Neuron Replating, a Powerful and Versatile Approach to Study Early Aspects of Neuron Differentiation

Abstract Neuron differentiation includes formation and outgrowth of neurites that differentiate into axons or dendrites. Directed neurite outgrowth is controlled by growth cones that protrude and retract actin-rich structures to sense environmental cues. These cues control local actin filament dynamics, steer growth cones toward attractants and away from repellents, and navigate neurites through the developing brain. Rodent hippocampal neurons are widely used to study the mechanisms underlying neuron differentiation. Genetic manipulation of isolated neurons including gene inactivation or reporter gene expression can be achieved by classical transfections methods, but these methods are restricted to neurons cultured for several days, after neurite formation or outgrowth. Instead, electroporation allows gene manipulation before seeding. However, reporter gene expression usually takes up to 24 h, and time course of gene inactivation depends on the half live of the targeted mRNA and gene product. Hence, these methods do not allow to study early aspects of neuron differentiation. In the present study, we provide a detailed protocol in which we combined electroporation-based gene manipulation of mouse hippocampal neurons before initial seeding with a replating step after 2 d in vitro (DIV) that resets neurons into an undifferentiated stage. By categorizing neurons according to their differentiation stage, thorough morphometric analyses, live imaging of actin dynamics in growth cones as well as guidance cue-mediated growth cone morphologic changes, we demonstrate that differentiation and function of replated neurons did not differ from non-replated neurons. In summary, we provide a protocol that allows to thoroughly characterize differentiation of mouse primary hippocampal neurons.


Introduction
During differentiation, neurons undergo striking morphologic changes from spheres to polar cells possessing an axon and a highly branched dendritic compartment (Dotti et al., 1988;da Silva and Dotti, 2002). Essential steps during early neuron differentiation include the formation and outgrowth of neurites, which later differentiate into axons or dendrites. Directed neurite outgrowth depends on growth cones, structures at neurite tips enriched in actin filaments (F-actin) that steer neurites toward attractants and away from repellent cues and, hence, navigate neurites through the developing brain (Gomez and Letourneau, 2014). Cultured hippocampal neurons isolated from mice or rats are widely used cellular systems to study neuron differentiation as they readily polarize on a two-dimensional substrate at very low densities (Dotti et al., 1988;da Silva and Dotti, 2002). Genetic manipulation including gene silencing, gene deletion or reporter gene expression provide powerful approaches to study virtually all biological processes in cellular systems, including neuron differentiation. Electroporation-based nucleofection as well as classical transfection procedures such as liposome-based transfection or calcium phosphate precipitation are the most commonly applied methods for gene transfer into cultured hippocampal neurons as they are far less labor-intensive when compared with virus infection (Dudek et al., 2001;Ohki et al., 2001;Zeitelhofer et al., 2009;Viesselmann et al., 2011;Sun et al., 2013). Unfortunately, efficiency of classical transfection procedures is rather low and these approaches are convenient only for hippocampal neurons cultured for several days, e.g., at around 6 d in vitro (DIV) or later. Instead, nucleofection allows genetic manipulation of hippocampal neurons before seeding. However, expression of reporter genes usually takes up to 24 h, and more importantly, time course and efficiency of gene silencing or gene deletion depends on the half live of the targeted mRNA and gene product. Consequently, nucleofection of hippocampal neurons does not allow a thorough analysis of neuron differentiation, specifically not of early processes during neuron differentiation. Thus, experimental approaches are needed to circumvent these limitations. We here report a protocol to reset primary hippocampal neurons from embryonic mice at DIV2 into an undifferentiated stage. Before initial seeding, these neurons can be manipulated genetically by means of nucleofection. We show that a combination of nucleofection and replating allows to study early aspects of neuron differentiation.

Mice
Generation of ADFÀ/À/Cfl1 flx/flx mice has been reported before (Bellenchi et al., 2007;Wolf et al., 2015;Zimmermann et al., 2015). Mice were housed with food and water available ad libitum on 12/12 h light/dark cycles. Treatment of mice was in accordance with the German law for conducting animal experiments and followed the guidelines for the care and use of laboratory animals of the National Institutes of Health. Killing of mice has been approved by internal animal welfare authorities (references: AK-5-2014, AK-6-2014, AK-12-2020. Genetic inactivation of Cfl1 in neurons from ADFÀ/À/Cfl1 flx/flx mice was achieved by nucleofection of catalytic active mCherry-Cre. ADFÀ/À/Cfl1 flx/flx neurons expressing a mutant, catalytic inactive mCherry-Cre served as controls. Both constructs have been achieved from the Solecki lab (Kullmann et al., 2020).

Hippocampus dissection and neuron isolation
One day before neuron isolation, glass cover slips (13 mm in diameter, VWR) were placed into 24-well plates and coated overnight with 0.1 mg/ml poly-L-lysine hydrobromid (dilution of 1 mg/ml poly-L-lysine with 0.1 M boric acid at pH 8.5) in a humidified incubator at 37°C and 5% CO 2 . For replating, 24-well plates without cover slips were coated with 0.05 mg/ml poly-L-lysine hydrobromid and similar incubated as above. On the day of neuron isolation, plates were washed twice with ddH 2 O and equilibrated either with 500-ml nucleofection medium (DMEM-31966; Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen) or for non-nucleofected neurons with neurobasal (NB; Invitrogen) medium. Mice of either sex were killed at embryonic day (E)18.5 by decapitation, and brains were dissected on ice in Leibovitz's L15-Medium with 7 mM HEPES (L151H, Invitrogen). After removal of the meninges, hippocampi of each embryo were isolated and collected in a tube containing cooled L151H. Thereafter, medium was replaced by 500-ml prewarmed TrypLE Express (Invitrogen) per embryo and incubated for 6 min at 37°C. Subsequently, hippocampi were washed twice with NB medium containing 2% B27, 2 mM GlutaMax, 100 mg/ml streptomycin, and 100 U/ml penicillin (NB1, Invitrogen). After washing, neurons were triturated in 1 ml NB1 by pipetting seven times up and down with a P1000 pipette. Neuron solution was filled up to 1 ml NB1 medium per embryo and density was calculated by using a hemocytometer. Thereafter, neurons were plated at a density of 60,000 cells per well. 5 h after plating, medium was completely replaced by NB1 medium.

Electroporation of hippocampal neurons
In some experiments, neurons were electroporated before plating. In these experiments, electroporation was performed according to manufacturer's protocol by using the Amaxa P3 Primary Cell 4D-Nucleofector X kit L (Lonza) and 4D-Nucleofector (Lonza). For nucleofection, 250,000 neurons were transfected with 3-mg plasmid and the entire neuron suspension was plated in a single well of a 24-well plate in nucleofection medium; 5 h after plating, medium was completely replaced by NB1 medium.

Replating of hippocampal neurons
At DIV2, neurons were detached and plated again (replated) on cover slips. Before replating, coverslips were prepared as described above. For replating, condition medium (350-ml medium from each well 1 200 ml fresh NB1 medium for each well) was collected and kept in the water bath at 37°C. Remaining medium was aspirated, replaced with prewarmed 500-ml TrypLE Express per well and incubated for 15 min in the humidified incubator. To detach the cells after incubation, the bottom of the well was rinsed twice with the TrypLE Express, and 500-ml prewarmed NB1 medium was added to stop enzymatic reaction. Again, the bottom of the well was rinsed twice with the medium-enzyme solution and then completely transferred in to 1.5-ml cups and centrifuged for 5 min with 7000 rpm. Thereafter, pelleted neurons were re-suspended in 500-ml condition medium and plated on cover slips in 24-well plates and incubated at 37°% with 5% CO 2 until further processing.

Immunocytochemistry
One or 2 d after seeding or replating, neurons were fixed for 10 min in 4% paraformaldehyde in PBS under cytoskeleton preserving conditions (pH 7-7.5). After washing with PBS, neurons were incubated with 0.4% gelatin with 0.5% Triton X-100 in PBS (carrier solution) for 5 min, followed by incubation with the primary antibody rabbit anti-Dcx (1:500, Abcam; in carrier solution). After 90 min incubation, neurons were washed with PBS and incubated with Alexa Fluor 488-coupled phalloidin (1:100, ThermoFisher Scientific) to visualize F-actin and the secondary antibody anti-rabbit IgG coupled to Alexa Fluor 546 (1:500, Invitrogen; in carrier solution). After 60 min of incubation, neurons were washed with PBS and nuclei were stained with the DNA dye Hoechst (1:1000 in PBS, Invitrogen). Neurons were imaged with a Leica TCS SP5 II confocal microscope setup.

Live cell imaging
For live cell imaging, neurons were seeded either directly after nucleofection or after replating in a poly-L-lysine hydrobromid-coated 22-mm glass-bottom dish and cultured for 1d. To measure actin turnover via fluorescence recovery after photobleaching (FRAP), neurons were transfected with GFP-actin (Robert Grosse lab) and imaged with a Leica TCS SP5 II in a chamber heated to 35°C. For imaging, neurons were washed once and then imaged in CO 2 -saturated HBS solution (Invitrogen), supplemented with 4.16 mM NaHCO 3 and 2 mM CaCl 2 . For prebleaching condition, five images of growth cones were acquired and in total 65 images over a time course of 5 min during fluorescence recovery. Images were analyzed with ImageJ (Schindelin et al., 2012) and recovery curve and parameters were calculated with R. To assess retrograde F-actin flow of growth cones neurons were transfected with LifeAct-GFP (Robert Grosse lab) and imaged in a CO 2 -regulated chamber maintained at 37°C. Image acquisition was done with a Leica DMi8 Thunder microscope system and a Leica DFC9000 GTC camera, which acquired images every 5 s for 5 min. Kymograph generation and analysis was performed with ImageJ (Schindelin et al., 2012).

Growth cone collapse assay and BDNF treatment
Neurons were treated for 60 min with 100 ng/ml BDNF (PeproTech), 1 mg/ml Ephrin A5 (R&D Systems) or 1 mg/ml Slit-1 (R&D Systems) before fixation. Images were acquired with a Leica TCS SP5 II microscope system and analyses were done with ImageJ (Schindelin et al., 2012). Growth cone size was measured for determining BDNF effects, whereas repellent cues treated growth cones were categorized into collapsed and non-collapsed according to previous studies (Müller et al., 1990).

Statistics
Statistical tests were done in R or Sigma Plot. For comparing mean values between groups, Student's t test or Mann-Whitney U test was performed. Analyzing the rescue conditions, ANOVA with post hoc test was used. Stage distribution and non-collapsed versus collapsed growth cones were tested for differences with x 2 test.

Replating does not alter actin dynamics in growth cones
Next, as functional readouts, we assessed actin dynamics in replated neurons. We electroporated neurons before seeding to express GFP-actin that allowed us to determine actin turnover in growth cones by FRAP, similar to previous studies (Flynn et al., 2012). We performed FRAP experiments in growth cones from DAR1 neurons and compared actin turnover to growth cones from nonreplated DIV1 neurons. In growth cones from DIV1 neurons, GFP-actin rapidly recovered with a mean half-recovery time (t ½ ) of 77.36 6 12.29 s (n . 20/3; Fig. 4A-C; Movie 1). We noted a similar GFP-actin recovery in growth cones from DAR1 neurons, with no difference in t ½ (74.04 6 10.00 s, n . 20/3, p = 0.83; Fig. 4A-C; Movie 2). Further, we calculated the stable actin fraction that did not recover within the time frame of 300 s. This fraction was not different between growth cones from DIV1 and DAR1 neurons (DIV1: 0.78 6 0.03, DAR1: 0.75 6 0.03, p = 0.500; Fig. 4D). Additionally, we electroporated neurons before plating to express LifeAct-GFP, which allowed us to visualize F-actin in living neurons (Riedl et al., 2008;Flynn et al., 2012). Factin appeared similarly dynamic in growth cones from (2) nucleofection-based genetic manipulation before seeding that could be either reporter gene expression or gene inactivation; (3) culture of hippocampal neurons for 2 d; (4) replating of hippocampal neurons at DIV2 to reset them into an undifferentiated stage; (5) culture of replated neurons until further analyses. DAR1 and DIV1 neurons (Movies 3, 4). Indeed, kymograph analysis revealed similar average retrograde flow velocity of F-actin in growth cones from both groups (DIV1: 8.18 6 1.58 mm/min, n . 20/3, DAR1: 7.73 6 0.82 mm/min, n . 50/3, p = 0.80; Fig. 4E,F). Together, replating neither affected actin turnover nor retrograde F-actin flow in growth cones.

Discussion
In the present study we report a protocol to reset DIV2 primary mouse hippocampus neurons into an undifferentiated stage. We combined replating with nucleofectionbased genetic manipulation (both reporter gene expression as well as gene inactivation by exploiting the Cre/ loxP system) before initial seeding of primary neurons. This approach allows a thorough analysis of neuron  differentiation including early processes such as neurite formation and outgrowth or growth cone function.
Replating of cultured neurons has been reported for various neuron subtypes including primary dorsal root ganglia (DRG) neurons, primary cortical neurons or stem cell (SC)-derived neurons (Caviedes et al., 1990a,b;Koechling et al., 2011;Saijilafu et al., 2013;Frey et al., 2015;Biswas and Kalil, 2018;Calabrese et al., 2019;Lee et al., 2020a). Neuron replating has been implemented to reduce neuron complexity and cell membrane surface area, thereby improving accessibility for electrophysiological recordings, because passive membrane properties such as membrane capacitance or resistance were altered (Caviedes et al., 1990a,b). Further, it has been implemented to transfer SC-derived neurons from normal cell culture dishes onto 384 wells before experiments (Calabrese et al., 2019), and it has been exploited as a paradigm of axon regeneration (Saijilafu et al., 2013;Frey et al., 2015;Lee et al., 2020a). These studies differed in the procedure applied, and some of them only included a brief and rather superficial description of the method. Moreover, these studies either did not focus on early aspects of neuron differentiation, did not systematically compare non-replated and replated neurons or did not combine replating with genetic manipulation. Hence, it remained unknown whether differentiation of replated neurons differed from non-replated neurons and whether a combination of genetic manipulation before initial seeding and replating allowed to study early aspects of neuron differentiation.
We compared cultured mouse hippocampal neurons that have been replated at DIV2 with non-replated neurons, focusing on early aspects of neuron differentiation up to 2 DAR. Our comparison included a categorization of neurons according to their differentiation stage as well as a thorough morphometric analysis. Neuron categorization did not reveal any differences between non-replated and replated neurons, thereby demonstrating that differentiation was largely preserved in replated neurons. Likewise, gross morphology was normal in replated neurons. However, they displayed some changes in neuron Further, we combined our replating procedure with nucleofection-based transfection of hippocampal neurons before initial seeding. We expressed reporter genes such as GFP-actin or LifeAct-GFP that allowed us to determine actin turnover as well as F-actin dynamics in growth cones as functional readouts. By FRAP analysis, we found that actin turnover in growth cones was not different between replated and non-replated neurons. Similarly, retrograde F-actin flow was unchanged in replated neurons. These finding demonstrated that our replating procedure did not alter actin dynamics in growth cones and let us suggest normal growth cone functions in replated neurons. Indeed, growth cones from replated neurons did not differ to those from non-replated neurons in their response to the neurotrophin BDNF or the repellent cues EphA5 and Slit-1. Together, our analysis in hippocampal neurons did not reveal any gross defects in differentiation, morphology or growth cone function in hippocampal neurons induced by the replating procedure. In contrast to our findings, a recent study revealed functional differences between non-replated and replated DRG neurons. Specifically, this study showed that axon regeneration occurred in replated adult DRG neurons even when gene transcription was inhibited by blocking RNA Polymerase II, while axon formation and Figure 5. Normal response to guidance cues in growth cones from replated neurons. A, Representative micrographs of phalloidinstained growth cones from non-replated and replated neurons treated with either PBS or BDNF. B, Growth cone size in non-replated and replated neurons treated with either PBS or BDNF. C, Representative micrographs of phalloidin-stained collapsed and non-collapsed growth cones from non-replated and replated neurons. D, Fractions of collapsed and non-collapsed growth cones in non-replated and replated neurons before and after treatment with EphA5 and Slit-1. Scale bars: 2 mm (A, C); ns: p . 0.05, ***p , 0.001. Green dots in A indicate mean values with SEM. Figure 6. Replating allows studying the relevance of ADF/cofilin for early aspects of neuron differentiation. A, Representative micrographs of phalloidin-stained growth cones from non-replated and replated ADFÀ/À/Cfl1 flx/flx neurons expressing either Cre or Cre-mut. B, Growth cone size in non-replated and replated ADFÀ/À/Cfl1 flx/flx neurons expressing either Cre or Cre-mut. Scale bar: 2 mm (A); ns: p . 0.05, ***p , 0.001. Green dots indicate mean values with SEM.
Open Source Tools and Methods