Primary hippocampal neurons were derived from C57BL/6JBomTac mice (Taconic) mice, bred and raised in the animal facility at the Institute of Biology, Otto-von-Guericke University Magdeburg. Animal maintenance were done according to the guidelines of State of Saxony-Anhalt, Germany and approved by the Landesverwaltungsamt Sachsen-Anhalt (permission number: 42502-2-1177 Uni MD).

Rat pheochromocytoma (PC12) cells were maintained in RPMI medium supplemented with 10% horse serum, 5% fetal bovine serum and 100 U/mL Penicillin-Streptomycin (Gibco). EGFP expressing control PC12 cells (EGFP PC12) and EGFP-Ndr2 expressing PC12 cells (Ndr2 PC12) were previously established (Stork et al., 2004), stably transfected cells were maintained with 200 μg/ml G418 (Invitrogen) at 37°C under 5% CO₂ incubation. For differentiation, cells were seeded (35,000 cells/cm² for neurite growth assays and 5000 cells/cm² for immunocytochemistry) on substrate coated glass coverslips (either with poly-d-lysine only (10 μg/cm² in 0.15 M Borate buffer (pH 8.4), Sigma-Aldrich) or following substrates were added on poly-D-lysine (PDL) coated coverslips: fibronectin (3 μg/cm² in water, Biochrom), laminin (5 μg/cm² in serum-free RPMI media, BD Bioscience), gelatin (100 μg/cm² in water, Biochrom), collagen I (10 μg/cm² in 30% Ethanol, Sigma-Aldrich) and collagen IV (10 μg/cm² in 30% Ethanol, Sigma-Aldrich) and cultured in low RPMI medium (RPMI with 0.2% horse serum, 1× PS and 200 μg/ml G418) supplemented with 50 ng/ml neurite growth factor (NGF, Gibco). To enhance general integrin binding, 0.3 mM MgCl₂ were added to the low RPMI medium. In another set of experiments, a synthetic peptide bearing the α₁β₁ integrin-specific binding motif KTS (CWKTSLTSHYC, generated and purified by Pineda Antibody Service, Berlin, Germany) was applied to the cell media at a concentration of 140 μM to stimulate α₁β₁ integrin heterodimers specifically without altering the cell adhesion on collagen IV substrate (Moreno-Murciano et al., 2003).

After 4 days of NGF treatment, cells were briefly washed with phosphate buffered saline (PBS) and fixed with 4% para-formaldehyde (PFA)/Sucrose in PBS. Images of PC12 cells were taken using an inverted Axiovert 200 M microscope (Zeiss) at 10× objective under phase-contrast filter. Cells with neurites and cells with neurites longer than 100 μm were counted separately using AxioVision 4.0 (Zeiss) and values of each trial were normalized against PDL-control conditions.

To stain for β₁ integrin phosphorylated at threonine788/789 (β₁^pThr788/789), PC12 cells were stimulated with NGF as in neurite growth assays. Cells were fixed with 4% PFA/sucrose in PBS and permeabilized with 0.3% TritonX for 10 min at room temperature. Coverslips were incubated with 5% bovine serum albumin (Roth) and 5% donkey serum (Linaris) for 2 h at room temperature to block unspecific binding. Immunocytochemistry was done using a pThr^788/789 β₁ integrin antibody (1:200, Abcam ab5189). Cells were incubated with the primary antibody overnight at 4°C, washed with PBS and incubated with Alexa-conjugated secondary antibodies (1:1000 dilution in PBS, Invitrogen) for 1 h at room temperature. For staining of α₁ integrin, PC12 cells were grown for a prolonged period of 6 days, according to Zhang et al. (1993), to allow for high expression levels. Cells were fixed, permeabilized and stained using α₁ integrin primary antibody (Abcam, ab78479; diluted 1:200 in PBS). Alexa-coupled secondary antibody was applied together with rhodamine-phalloidine (1:40, R415- Molecular Probes) according to the manufacturer’s protocol to detect the actin cytoskeleton. Cells were finally mounted to slide glasses using ImmuMount (Thermo Scientific) and images were captured with Meta 510 (Zeiss) and DMI 6000 (Leica) microscopes. Images were randomized and evaluated by a researcher who was blinded with respect to the cell line. The most prominent growth tip of differentiated PC12 cells was selected and fluorescent intensities in the growth tips and cell bodies were quantified using ImageJ (NIH) software with built-in histogram and measure functions.

NGF-treated PC12 cells (0 day, 3 days and 6 days NGF treatment (50 ng/ml)) were briefly washed with pre-warmed PBS and lysed in lysis buffer (1% lauryl maltoside N-dodycyl-β-D-maltoside (Merck), 1% NP-40 (Sigma Aldrich), 1 mM Na-orthovanadate, 2 mM EDTA, 50 mM Tris-HCl, 150 mM NaCl, 0.5% deoxycholic acid, 1 mM 4-(2-Aminoethyl)benzenesulfonyl fluoride hydrochloride, 1 μM Pepstatin A, 1 mM NaF and protease inhibitor cocktail (one tablet per 50 ml, Pierce)). Proteins were separated according to their size on polyacrylamide gels and transferred to PVDF membranes (Immobilon FL, Millipore). Membranes were blocked with 5% milk powder (Roth) in PBS for 2 h in room temperature and incubated with α₁ integrin (1:1000 dilution in PBS+0.1% Tween; Abcam78479) and α-tubulin (1:5000 dilution in PBS+0.1% Tween; Sigma Aldrich T6199) antibodies overnight at 4°C. Membranes were briefly washed with PBS and fluorescently labeled IRDye secondary antibodies (Licor) was used with 1:15,000 dilution in PBS+0.1% Tween. After 1 h incubation at room temperature, membranes were washed three times with PBS+0.2% Tween and visualized using an Odyssey scanner (Licor).

Hippocampi from embryonic day 18 (E18) to E19 mice were dissected and disassociated using MACS Neural Tissue Dissociation Kit (Milteny). Disassociated cells were plated either on poly-D-lysine (10 μg/cm², Sigma-Aldrich) or poly-D-lysine+ Laminin111 (0.45 μg/cm², in PBS, Biolamina) coated coverslips (25,000 cells/cm²). Except for the plating media (DMEM+ 10% FBS (v/v) + L-GlutaMax (2 mM)) at day in vitro (DIV0), neurons were kept with neurobasal media containing B27 (2% v/v) and L-GlutaMax (0.5 mM) (all from Invitrogen) that had been glia conditioned for 72 h. At DIV3, neurons were transfected either with EGFP-C1 (ClonTech), EGFP-Ndr2 (Stork et al., 2004), pLL3.7_shLuc (TTG GAG AAA AGC CTT GTT T) or pLL3.7_shNdr2 (CCT CAT CTG CCA ATC CCT C; Rehberg et al., 2014) using CalPhos Mammalian Transfection kit according to the manufacturer instructions (ClonTech). All neurons were also co-transfected with pCMV-tdTomato (ClonTech) to outline the dendritic morphology. At DIV7, primary neurons were fixed with 4% paraformaldehyde and 4% sucrose in 0.1 M PBS, mounted on slides using Immuno-Mount (Thermo Scientific) and images were taken using a DMI 6000 epifluorescence microscope (Leica). Dendrites of the double-transfected neurons were selected according to their morphological features (Kaech and Banker, 2006) and were traced using QWin software (Leica). Sholl analysis of the traced dendrites was done using Fiji (ImageJ) with 10 μm radius intervals.

Statistical analysis was performed using two-way ANOVA followed by Fisher Least Significant (LSD) test or Kruskal-Wallis test for multiple comparisons. Immunofluorescence data were log-transformed before statistical analysis. The number of cells with long neurites (>100 μm) and short neurites (<100 μm) were compared using Chi-square tests. Pairwise comparisons were done using Student’s t-test or Mann-Whitney test when appropriate and a statistical threshold for significance was set at p < 0.05. Normality of the datasets was tested using Shapiro-Wilk test.

To investigate the role of Ndr2 kinase in neurite growth, EGFP transfected PC12 cells (EGFP PC12) and Ndr2 transfected PC12 cells (Ndr2 PC12) were cultured on different ECM substrates and treated with NGF. Both cell lines showed efficient neurite formation on PDL and collagen IV with >95% of cells forming discernable neurites under each condition. Laminin also was a favorable substrate for both lines, whereas less than 50% of control cells showed neurites on fibronectin, collagen I and gelatin. On those less efficient substrates, an increase of outgrowth in Ndr2 cells was evident, as the proportion of cells with neurites were significantly increased on fibronectin, gelatin and collagen I substrates (Figure 1A, two-way ANOVA genotype × substrate interaction: F_(5,24) = 4.575, p = 0.0045). To address potential differences in neurite extension, we further analyzed cells with neurites longer than 100 μm (Figure 1B). Here we have selected PDL, Laminin and Collagen IV substrates since they resulted in neuronal differentiation in more than 90% of cells in both cell lines. Here we confirmed the previously reported increase in Ndr2 PC12 cells neurite extension of on PDL (X²₍₁₎ = 4.276, p < 0.05, Stork et al., 2004), but did not observe any change in neurite extension on laminin substrate between genotypes (X²₍₁₎ = 0.1364, p = 0.71). In contrast to PDL, on collagen IV substrate Ndr2 PC12 cells displayed significantly reduced rather than increased neurite growth compared to EGFP PC12 cells (Figures 1B,C; X²₍₁₎ = 81.45, p < 0.001). Therefore, PDL and collagen IV were chosen as substrates for further experiments to investigate how Ndr2 kinase controls substrate specific neural differentiation of PC12 cells.

Integrin receptors on the cell surface are recruited by these ECM substrates and previous studies showed that β₁ integrin containing integrin heterodimers are the main receptors for their signal transmission in developing neurons (Denda and Reichardt, 2007; Lei et al., 2012). Since we have previously described the role of Ndr2 in the “inside-out” activation of β₁ integrins, we next tested whether this process may be involved in the differential response of Ndr2 PC12 cells to integrin substrates. Immunohistochemical staining of β₁^pThr788/789, which is required for β₁ integrin activation (Nilsson et al., 2006), revealed an expression of this activated β₁ integrin form in both somata and neurite tips of both control cells and Ndr2 PC12 cells (Figures 2A–F). Quantification of the relative amount of β₁^pThr788/789 revealed no main effect of the coating substrate (two-way ANOVA, substrate effect: F_(1,36) = 0.05323, p = 0.8188). However, an increase of β₁^pThr788/789 was observed in the neurite tips of Ndr2 PC12 cells compared to control cells (two-way ANOVA genotype effect: F_(1,36)= 7.407 p < 0.01), regardless of the coating substrate (genotype × substrate interaction: F_(1,36)= 0.1485, p = 0.7022; Figure 2G). Furthermore, while collagen IV substrate decreased the growth cone size of control cells, Ndr2 PC12 cells did not change their growth cone size upon collagen IV coating (genotype × substrate interaction: F_(1,35) = 15.67, p = 0.0004; Figure 2H). However, β₁^pThr788/789 raw fluorescence results were independent of the growth cone sizes, since fluorescence intensity of β₁^pThr788/789 per μm² of growth cones revealed only a genotype difference between EGFP and Ndr2 PC12 cells as in raw fluorescence results (two-way ANOVA genotype effect: F_(1,36)= 4.867, p = 0.0338; data not shown).

α₁β₁ integrin heterodimers are the main receptors on the cell surface that respond to collagen IV substrate and α₁β₁ integrins show increased binding to collagen IV, rather than collagen I, during neurite extension of human neuroblastoma cells (Carmeliet et al., 1994). To identify the recognition process that underlies the growth deficiency on collagen IV, we tested the effect of integrin stimulation via Mg²⁺ and a KTS ligand specific to α₁β₁ integrins on PDL or Collagen IV substrates. Here we observed a general enhancement of outgrowth in the Ndr2 PC12 cells by Mg²⁺, but a complete failure to increase growth by the α₁β₁ integrin ligand. A significant interaction was found between Ndr2 expression and integrin stimulation (Figure 3, genotype × treatment interaction: F_(6,24)= 41.05 p < 0.001).

As in the previous experiment a significant enhancement of neurite growth by collagen IV was observed in EGFP PC12 control cells, but not by Ndr2 PC12 cells upon mere NGF stimulation. Under 0.3 mM Mg²⁺ treatment along with NGF, EGFP PC12 cells slightly further increased growth on collagen IV, while Ndr2 cells increased growth significantly regardless of the substrate, but remained lower than control cells on collagen IV. To specifically address α₁β₁ integrin function we synthesized a KTS ligand peptide (CWKTSLTSHYC), derived from the larger disintegrin obtustatin (Marcinkiewicz et al., 2003; Moreno-Murciano et al., 2003) and applied it to the cell culture medium. At the concentration used in our experiments (140 μM; based on Marcinkiewicz et al., 2003), this KTS ligand did not induce any change in the flattening of the cells that would indicate a generally altered adhesion (Kruskal-Wallis test between treatments: p = 0.93), but a maximal neurite growth of PC12 control cells on both PDL and collagen IV was observed. As a side note, under 140 μM obtustatin treatment, no significant difference was observed between control cells plated on PDL and Collagen IV, suggesting that obtustatin peptide did not cause any inhibition on Collagen IV substrate. In Ndr2 PC12 cells, however, in spite of the increased expression of β₁ integrin phosphorylation (Thr^788/799) in the neurite tips (Figure 2G), the KTS agonist peptide like collagen IV entirely failed to induce neurite growth (Figure 3).

As those data indicated a specific impairment of α₁β₁ integrin function in Ndr2 PC12 cells, next we examined the expression and distribution of the α₁ integrin subunit during differentiation (Figures 4A–F). In fact, we observed that Ndr2 PC12 cells show significantly less α₁ integrin labeling at their neurite tips than EGFP PC12 cells (Figure 4G, two-tailed Student’s t-test, p < 0.001). At the same time, labeling for F-actin was not different in the neurite tips. To distinguish whether Ndr2 regulates the α₁ integrin trafficking or its overall expression, we also determined the total α₁ integrin expression in EGFP PC12 and Ndr2 PC12 cells (Figures 4H,I). Following either 3 days or 6 days of NGF treatment, a profound increase in α₁ integrin levels was observed that was as pronounced in Ndr2 PC12 cells as in EGFP PC12 controls, similarly on PDL (two-way ANOVA, genotype effect: F_(1,18)= 0.2754, p = 0.6062) and Collagen IV (two-way ANOVA, genotype effect: F_(1,18)= 1.74, p = 0.2036) substrates (Figures 4J,K).

To investigate the role of Ndr2 in neuronal differentiation, we acutely transfected primary hippocampal neurons with Ndr2 constructs (overexpression and silencing) on different substrates. Neurons were plated on coverslips that are coated either with PDL or Laminin-111 (LN111), a previously established ECM molecule of primary neurons and a known α₁β₁ integrin substrate (Desban et al., 2006; Tulla et al., 2008). When control neurons (EGFP hippocampal primary culture (HPC)) were plated on LN111, their dendritic branching significantly increased compared to PDL (two-way ANOVA substrate effect: F_(1,57)= 12.59, p = 0.0008; Figures 5A–C). However, Ndr2 overexpressing neurons (Ndr2 HPC) showed less dendritic branching on LN111 compared to PDL (two-way ANOVA substrate effect: F_(1,57)= 15.46, p = 0.0002; Figures 5D–F), suggesting an impairment of α₁ integrin dependent growth, in line with our PC12 cell results. Furthermore, analysis of the total dendritic length of these acutely-transfected neurons revealed that Ndr2 overexpression increases total dendritic length on PDL, while it reduces the total dendritic length on LN111 compared to the control neurons (one-way ANOVA: F_(3,115) = 7.72, p < 0.0001; Figure 5G).

Finally, we tested how silencing of Ndr2 kinase in neurons affects the morphology of dendrites. shLuc transfected neurons (shLuc HPC) serving as controls increased their dendritic branching upon LN111 substrate as expected (two-way ANOVA substrate effect: F_(1,54)= 12.16, p = 0.0010; Figures 6A–C). By contrast, Ndr2 knockdown neurons (shNdr2 HPC) entirely failed to increase their growth in response to LN111 substrates (two-way ANOVA substrate effect: F_(1,58) = 1.262, p = 0.2660; Figures 6D–F). In fact, total dendritic lengths of shNdr2 neurons were significantly shorter than shLuc neurons on both PDL and LN111 substrates (one-way ANOVA: F_(3,113)= 17.33, p < 0.0001; Figure 6G).