Cell Tracking of Adult Neural Stem Cells From The Subependymal Zone In Vitro Reveals Cell Type-Specic Impacts On Cell Cycle Progression By The Extracellular Matrix Molecule Tenascin-C

Adult neurogenesis has been described in two canonical regions of the adult central nervous system (CNS) of rodents, the subgranular zone (SGZ) of the hippocampus and the subependymal zone (SEZ) of the lateral ventricles. The stem cell niche of the SEZ provides a privileged environment composed of a specialized extracellular matrix (ECM) that comprises the glycoproteins tenascin-C (Tnc) and laminin-1 (LN1). In the present study, we investigated the function of these ECM glycoproteins in the adult stem cell niche. Adult neural stem cells (aNSPCs) of the SEZ were prepared from wildtype (Tnc +/+ ) and Tnc knockout (Tnc -/- ) mice and analyzed using molecular and cell biological approaches. A delayed maturation of aNSPCs in Tnc -/- tissue was reected by a reduced capacity to form neurospheres in response to epidermal growth factor (EGF). In order to examine a potential inuence of the ECM on cell proliferation, aNSPCs of both genotypes were studied by cell tracking using digital video microscopy. aNSPCs were cultivated on three different substrates, namely poly-D-Lysine (PDL) and PDL replenished with either LN1 or Tnc for up to six days in vitro. On each of the three substrates aNSPCs displayed lineage trees that could be investigated with regard to cell cycle length. The latter appeared reduced in Tnc -/- aNSPCs on PDL and LN1 substrates, less so on Tnc that seemed to compensate the absence of the ECM compound to some extent. Close inspection of the lineage trees revealed a subpopulation of late dividing aNSPCs late that engaged into cycling after a notable delay. aNSPCs late exhibited a clearly different morphology, with a larger cell body and conspicuous processes. aNSPCs late reiterated the reduction in cell cycle length on all substrates tested, which was not rescued on Tnc substrates. When the migratory activity of aNSPC-derived progeny was determined, Tnc -/- neuroblasts displayed signicantly longer migration tracks. This was traced to an increased rate of migration episodes compared to the wildtype cells that rested for longer time periods. We conclude that Tnc intervenes in the proliferation of aNSPCs and modulates the motility of neuroblasts in the niche of the SEZ. time series the identical eld of view. (A-C) Phase contrast images of wildtype cells in all three conditions. The cells generally showed good adherence and survival rates. Proliferation events were visible after 72 h, when rst cell assemblies could be observed. After 140 h, numerous cells appeared branched. Cultivation on Tnc coated PDL also led to cell proliferation. However, increased cell loss and the tendency to cell aggregation with diminished adhesion to the culture plate were observed in this condition. (D-F) Tnc-decient cells also showed satisfying adhesion in the control condition and fast proliferation within 72 h. In the next three days, cells on LN1 and Tnc started to cluster, which reduced the adhesion to the culture plate (E’’ and F’’). Scale bar: 150 µm. (G-L) Exemplary pedigrees of individual mother cells in each condition monitored by lineage tracking are shown. Each lineage tree traces the proliferation of an initial mother cell to map the whole progeny that was generated within 140 hours of time-lapse video microscopy. Every branching point marks a cell division event and the length of the branches is proportional to the corresponding cell cycle duration. An ‘X’ marks the death of a daughter cell. Analyses of gene expression in the lateral ventricular wall tissue from 10-week old mice. (A) Investigation of protein levels by Western blot technique. The analysis comprised ve wildtype samples (N=5) and three Tnc-decient samples (N=3), as conrmed by pk anti-Tnc staining. The levels of GFAP and Sox2 in each sample were normalized to α-tubulin. (B) Quantication of GFAP signals demonstrated strongly decreased levels in Tnc-decient SEZ tissue of animals. (C) Examination of Sox2 signals showed a slight increase of Sox2 expression levels in the knockout tissue. (D-E) Analysis of signaling pathways in the lateral ventricular wall tissue from 10-week old mice. Quantication of pERK signals demonstrated strongly increased phosphorylation levels in Tnc-decient SEZ tissue (D). Notch intracellular domain (NICD) expression in each sample was related to the corresponding Notch1 level (wildtype N=5, knockout N=3). The ratio reected an unaltered Notch pathway in knockout tissue (E). (F) Analysis of mRNA levels by RT-PCR suggested an implication of the Tnc-knockout in the Notch pathway activation and cell cycle progression. Tnc-decient mice at 10 weeks of age showed an increased expression level of the target gene Hes5 in tissue isolated from the lateral ventricular wall (N=4). (G) mRNA samples isolated from E15 neurospheres after 7 days in vitro also displayed increased Hes5 expression in Tnc-decient cells (N=4). (H) E15 neurospheres after 7 days of cultivation in suspension were tested for differences in the expression level of the cytoskeleton associated protein 2-like (Ckap2l) that encodes for a mitotic spindle protein. Ckap2l expression in Tnc-decient neurospheres was reduced compared to wildtype neurospheres (N=4). Data show mean ± s.e.m.


Introduction
The adult mammalian brain retains neural stem cells in closely de ned areas, the so-called stem cell niches at the lateral ventricle and the subgranular zone of the dentate gyrus [1,2]. The stem cells from the subependymal zone (SEZ) of the lateral ventricle are surrounded by niche astrocytes, blood vessels and their own generated progeny underneath a dense layer of ependymal cells facing the lumen of the ventricle [3][4][5]. These stem cells have astroglial characteristics and are also called SVZ astrocytes or type B cells [6,7,5]. They give rise to type C cells, which are fast-cycling transient amplifying progenitors (TAPs) and in the end produce type A cells which are neuroblasts generated by a nal symmetric division [8,9]. These newborn neurons are integrated into the neuronal network of the olfactory bulb, which they reach by migration through the rostral migratory stream [10].
At the beginning, adult neural stem cells (aNSPCs) were investigated in vitro in neurosphere cultures [11] or in adherent cultures by addition of growth factors or astroglial feeder layers [12,13]. Subsequent studies could show that the addition of such factors considerably in uences the proliferation mode of aNSPCs [14,9]. On these bases, protocols were developed that allow to study the behavior of aNSPCs from the SEZ in absence of mitogenic or cell fate-changing factors [8]. This approach unraveled the intrinsic program of lineage progression and opened that possibility to investigate extracellular matrix (ECM) effects by presenting de ned niche molecules in vitro [9,8].
The ECM has been shown to be of great importance for the behavior of diverse cell types from different tissues [15][16][17]. The glycoprotein tenascin-C (Tnc) is part of the ECM in stem cell niches of the central nervous system in development and adulthood [18][19][20][21][22][23]. The close association of Tnc with stem cell environments prompted the analysis of Tnc impact on stem cell behavior during development [24][25][26].
The use of Tnc-knockout cells [26] as well as the cultivation on puri ed Tnc substrates [27] or the Tncderived peptide VSWRAPTA [28] could con rm the hypothesis that Tnc does strongly in uence processes such as cell cycle progression, differentiation and neurite growth in speci c contexts of early development.
Another important member of the ECM are laminins as part of the basal lamina. They are expressed in the subventricular zone of adult mice [29] and a promoting in uence of laminins on cell behavior has already been reported for myoblasts [30] or neural progenitor cells [31,32] in terms of proliferation and neurogenesis. We present here a study that investigates the in uence of the two extracellular matrix compounds Tnc and laminin-1 on the lineage progression of aNSPCs from the SEZ by time-lapse video microscopy and cell tracking to obtain deeper insights into the direct in uence of the niche environment on adult stem cell behavior under speci cally de ned conditions.

Material And Methods
Legal issues and animal housing The present study was carried out in accordance with the European Council Directive of September 22, 2010 (2010/63/EU) for care of laboratory animals and approved by the animal care committee of North Rhine-Westphalia, Germany, based at the LANUV (Landesamt für Umweltschutz, Naturschutz und Verbraucherschutz, Nordrhein-Westphalen, D-45659 Recklinghausen, Germany). The study was supervised by the animal welfare commissioner of Ruhr-University. All efforts were made to reduce the number of animals in the experiments. Embryos of both sexes were used. Data of this publication are based on experiments performed with Mus musculus with tenascin-C de ciency. The systemic knockout has initially been generated and published by Forsberg and colleagues [33]. The Tnc −/− knockout line was compared to wildtypes from the 129sv mouse strain that shares the same genetic background. Handling of the animals was conducted according to German animal protection laws and Federation for Laboratory Animal Science Associations (FELASA) standards. Mice were kept under controlled conditions with a 12-hour light/dark cycle and had access to water and food ad libitum. Adult animals at the age of 9-to 11-weeks were anesthetized with Iso uran (3% (v/v), CP-Pharma) and sacri ced by cervical dislocation. The genotypes were con rmed with PCR using genomic DNA from tail biopsies with the Page 4/31 following primers (forward: CTGCCAGGCATCTTTCTAGC; reverse: TTCTGCAGGTTGGAGGCAAC; neo reverse: CTGCTCTTTACTGAAGGCTC).
Isolation and culture of adult neural stem cells from the murine subependymal zone The following isolation protocol was adapted from a published procedure [8]. 9-to 11-week old mice were sacri ced as described above. After decapitation, the brain was dissected and transferred into a 10 cm petri dish lled with ice-cold HBSS (#24020-091 Gibco, Darmstadt, DE) with 10 mM HEPES (#H0887-100 Sigma, Steinheim, DE), pH 7.4. The brain was split into its two hemispheres and cut coronally behind the optic chiasm with a surgical blade. Uncovering the ventricle from the caudal side using forceps, the lateral wall of the ventricle becomes visible. Ventricular wall tissue was dissected carefully and as thin a sheet as possible to avoid contamination with striatal tissue or myelin from the corpus callosum. The tissue was placed into a 15 ml tube containing 10 ml of fresh dissection medium (see above) until all brains were dissected. Afterwards, the medium was replaced by 5 ml of dissociation medium (HBSS with 15 mM HEPES and 0.54% (w/v) D-(+)-glucose (#D1349 AppliChem, Darmstadt, DE) supplemented with 6.7 mg trypsin (#T9201-100 Sigma) and 3.3 mg hyaluronidase (#H3884-50 Sigma) and incubated for 15 min at 37°C. The tissue was gently triturated with a 5 ml pipette up to 10 times and then incubated for another 15 min at 37°C. Tissue dissociation was stopped by adding the same volume of EBSS (#24010-043 Gibco) with 20 mM HEPES and 4% (w/v) BSA (#A9418 Sigma). The cell suspension was mixed gently before it was passed through a 70 µm cell strainer (#2350 Falcon). Cells were centrifuged at 200 g for 5 min. After aspirating the supernatant with a Pasteur pipette, cells were re-suspended in 10 ml of icecold HBSS with 0.9 M saccharose (#S/8600/60 Fisher Chemicals). Cells were centrifuged again for 15 min at 450 g. The resulting cell pellet was re-suspended in 2 ml ice-cold EBSS with 20 mM HEPES and 4% (w/v) BSA. 5 ml of the same solution were lled into a fresh 15 ml tube. The 2 ml cell suspension was gently applied to the top of the fresh solution. A nal centrifugation step at 250 g for 9 min was carried out, followed by re-suspension of the pellet in culture medium containing DMEM/F12 (#11320-074 Gibco) with 1x B27 (#17504-044 Gibco), 1x penicillin/streptomycin (#P4333 Sigma) and 8 mM HEPES. The yield of one entire brain was seeded onto a poly-D-lysine (20 µg/ml, #P0899 Sigma) coated well of a 24-well culture plate. As optional treatment, 24-well culture plates were coated further with 10 µg/ml Laminin-1 (LN1) (#354259 Corning) or 25 µg/ml Tenascin-C (non-commercial, self-produced as described [34]). The cells were incubated at 37°C and 5% (v/v) CO2.

Time-lapse video microscopy
For the detailed investigation of adult neural stem cell proliferation behavior, the 24-well plate was placed into the closed system of an Axiovert 200M equipped with an AxioCam HRm and AxioVision-4.8.1 software (all from Carl Zeiss, Oberkochen, DE). Additionally, the two controlling elements 'Tempcontrol 37 − 2 digital' and 'CTI-Controller 3700 digital' (PeCon GmbH, Erbach, DE) were used for stable temperature and pH conditions, respectively. Conditions were adjusted 1 h before use. Over a period of 6 days, 8 visual elds per well were documented every 5 min with de ned XYZ-coordinates controlled by a moving stage (Märzhäuser, Wetzlar, DE) to ensure a sharp focus over the whole period of time. Thereby, a stack from about 1680 single images can be combined to obtain a time-lapse video with detailed information about the cell behavior in vitro.

Cell lineage tracking
To follow the lineage of a single mother cell, the obtained stack from time-lapse video microscopy was loaded into Fiji software. First, the visual eld was scanned for proliferation events. When there was a proliferative cell, it was tracked from the beginning and each time point of cell division was noted with respect to the direct precursor. With this information, a lineage tree could be drawn that gave precise evidence for the cell cycle length of each generation. Each bifurcation in this tree means a divisional event and the length of the vertical lines corresponds to the duration of the observed cell cycle. Thus, each mother cell establishes its own individual lineage tree. Every dividing cell from ve biologically independent experiments was included into the analysis (N = 5; wildtype n = 497 and knockout n = 350).

Migration analysis
Another read-out of the time-lapse videos focused on the migratory behavior of the generated neuroblasts. In most cases, neuroblasts arose from the nal division of a mother cell in this culture system, which could be identi ed by a bipolar morphology and the gain of migratory properties. With Fiji software, cell movement could be evaluated by using the 'Manual Tracking' Plugin. Parameters like distance (in arbitrary units, AU) and migration time (in min) were directly analyzed, while the speed parameter was calculated as distance per time (in AU/min).

Neurosphere assay
To calculate the number of stem cells within the wildtype and knockout SVZ, tissue was dissected and digested as described above and then seeded into T25 asks with a density of 750 cells/ml in DMEM/F12 medium supplemented with B27, penicillin/streptomycin, 8 mM HEPES and epidermal growth factor (20 ng/ml, #100 − 15 PeproTech, Hamburg, DE). Without a coated surface, the stem cells grow in suspension and build neurospheres upon proliferation. After cells were incubated at 37°C and 5% (v/v) CO2 and allowed to grow for 7 days, they were xed by the addition of 1% (w/v) PFA into the culture medium. The total number of spheres per ask was counted under an inverted cell culture microscope with the help of a grid. As the cells were seeded in clonal density, one sphere was expected to originate from one stem cell and should not result from fusion events due to agitation of the cultivation medium. Thereby, this method was used as an in vitro approach to determine stem cell numbers in both genotypes.
mRNA expression analysis via RT-PCR To isolate mRNA from the tissue, three pieces of lateral ventricular wall tissue were pooled for one sample and covered with 250 µl lysis buffer supplemented with 2.5 µl ß-mercaptoethanol from the GenEluteTM Mammalian Total RNA Miniprep Kit according to manufacturer's instructions (#RTN350 Sigma). An oncolumn DNAse digestion step was inserted (#DNASE70 Sigma). After elution with 30 µl H2O, RNA concentrations were measured with a photometer (Hellma Analytics). Puri ed mRNA was used for cDNA synthesis. If possible, 0.5 µg RNA was taken for one reaction. When concentrations were minute, the whole yield of mRNA was used into the synthesis reaction. Manufacturer's instructions for the First Strand cDNA Synthesis Kit were followed (#K1612 Thermo Scienti c). The RT-PCR method was used to analyze changes in gene expression levels between the wildtype and Tnc knockout mice. The regulation was estimated in a semi-quantitative manner by setting amplicon levels in relation to a housekeeping gene that does not underlie regulation in the corresponding tissue. ß-actin was generally used as reference gene. For detailed information concerning the used primers please see the supplementary material section (supplemental table S3).

Protein expression analysis via Western blot
For protein analysis, tissue was lysed with 100 µl of protein lysis buffer (50 mM Tris/HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 5 mM EGTA, 1% (v/v) Triton-X 100, 0.1% (v/v) Na-deoxycholate, 0.1% (v/v) SDS, 1 mM orthovanadate, 40 mM sodium uoride) supplemented with 10 µl/ml PMSF (#195381 MP Biomedicals) and 10 µl/ml Aprotinin (#10 981 532 001 Roche) as protease inhibitors. After mechanical trituration of the tissue, samples were incubated for 15 min on ice followed by centrifugation at 16.000 rpm for 15 min at 4°C. The supernatant was transferred to a new tube and concentrations were determined with PierceTM BCA Protein Assay Kit (#23225 Thermo Scienti c). A protein amount of 10 µg was loaded onto a 10% (v/v) polyacrylamide gel. Gels ran with a current of 15 mA for 1 h. Proteins were transferred to a PVDF membrane by applying a current of 75 mA for 1 h. Membranes were blocked with 5% (w/v) milk powder (Heirler, Radolfzell, DE) in TBS buffer (10 mM Tris/HCl pH 7.4 with 150 mM NaCl) for 2 h, before primary antibodies were diluted in blocking buffer and incubated overnight at 4°C. After three cycles of washing with TBST (TBS with 0.05% Tween-20 (VWR Chemicals, Darmstadt, DE)), membranes were covered with secondary antibodies diluted in blocking buffer for 1 h and rotated on an orbital shaker.
Before the blot was developed, membranes were washed again for three times with TBST and once with TBS to remove Tween residues. Afterwards, both solutions of the Clarity Western ECL Substrate (#1705061 BioRad) were mixed in equal parts and applied onto the membrane to incubate for 5 min. The solution was drained from the membrane to document chemiluminescent signals using the MicroChemi (DNR Bio-Imaging systems).

Differentiation assay and immunocytochemistry
To reveal differences in the cell fate of adult neural stem cells driven by the different extracellular matrix components which they were exposed to, cells were incubated for 7 days at 37°C and 5% (v/v) CO2. Within this time period, the progeny generated by stem cells in the culture was able to differentiate to neuroblasts (Dcx+), astrocytes (GFAP+) or oligodendrocytes (O4+). To stain the culture after 7 div, cells were xed with 4% (w/v) PFA for 10 min. After three washing steps with PBS for 5 min, cells were incubated with the primary antibodies at 4°C over night (mouse IgM anti-O4 (RRID: AB_94872; 1:30), rabbit anti-GFAP (RRID: AB_477010; 1:300) and mouse anti-bIII-tubulin (RRID: AB_477590; 1:300)) and 0.5% (v/v) Triton X-100 (#A4975 AppliChem). After three washing steps with PBS for 15 min each, secondary antibodies were diluted in the same buffer as primary antibodies and incubated for 1 h at room temperature (goat anti-mouse IgM AF488 (RRID: AB_2338849), goat anti-mouse IgG Cy3 (RRID: AB_2338686) and goat anti-rabbit IgG Cy5 (RRID: AB_2338013), all 1:300 from Dianova; supplemental table S4). Coverslips were placed in a drop of ImmuMount on an object slide with the cells facing downwards. Stainings were documented at an Axiophot (Zeiss).

Tissue xation and preparation of cryosections
For immunohistochemical stainings, the brain was taken out of the skull and directly stored in 4% (w/v) PFA at 4°C for 48 h. This xation was followed by a sucrose gradient: brains stayed at 10% (w/v) sucrose for 6 hours, 20% (w/v) sucrose for 24 hours and 30% (w/v) sucrose for another 24 hours, before they were embedded in tissue freezing medium (Leica Microsystems, Wetzlar, DE) and placed onto dry ice to ensure fast freezing of the tissue. Cryosections were cut at the Leica Cryostat CM3050S with 14 µm thickness .

Immunohistochemistry
Sections were thawed and allowed to dry. The tissue was encircled with a hydrophobic barrier marker (ROTI ®Liquid Barrier marker, Carl Roth, Karlsruhe, DE). Sections were rehydrated in PBS with 1.7% (w/v) NaCl (Fisher Chemical) for one hour. Afterwards, sections were covered for 2 h with blocking buffer containing 10% (v/v) normal goat serum (Dianova), 1% (w/v) BSA (Sigma) and 0.1% (v/v) triton X-100 (AppliChem) in PBS. Primary antibodies were diluted in blocking buffer and incubated overnight at 4°C.
After three washing steps with PBS for 15 min each, secondary antibodies were diluted in blocking buffer and applied to the sections for 2 h at room temperature. Another three cycles of washing with PBS were conducted before the sections were covered with ImmuMount (Thermo Scienti c) and a 24x50 mm coverslip. Stainings were documented at the AxioZoom V16 (Carl Zeiss).

Statistical analysis
Results in bar charts are depicted as mean ± s.e.m., unless otherwise speci ed. Results that are presented as Box-Whisker-Plots show the median as line, the upper and lower quartile as box and the 5-95 % percentile as whiskers. The statistical test and the number of performed experiments are indicated in the gure legends, where biological replicates are referred to as "N" and technical replicates are referred to as "n". The Kolmogorov-Smirnov test was used to verify Gaussian distribution. Statistical tests were done with the GraphPad Prism 7 software (Graphpad Software, San Diego, CA). Signi cances were graphically illustrated by * for p < 0.05, ** for p < 0.01 and *** for p < 0.001.

Results
Expression pattern of Tnc in the adult murine brain In the course of murine development, Tnc expression reaches a maximum right before birth to be strongly reduced afterwards [18]. Nevertheless, an immunohistochemical staining of frontal sections from wildtype mice at the age of 10 weeks (Fig. 1A) with polyclonal anti-Tnc revealed a restricted but distinct expression of Tnc around the lateral and the third ventricle (Fig. 1B). The speci city of the antibody was con rmed by controls with Tnc-de cient brain sections where staining was completely absent (Fig. 1B').
The close association of the secreted Tnc protein with stem cell niches in embryos that is maintained in the adult stage led to the assumption that this ECM component provides a bene cial environment for stem cell maintenance [35,25].
Tnc-de cient aNSPCs display a reduced response to EGF in vitro In order to probe the asserted functional role, a classical experimental approach was used to determine differences in aNSPC numbers between wildtype and Tnc-de cient mice in vitro. The clonal density assay was performed with single cells isolated from the lateral ventricular wall that were cultivated in suspension cultures in presence of the cytokine epidermal growth factor (EGF). After 7 days in vitro, the number of generated neurospheres was evaluated relatively to the sum of plated cells. Cells were seeded in clonal density to ensure that each neurosphere results from the progeny of a single stem cell. Tncde cient cells produced signi cantly less neurospheres after 7 days than wildtype cells (15.35% ± 1.2% neurosphere-forming cells in wildtype cultures compared to 9.32% ± 1.2% in knockout cultures, N = 5, p = 0.007 in unpaired students t-test, Fig. 1C).
The reduced capacity to form neurospheres could result from changes of EGF receptor (EGFR) expression. Therefore, the EGFR level in SEZ tissue in vivo was analyzed using PCR and Western blot. The expression of EGFR on mRNA level normalized to b-actin showed a decrease in Tnc-de cient animals (1.06 ± 0.39 compared to 1.60 ± 0.46 in wildtypes, mean ± s.e.m., N = 4, Fig. 1D). The protein expression paralleled this result, as Tnc-de cient tissue seemingly displayed reduced EGFR levels compared to the control (0.67 ± 0.20 in knockout SEZ (N = 3) compared to 1.13 ± 0.13 in wildtype SEZ (N = 5); however, not signi cant, p = 0.14 in Mann-Whitney U-test, Fig. 1E). In conclusion, the aNSPC population of the SEZ of Tnc-knockout mice exhibited a tampered proliferative capacity in response to EGF. The role of the EGFR in this context remains to be established.
Cell lineage tracking of adult NSPCs by time-lapse video microscopy permits measurement of cell cycle length A detailed evaluation of cell cycle duration in wildtype and Tnc-de cient aNSPCs was conducted with the help of time-lapse imaging over a period of 6 div. The single cells were plated onto PDL coated control dishes or to substrates additionally replenished with laminin-1 (LN1) or tenascin-C (Tnc). Each aNSPC mother cell constructed its own individual pedigree depending on the number of divisions and the length of the cell cycle (exemplary lineage trees for each condition are mapped in Fig. 2G-L). A striking attribute of lineage trees generated by aNSPCs was their synchronized sequence (Supplemental movies 1-4).
Daughter cells mostly divided again after comparable cycle lengths. Note that the mapped lineage trees all showed three rounds of division. This happened in 10% of the mother cells (data not shown), but served better for illustration of the proliferation dynamics. Most of the generated trees were slenderer and displayed fewer branching points. This mapping provided a good overview of the sequential divisions of a single mother cell, but it did not yield su cient information that permitted precise quanti cation and statistical evaluation. For this reason, the complete data set relating to cell cycle lengths was assembled in box-whisker-plots.

Tracking of adult NSPCs by time-lapse video microscopy reveals differences between Tnc and LN1 substrates
Previous studies using this tracking approach already revealed signi cant differences in the cell cycle lengths of subsequent generations of embryonic spinal cord NSPCs [26,8]. Therefore, each generation was considered separately. The rst generation represented a peculiar condition due to the fact that the beginning of the cycle could not be determined with precision as it laid outside of the observed time frame. This accounted for a broader distribution of the data points and no signi cant differences were hence observed for the rst generation of wildtype aNSPCs between PDL (median of 9.7h, n = 171, N = 5), So far, the study was focused on the effect of supplemented ECM-glycoproteins presented as substrates in culture. It is also of interest to consider the behavior of wildtype and knockout aNSPCs under identical culture conditions to evaluate the relative contribution of Tnc genes. Because of the impossibility to x the time point of the rst division mentioned previously, no signi cant changes could be traced in the rst generation (Fig. 3A). In the second generation, Tnc-de cient aNSPCs cycled signi cantly faster on PDL (18.7h, n = 117, N = 5, Fig. 3B) compared to wildtype aNSPCs (21.8h, n = 157, N = 5, p < 0.001). An analogous phenomenon was also seen on LN1 (wildtype: 18.1h, n = 106, N = 5; knockout: 21.0h, n = 176, N = 5, p = 0.003). The result on the Tnc substrate was not signi cant in the second generation (wildtype: 19.3h, n = 145, N = 5; knockout: 17.6h, n = 103, N = 5, p = 0.11; Fig. 3B). In the third generation, cell cycle lengths did not show signi cant alterations on PDL (wildtype: 17.2h, n = 81; N = 5; knockout: 17.0h, n = 106, N = 5) and Tnc (wildtype: 16.6h, n = 67; N = 5; knockout: 16.1h, n = 96; N = 5, p = 0.15; Fig. 3C). On LN1, the decrease in cell cycle length of Tnc-knockout cells was signi cant (16.3h, n = 79; N = 5; Fig. 3C) compared to the wildtype (16.8h, n = 125, N = 5, p = 0.041). The situation in the fourth generation was comparable in that Tnc-de cient aNSPCs (18.3h, n = 54, N = 5) divided faster than the wildtype on PDL (25.3h, n = 21, N = 5, p = 0.007; Fig. 3D). The outcome was comparable on the LN1 substrate (knockout: 15.0h, n = 29, N = 5; wildtype: 17.0h, n = 57, N = 5, p < 0.001; Fig. 3D). On the Tnc coating, however, the difference was not signi cant (knockout: 15.8h, n = 52, N = 5; wildtype: 16.3h, n = 7, N = 5, p = 0.12; supplemental table S1). Taken together, both the addition of Tnc as culture substrate and, alternatively, its depletion by genetic recombination in the knockout resulted in alterations of the cell cycle length in aNSPCs. Based on these results we proposed the hypothesis that the Tnc concentration is subtly balanced in the niche. Consequently, its dysregulation unsettled the proliferation behavior of aNSPCs in vitro.
Identi cation of a subpopulation of slowly dividing aNSPCs using time-lapse video microscopy When the proliferation activity of aNSPCs was examined in more detail, a subset of proliferating cells appeared to engage remarkably late into cell division during the observation period of 6 days. This subset remained notably inert while the majority of proliferating cells performed two rounds of mitosis. This resulted in a remarkable extension of the temporal delay until the rst round of mitosis that was clearly visible when the temporal sequences until the rst division were plotted for the individual cells monitored in the cell tracking experiment (Fig. 4A). Whereas the majority of proliferating cells started their rst division within the rst 24h of tracking, a minority stalled for up to 48h. In order to distinguish this subpopulation, a threshold was set at 50h, following the assumption that a subclass of slowly dividing aNSPCs would cluster above this hypothetical barrier. This procedure was supported by the observation that early dividing aNSPCs exhibited a different morphology compared to late dividing ones (Fig. 4B). The rapidly dividing cells were characterized by a rounded small cell body, while the late dividers presented several elongated processes emanating from a cell body that was nearly twice as large (top). An exemplary scheme of both cell morphologies was added to the phase contrast images. The durations of the rst completed cell cycle were compared for the rapidly dividing as compared to the late-dividing population. On PDL, the groups differed signi cantly as early dividing cells needed 21.8h (median, n = 157, N = 5; Fig. 4D 4C). We propose that the late dividing aNSPCs correspond to the slow-dividing astro/radial glia that has been described previously [9]. This cell type is placed at the top of a lineage tree and generates complex pedigrees. Most of these trees are characterized by the occurrence of asymmetric cell divisions and give rise to neuronal as well as astroglial progeny. In this sense they have been perceived as genuine aNSPCs, different from already fate-restricted progenitors [8]. Because we studied this subpopulation in more detail, it will heretofore be referred to as aNSPCs late .
Examination of cell cycle lengths from the late-dividing subpopulation of aNSPCs The aNSPCs late subpopulation de ned by a prolonged inactive phase during the rst 50 hours of cell tracking was analyzed separately in more detail with regard to the behavior on different substrates ( Remarkably, the prolongation of the cell cycle on Tnc was not seen with knockout aNSPCs late , which may indicate that these cells have lost sensitivity to the ECM component. The asserted differences between wildtype and Tnc −/− aNSPCs late were followed further by direct comparison of the respective cell cycle lengths under the different conditions tested. Thus, in the second generation the knockout cells manifested a signi cant decrease of the cell cycle lengths on all substrates (p < 0.01 for each condition; Fig. 5A). In the third generation, no statistically signi cant differences could be discerned on the substrates PDL and LN1. In contrast, on Tnc the stretched cell cycle in the wildtype compared to Tnc −/− aNSPCs late was obvious (wildtype: 18.8 h, n = 24, N = 5; knockout: 14.7h, n = 23; N = 5, p < 0.001; Fig. 5B).
In the fourth generation the situation was analogous to the third generation. In particular, the signi cantly reduced cell cycle length of Tnc-de cient cells on a Tnc substrate persisted into the fourth generation (Fig.  5C). Taken together, this analysis indicated that Tnc −/− aNSPCs late deriving from Tnc-de cient stem cell niches could not respond to distinct ECM signaling cues in the same manner as their wildtype counterparts.

Functional analysis of aNSPC progeny in wildtype and Tnc-de cient cultures
In the adult subependymal zone aNSPCs give rise to transient amplifying precursor cells and, eventually, to neuroblasts [14]. The latter migrate to the olfactory bulb where they differentiate to dopaminergic interneurons [36]. When the generation of neuroblasts was rated in our culture system, the incidence after 6 div ranged between 11,6 ± 7,0% (mean ± SD, N = 3) and 16 Fig. 6C). In order to explain this difference of the extent of cellular itineraries, the durations of migratory activity were evaluated for both genotypes. It emerged that the Tnc-de cient neuroblasts were motile for a mean time sequence of 5088 ± 124 min (n = 32, N = 3), whereas wildtype neuroblasts moved for a mean of 3530 ± 206 min (n = 34, N = 3, p < 0.001; Fig. 6D) during the observation period. It appears that the wildtype cells paused signi cantly more often than the Tnc-de cient neuroblasts, which may reveal an anti-migratory effect of the glycoprotein on cell motility. While this functional trait of the generated neuroblasts was altered between the two genotypes, differentiation assays did not indicate a notable impact on cell fate (supplemental Fig. S1).
Comparison of signaling pathway in the SEZ of Tnc-knockout mice in vivo It was known from previous studies that Tnc intervenes in critical signaling pathways [18]. Therefore, protein samples collected from the lateral ventricular wall of 10-week old wildtype and knockout mice were analyzed by Western blotting (Fig. 7A). The astrocyte marker GFAP that is strongly expressed in type B stem cells [6] was decreased in Tnc-knockout samples (wildtype: 15.8 ± 5.8 AU, n = 5, N = 5; Tnc knockout: 5.8 ± 1.9 AU, n = 3, N = 3, p = 0.14; Fig. 7B). The stem cell marker Sox2 varied barely between wildtype and knockout (wildtype: 0.61 ± 0.06 AU, n = 5, N = 5; knockout: 0.73 ± 0.03 AU, n = 3, N = 3; Fig.  7C). Thus, the classical stem cell markers were negligibly affected by the depletion of Tnc from the aNSPC niche of the lateral ventricle. The same set of protein samples was also analyzed with emphasis on developmentally relevant signal transduction pathways. A particular attention was devoted to mechanisms that regulate cell cycle progression, as this process was modi ed by Tnc de ciency in the in vitro assays. The ERK signaling path was assessed by determining the level of phosphorylated ERK (pERK), as normalized to the corresponding total ERK1/2 (tERK) signal (wildtype: 2.6 ± 1.0 AU, n = 5, N = 5; Tnc knockout: 13.8 ± 4.7 AU, n = 3, N = 3, p = 0.07 in Mann-Whitney U test; Fig.  7D). The seeming difference of activation in the knockout did not, however, achieve signi cance (Fig. 7D).
Furthermore, the Notch/delta pathway was examined that is involved in the maintenance of the aNSPC compartment [37]. As Tnc affects the duration of the cell cycle, genes involved in its regulation were of interest. In a previous study using gene arrays, the expression level of some cell cycle-related genes was found modi ed in the spinal cord of embryonic Tnc-de cient mice [21]. The cytoskeleton-associated protein 2-like protein (Ckap2l/Radmis) is required for mitotic spindle formation and cell cycle progression in neural progenitor cells [39]. We found a reduction of Ckap2l mRNA expression in cortical neurosphere cultures of E15 Tncknockout mice (wildtype: 0.32 ± 0.09 AU, n = 4, N = 4; Tnc knockout: 0.19 ± 0.02 AU, n = 3, N = 3, p = 0.847; Fig. 7E'). From our results, we assume that Tnc de ciency has an impact on signaling pathways that are relevant for the regulation of cell proliferation and cell cycle progression. The detailed mechanism and participating proteins remain to be investigated in the future.

Discussion
Our results show that the glycoprotein Tnc has a strong in uence on the proliferation of aNSPCs obtained from the SEZ investigated by time-lapse video microscopy and lineage tracking in vitro. The cultivation of SEZ cultures in suspension revealed a reduced neurosphere formation capacity in Tnc-de cient cultures in response to EGF treatment. This could result from a decreased expression level of the EGFR documented on the mRNA and protein level. A link between Tnc and EGFR expression has been reported previously in embryonic neural progenitors, as mice lacking Tnc experienced a delayed acquisition of this receptor [25]. In adult mice, the presence of the EGFR discriminates between quiescent (EGFR-negative) and active (EGFR-positive) aNSPCs [40,41]. Thus, a reduced responsiveness to EGF might re ect a shifted proportion between these two subpopulations of aNSPCs and does not necessarily point towards a generally reduced stem cell pool in Tnc-de cient mice. This explanation is in agreement with an in vivo approach of stem cell quanti cation in the adult Tnc-knockout niche using BrdU labeling, where no striking changes were detected in comparison to Tnc +/− heterozygous mice [42].
A recent study reported a faster cell cycle progression in Tnc-de cient spinal cord progenitors isolated from E15 mice [26]. This nding obtains further support by our observation of shorter cell cycles also in aNSPCs from the SEZ in adult Tnc-knockout mice. The in uence of Tnc on cell proliferation is controversially being discussed, as Tnc elimination caused opposite results, depending on the experimental context and cell type [43,44]. The cultivation on a Tnc substrate accelerated the cell cycle in the majority of proliferating cells but elongated it in the subset of slow-dividing aNSPCs late . This subpopulation behaved as expected for genuine aNSPCs, distinct from fate-restricted progenitors [8].
Both proliferating populations may be considered as progenitors isolated at different stages of their respective lineage progression [9]. The opposite response to the ECM component Tnc might indicate differences in the receptor expression that eventually impact signaling mechanisms to alter cell cycle progression. This phenomenon was not noted in Tnc-knockout cultures where a compensation caused by coated Tnc failed. It seems that the absence of Tnc in the adult neural stem cell niche of knockout mice desensitized the aNSPCs for this compound. Another possible explanation would be a threshold effect, where the coating protein on the culture surface has to be supplemented by cell secreted Tnc protein to reach an effective concentration. It remains still elusive whether Tnc expression in the SEZ of adult mice has the function of maintaining the stem cell pool. A decreased amount of asymmetric cell divisions and accompanied increased percentage of neuroblast generation after 6 days in vitro (our unpublished observations) corroborates this hypothesis, because this would favor a faster depletion of the stem cell pool in Tnc-de cient populations. Furthermore, we can conclude from our data that Tnc-knockout aNSPCs late enter the cell cycle earlier than their wildtype counterparts, indicating that their activation happens faster. This strengthens the idea of Tnc as a promoter of the quiescent state. This effect could be revoked when Tnc-knockout aNSPCs late were plated onto Tnc coated dishes.
It has been described for several cell types that LN1 promotes cell proliferation and neurogenesis [30][31][32]. The accelerating effect on cell cycle progression was also observed in our experiments for the mitotic population of the primary SEZ cultures from wildtype and Tnc-knockout mice. For both ECM compounds of interest investigated in this study, integrins represent important receptors [45,46]. Thus, it is conceivable that the regulatory effects enacted by these substrate proteins are mediated by integrins. This most probably involves different heterodimeric receptors, as the reaction of slow-dividing aNSPCs late to the LN1 and Tnc substrates was not congruent. In support of this assumption, a distinct expression pattern of integrin variants during lineage progression has been reported [47].
Several reports suggest that Tnc has an impact on migration [48,49]. In particular in pathologic contexts such as cancer, a promoting effect on migration and metastasis has been attributed to Tnc [50,51], which might be mediated via integrins [52,53]. Our in vitro experiments with aNSPCs from the Tnc −/− de cient SEZ revealed increased migration trajectories of neuroblasts that were caused by elongation of movement episodes. This is reminiscent of increased migration of oligodendrocyte precursors observed in Tnc-de cient rats [24]. Tnc expression was found in the pathway of the rostral migratory stream [42] where neuroblasts migrate along to colonize the olfactory bulb and integrate into local neuronal networks [54]. The Tnc glycoprotein might have a guidance function there as it serves as inhibitory boundary molecule for olfactory sensory neurons [55]. Along these lines, early studies on Tnc in cell cultures characterized the glycoprotein as anti-adhesive [56,34]. Anti-adhesion was also seen in the videos generated in this study with respect to aNSPCs, although not as pronounced as for embryonic neural progenitors (unpublished observations). Changes of the intensity of substrate adhesion might be an indirect cause of the observed results and has been discussed as a potential mechanism to in uence migration and proliferation [57].
The altered GFAP protein level observed in in vivo SEZ samples points to structural changes in the niche of Tnc-knockout mice. GFAP is a marker that is expressed by niche astrocytes and aNSPCs. Therefore, both cell populations could be responsible for variations of GFAP expression. Minor structural changes have already been described by previous studies of the SEZ niche in Tnc −/− knockout animals [42]. The latter study for example described an increased number of neuroblast clusters around the lateral ventricle of Tnc −/− null mice. This is in agreement with our observation of a higher rate of neuroblast generation by Tnc −/− aNSPCs in vitro (E. Schaberg, unpublished observations).
In order to gain mechanistic insight, we investigated a potential intervention of Tnc in signaling pathways in the adult SEZ. Interestingly, the ERK pathway activity was altered upon Tnc knockout. This pathway is linked to the regulation of migration in numerous contexts [58], which may mechanistically explain the enhanced migration of knockout cells observed here. The increased rate of phosphorylated ERK may also be the cause for faster cell cycling as this pathway is known to regulate cell cycle progression and transition from G1 to S-phase [59,60]. Activation of the ERK1/2 pathway in primary embryonic cortical neurons by the synthetic biomimetic VSWRAPTA peptide derived from the primary Tnc sequence has been reported [28]. Taken together, these ndings suggest of the ERK1/2 pathway in mediating reactions to Tnc in the niche environment.
With regard to further downstream mechanisms related to cell cycle control, the cytoskeleton-associated protein 2-like protein (Ckap2l/Radmis) appeared reduced in Tnc −/− SEZ knockout tissue, as had been described in embryonic spinal cord based on transcriptome analysis [21]. The PCR result is supported by Western blot analysis of the EGF receptor, which veri ed a trend towards reduced expression also on the protein level (wildtype N=5, Tnc knockout N=3). Data are presented as mean ± s.e.m. For D-E, the Mann-Whitney U-test was used for statistical analysis.   was used for statistical analysis (* p<0.05, ** p<0.01).

Figure 5
Cell cycle lengths of the wildtype and Tnc-knockout subset of slow-dividing aNSPClate. (A-C) Cell cycle lengths of late-dividing wildtype (gray) and knockout aNSPClate (white) are presented in box-whiskerplots with 5-95% percentiles from the 2nd (A), 3rd (B) and 4th (C) generation (N=5, n=2-37). (A) In the 2nd generation, the cell cycle length was reduced in Tnc-de cient cells in the control, on LN1 and on Tnc substrates compared to their wildtype counterpart. (B) Exposure to Tnc as substrate led to a signi cantly elongated cell cycle in the 3rd generation of wildtype cells compared to the LN1 substrate. (C) This effect was maintained up to the 4th generation, which underscored the opposite effect of the Tnc substrate on the slow-dividing subset of aNSPClate in comparison to the majority of the proliferative population. In Tnc-de cient aNSPClate, we could not detect differences between the different substrate conditions in any generation. Extensive information concerning the number of analyzed cells is presented in supplementary table S3. Statistical evaluation was conducted using the Kruskal-Wallis test with Dunn's multiple comparison for comparing substrates (* p<0.05, ** p<0.01). The Mann-Whitney U tests were performed to analyze differences between wildtype and knockout data (** p<0.01, *** p<0.001). longer distance compared to the control (N=3, n=32). (C) Analysis of the velocity in AU per minute reveal that the observed changes in distance are not caused by a higher velocity of Tnc-knockout cells, as the speed range is equal in both groups (N=3, n=32). (D) Evaluation of the temporal sequence (min) shows a signi cant increase of the duration of migratory phases for Tnc-de cient compared to wildtype neuroblasts. Duration is measured from initiation of migration to the moment the neuroblast arrests and remains immobile (N=3, n=32). Data are presented as mean ± s.e.m. The unpaired two-tailed t-test was used for statistical analysis (*** p<0.001). Figure 7