the CO 2 cell culture incubator. NOTE: Neuronal culture medium consists of Neurobasal medium containing 1-5% (vol/vol) heat-inactivated horse serum, 2 mM glutamine, 0.5% (wt/vol) glucose, 1% (wt/vol) Pen-Strep solution and 0.044% (wt/vol) NaHCO 3 . Ensure that the pH is between 7.2-7.3.

Generation of Embryonic Explant for Culture
1. Sterilize the surgical tools (scissors, scalpel blade handle, straight and curved forceps) by autoclaving following routine sterilization guidelines. In parallel prepare 500 mL of Hank's balanced salt solution-glucose buffer and 4-5 Petri dishes (100 mm diameter) containing 10 mL of HBSS-G. Place these plates on ice (4 °C). 2. Sacrifice the pregnant female rat (embryonic day 16.5) outside the sterile area, following the approved ethical procedures. Cut the embryo horns with scissors from the abdominal cavity and place it into a large Petri dish containing cold HBSS-G. 3. Place the dish in the laminar flow hood and extract the embryos with straight forceps. Place them into a new dish containing cold HBSS-G.
Next, remove the skin of the embryo using small forceps and carefully dissect the brain using the curved and straight forceps. Place them into a dish containing cold HBSS-G. 4. Under a dissecting microscope, cut the brain in half along the midline to separate both the hemispheres with the scalpel or fine scissors and remove the diencephalon, the meninges and blood vessels from the brain pieces with fine forceps. 5. Repeat steps 3.3-3.4 with the rest of the embryos. Do not delay the dissection for more than 2 h to preserve the tissue quality.

Preparation of 3-D Co-cultures in Collagen Hydrogels
1. Place several sterile 4-well culture plates in the laminar flow hood and prepare a collagen working mixture as previously indicated in steps 1.12.2-1.12.6. 2. Place 15-20 μL of the hydrogel mixture into the bottom of a well to produce a circular collagen base. Repeat this step for the rest of the wells.
Do not prepare more than five plates at the same time to avoid excessive liquid evaporation from the hydrogel base. 3. Keep the dishes in the incubator until the complete gelation (± 15-20 min) is observed and take the plates out of the incubator only when the gelation is completed. Check the quality of the gelled collagen. 4. Transfer a small piece of COS1 cell aggregate with a pipette. Place it onto the hydrogel base and place a tissue piece on the same base with a pipette close to the piece of cells aggregate at one explant-size. 5. Prepare a new working collagen mixture on ice as in steps 1.12.2-1.12.6. 6. Gently pipette 15-20 μL of this new mixture and cover the explant and cell aggregate. A sandwich-like hydrogel culture will be observed. At this moment, re-orientate the explant with a fine tungsten needle (do not touch the COS1 cell aggregate!), so it faces the cell aggregate at ± 500-600 μm. 7. Return the plate to the incubator until the gelation is observed (± 10-15 min), and 0.5 mL of complete NCM supplemented with 2% B27 supplement and keep cultures for 36 to 48 h in the incubator (37 °C, 5% CO 2 ).

Representative Results
Here, we present a widely accessible methodology to study axonal growth in 3-D hydrogel collagen cultures of embryonic mouse nervous system. To this end, we isolated collagen from adult rat tails to generate 3-D matrices in which we cultured genetically-modified cell aggregates expressing Netrin-1 or Sema3E confronted with embryonic neuronal tissue (e.g., CA region of the hippocampus). These cell aggregates formed a radially distributed gradient of the candidate molecule inside the collagen matrix. Finally, to evaluate the neuronal response to different molecules, we labeled the cultures using immunocytochemical methods (e.g., α-TUJ-1) and by applying a simple and easy quantification method, we obtained enough data to determine the effect of the putative candidate on axonal behavior.
In our experiment, when hippocampal axons were confronted with Netrin-1, these axons grew preferentially towards the source of Netrin-1 which indicates that Netrin-1 acts as a chemoattractive molecule for these axons (Figure 1B). In contrast, when hippocampal axons where confronted with Sema3E-secreting cells, most of them grew opposite to the cell aggregate indicating that Sema3E is a chemorepulsive molecule for them ( Figure 1C). In the control condition (mock transfection), all axons grew radially without any directional preference ( Figure 1A). Figure 1D-E are schematic representations of the axonal response and quantification method. After image acquisition, we drew a line in the middle of the explant which delimited the proximal (close to cell aggregate) and the distal (opposite to the cell aggregate) quadrants in order to calculate the proximal/distal ratio (P/D ratio). In control conditions, the axons were equally distributed in both quadrants (radial outgrowth) which indicated a ratio P/D = 1 (Figure 1D). When explants showed increased number of axons in the proximal quadrant in comparison to the distal (indicating chemoattraction) the ratio was P/D > 1 ( Figure 1E) and when the number of axons was higher in the distal quadrant than in the proximal one (indicating chemorepulsion) the ratio was P/D < 1 ( Figure 1F).
In order to achieve excellent results with this technique, we must make sure that collagen polymerization is homogenous, cell transfection is efficient, and the distance between the tissue explant and the cell aggregate is appropriate (see Discussion).
In conclusion, we can confirm that the generation of 3-D collagen-based hydrogels is a useful technique in order to evaluate axonal growth and behavior responses to candidate guidance molecules which can be playing essential roles in the axonal migration during nervous system development.

Discussion
The growth of developing axons is mainly invasive and includes ECM degradation and remodeling. Using the procedure presented here, researchers can obtain a homogenous 3-D matrix formed by the natural type I collagen in which axons (or cells) can respond to a chemical gradient secreted by genetically-modified cells as they do in vivo. Different axonal responses to gradients of attractive or inhibitory cues (protein, lipids, etc.) can be easily compared to specific control (mock transfected cells). As advantages, we must mention that tendons are easy to isolate and indeed they can be remnants of animal experimentation. In addition, tendons are highly collagen I concentrated compared to other tissues such as skin or lung 31 .
Although the methodology presented here is simple to perform, there are some steps that need special attention during the process. Concerning collagen extraction, it is imperative to remove unwanted blood vessels and skin debris from tail tendons in order to improve collagen purity and the quality of gelation. Also, it is mandatory to maintain sterile conditions by performing some steps under a sterile laminar flow hood and sterilizing the surgical tools before use. In addition, it is important to maintain the appropriate pH and temperature conditions of the solutions. For instance, if MEM 10x and bicarbonate solutions are not optimal, the collagen matrix will not polymerize homogeneously, and consequently, the axonal growth and result will be negatively affected. Moreover, if the collagen stock solution is too concentrated or too diluted, the matrices will not gel properly. In our experience, the best collagen stock concentration is approximately 5-5.5 mg/mL of protein (quantified by a colorimetric protein assay kit) and we use a 3:1 dilution (Collagen: 0.1x MEM) to obtain perfect hydrogel matrices. Regarding cell transfection and cell aggregate formation, it is important to maintain sterile conditions and avoid possible contamination, for example, purifying plasmid vectors with endotoxin-free plasmid DNA purification kits is mandatory. Also, we must emphasize that the transfection conditions vary depending on the cell type, passage number, and the plasmid characteristics. Here, we have reported the optimal and routine conditions in our hands. Therefore, researchers should test the recommended concentrations indicated by the manufacturer or adjust them to determine their own optimal conditions. Regarding the problems that may arise with this technique, we must consider that sometimes the 3-D matrices do not present the expected homogeneous gel-like structure. In this case, it is important to check the temperature and pH condition of the solutions and discard them in case it is incorrect. Also, it is recommended to perform a quality control test such as denaturing polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions to validate the purity of collagen stock preparation. With this approach, pure and undamaged type I collagen shows a typical migration pattern consisting of 2 monomeric α chains (α1 and α2), 3 dimeric β chains (β11, β12, variant β11), and 1 trimeric γ chain 32,33 . If the obtained collagen does not fit this pattern, it should not be used. Lastly, after immunostaining, axons can appear radially distributed when confronted with cells secreting a chemorepulsive or chemoattractive molecule. In this situation, the efficiency of transfection must be checked by performing a dot blotting technique on the proper co-culture system (if the DNA plasmids are alkaline phosphatase-tagged) or by processing the culture media after transfection by western blotting. A limitation to consider is that the distance between the cell aggregate and tissue explant is crucial. If they are very far apart, we will not be able to see any clear effect of the secreted molecule on the tissue explant, but if they are very close to each other, the effect will be too strong to be considered as a good result. From our experience, the appropriate distance is around one explant-size (400-500 μm) because the molecular gradient generated by the cell aggregate will extend radially from along 400-500 μm after 24 h in culture.
Alternatively, one can use commercial tumor-derived ECM extract instead of rat tail collagen. In that case, all the procedures must be performed at between 4 and 10 °C, since the gelation of commercial ECM extract is temperature-dependent. Thus, special care should be taken to ensure all culture dishes, pipette tips, culture media, and solutions are maintained at 4 °C.
Finally, although the method presented here is mainly associated with the analysis of neuronal functions such as axonal growth or neuronal migration, it also becomes a useful technique for the pharmacological screening, adhesion assays, in vitro fibrillation experiments and tissue engineering strategies 34,35,36 .

Disclosures
The authors have nothing to disclose.