All experiments were conducted in accordance with the approval of the Institutional Animal Care and Use Committee of Ajou University School of Medicine. The manuscript has been prepared following the ARRIVE guidelines. Mice of the Igf1r-knockout strain (B6.129-IGF-1r^tm1.2Mhz/Orl, #EM:00115) were purchased from European Mouse Mutant Archive (EMMA). Mice of the DsRed.T3 strain (B6.Cg-Tg(CAG-DsRed^*MST)1Nagy/J, #006051), which constitutively express the red fluorescent protein (RFP) variant DsRed.MST, were purchased from the Jackson Laboratory. To distinguish transplanted NSCs from resident cells in the spinal cord, Igf1r-heterozygous mice were crossed with the Ds.Red.T3 line. To generate Igf1r-heterozygous NSCs, heterozygous males were crossed with wild-type females; the resulting embryos were used to obtain NSCs. Mouse NSCs were cultured as previously described, with a slight modification for mice [23]. Briefly, the spinal cords of embryos at day 14 were dissected under a dissecting microscope (Carl Zeiss). The spinal cord tissue from each embryo was placed on a separate dish and incubated individually in Accumax dissociation medium (Thermo Fisher Scientific) at room temperature for 10 min. After centrifugation, dissociated cells were plated in StemPro® NSC serum-free culture medium (Thermo Fisher Scientific) containing basic fibroblast growth (bFGF, 20 ng/ml) and epidermal growth factor (EGF, 20 ng/ml). Immediately after plating, genotypes of the embryos were determined using REDExtract-N-Amp™ Tissue PCR Kit (Sigma). Subsequently, NSCs from the embryos of the same genotypes were collected and cultured together from then on.

Neurosphere assay was performed following a previously published protocol [25]. Tertiary neurospheres, which are formed after two passages, were used for measurement of the number and size of neurospheres. Secondary neurospheres from each group were collected and centrifuged, and the supernatant was removed. Then, Accumax dissociation medium was added, and the neurospheres were incubated for 5 min. Dissociated cells were plated at a density of 10⁵ cells per 100 mm petri dish. Culture medium was changed every day, with fresh bFGF and EGF added during the entire culture period. Neurospheres were allowed to grow for 7 days, after which the number and size of tertiary spheres were determined. Four independent cultures were used for the neurosphere assay. Neurospheres were counted under a confocal microscope (FV300, Olympus) at a 10× magnification, with a minimum cutoff diameter of 20 µm. Three images were randomly acquired from each plate, and the diameter of the neurospheres in these images was measured using the Image J software. For in vitro NSC survival assay, dissociated NSCs cultured in a 96-well plate were exposed to either 400 µM 3-morpholinosydnonimine (SIN-1, Enzo Life Sciences, Inc.) or 1 mM hydrogen peroxide (H₂O₂, Sigma), which produce reactive nitrogen species (RNS) or reactive oxygen species (ROS), respectively. Nitric oxide (NO)-associated radical species can kill transplanted NSCs in the injured spinal cord, and SIN-1 was used to produce RNS to damage cultured NSCs [26]. H₂O₂ has been frequently used to induce oxidative stress in stem cells [27,28]. Cultured NSCs were incubated in medium containing either SIN-1 or H₂O₂ for 48 h with or without IGF-1 at a concentration of 20 ng/ml. Three independent cultures were performed per experimental condition (N=3), with each culture replicated thrice. The percentage of surviving NSCs was determined using Cell Counting Kit-8 (Dojindo Laboratories) according to the manufacturer's instructions.

Nine-week-old C57BL/6 females (20~25 g) (Orient Bio), which had completed 1 week of treadmill pretraining (see below), were subjected to a contusion injury. Animals were anesthetized by intraperitoneal injection of ketamine (90 µg/g) and xylazine (10 µg/g) mixture. After laminectomy was performed, the dorsal surface of the spinal cord at the 9^th vertebral level was injured using the Infinite Horizon Impactor (Precision Systems and Instrumentation), with a force of 70 kdyn. Subsequently, animals were randomly divided into 4 groups depending on whether they received NSC grafts that were wild-type (+/+) or heterozygous (+/−) for Igf1r gene, or whether they underwent further TMT. Therefore, the 4 groups consisted of: 1) mice receiving +/+ grafts without TMT, 2) mice receiving +/+ grafts with TMT, 3) mice receiving +/− grafts without TMT, and 4) mice receiving +/− grafts with TMT. Transplantation of NSCs was performed 1 week after contusion injury. Immediately before transplantation, tertiary neurospheres were dissociated and resuspended at a density of 10⁵ cells/µl. Two injections were administered at 2 mm rostral and caudal to the epicenter using a Hamilton syringe configured with a glass micropipette (tip diameter <70 µm). The injection lasted for 3 min, being controlled by a nanoliter syringe pump (KD Scientific Inc.). The pipette was kept in situ for 2 more min to prevent leakage from the injection sites. Each injection consisted of 2.5×10⁵ dissociated NSCs suspended in 2.5 µl of PBS (5×10⁵ cells in total). After transplantation, animals were assigned randomized codes to ensure that experimenters were blinded to the graft genotype.

We performed TMT using a 10-channel Flat Treadmill System for mice (model IW-FT, IWOO Scientific). Each channel consists of a 50×450 mm runway. The TMT protocol adopted in our previous study [23] was slightly modified for mice. All animals received pretraining for 7 days before injury; TMT was initiated 3 days after injury for animals that were assigned to the groups with TMT. Most mice were able to exhibit spontaneous stepping by 7 days at a slow treadmill speed (5 m/min). Temporarily suspended after transplantation, TMT was restarted 3 days later and continued until the animals were sacrificed for histological analysis. TMT was performed at a slow speed (5 m/min) until animals regained weight-supported plantar stepping, and thereafter, the speed was gradually increased up to 10 m/min. The training was performed 6 days per week, with 3 sessions per day. Each session consisted of 20 min of training and 5 min of break. Mice assigned to the groups without TMT, they were handled and brought to the training room together with those in the groups with TMT, but excluded from the actual training.

A total of 36 mice were included for the behavioral analysis with 9 animals per group. Basso Mouse Scale for locomotion (BMS), ladder walk test, and digitized gait analysis using the Catwalk system (Noldus Information Technology) were used to assess recovery of locomotor function. We assessed BMS 24 h after injury and once a week thereafter. For the ladder walk test, the animals were pretrained to walk on ladder rungs for 7 days before surgery, and tested 4 and 8 weeks after injury. The number of hind-paw placement errors per run was counted and the average percentage of errors was obtained from four runs for each animal. The Catwalk analysis was performed 8 weeks after injury. Animals were pretrained for walking on the Catwalk runway before surgery and retrained for 5 days before final testing. On the test day, 4 uninterrupted crossings were recorded. Before computerized analysis, individual footprints from 4 different paws were manually confirmed. The following parameters were automatically calculated: stride length, base of support, and rotation angle. The angle of hind-paw rotation was defined as the angle (in degrees) of the hind-paw axis relative to the horizontal plane. As described previously [23], the relative position of the fore-paws and hind-paws was obtained by measuring the distance between the center pads of the fore-paw and hind-paw prints.

After cardiac perfusion with 4% paraformaldehyde, the spinal cord was dissected and post-fixed, followed by cryoprotection in a graded series of sucrose solutions. Longitudinal sections (20 µm-thick) of the spinal cord were cut using a cryostat (CM 1900, Leica) and thaw-mounted onto Super Frost Plus slides (Thermo Fisher Scientific). For immunohistochemistry, sections were incubated overnight at 4℃ with the following primary antibodies: anti-RFP (1:500, mouse polyclonal; #R10367, Invitrogen), anti-adenomatous polyposis coli (APC) clone CC1 (APC-CC1) (1:200, mouse monoclonal; #OP80, Calbiochem), anti-glial fibrillary acidic protein (GFAP) (1:500, rabbit polyclonal; #Z0334, DAKO) and anti-doublecortin (DCX) (1:400, rabbit polyclonal; #4604S, Cell Signaling Technology). After washing, slides were incubated with appropriate secondary antibodies conjugated to the Alexa Fluor fluorescent dyes. Slides were examined using a confocal laser scanning microscope (LSM 800, Carl Zeiss).

Stereological cell counting was performed as previously described [23]. Longitudinal sections in a horizontal plane were systematically selected at an intersection interval of 300 µm (section sampling fraction=1/15). The sampling grid dimension was 500×500 µm² with a 200×200 µm² counting grid (area sampling fraction=0.16). Using a dissector height of 10 µm in sections with an average post-processing thickness of 18 µm, the height sampling fraction was 0.56 (10/18). The migratory distance of RFP-positive grafts for each animal was determined as the longest longitudinal distance in either rostral or caudal direction from the epicenter out of 5 consecutive longitudinal sections with a 300-µm intersection distance. It is conceivable that the longitudinal migratory distance can be influenced by the extent of graft survival; the more NSCs survive, the longer they distribute longitudinally. To control for the different extents of graft survival, a migration index was generated by dividing the longest longitudinal migratory distance by the longest transverse diameter of the RFP-positive area.

The IGF-1-induced NSCs migration was assessed using Boyden chamber assay kit (CytoSelect™ 24-well cell migration assay kit, #CBA-107; CELL BIOLABS Inc.), where an upper chamber was separated from a well by a polycarbonate membrane insert (12-µm pore size). Dissociated NSCs were plated on a culture insert nested inside of a 24-well culture plate at a density of 5×10⁴ cells/well. StemPro® NSC serum-free medium containing bFGF and EGF was added to an upper chamber within an insert, while IGF-1 was added as a chemoattractant to the same medium in a 24-well with the same insert. The Boyden chamber was incubated at 37℃ with 5% CO₂ for 48 hours. After non-migratory cells on the upper surface of the insert were removed by a cotton swab provided in the kit, migratory cells attached to the bottom side of the insert were stained using the cell-staining solution provided in the kit for 10 min at room temperature. The rate of NSCs migration was calculated by counting cells in 3 random fields of each well using a 10× objective lens (BX51, Olympus). For in vitro scratch assay, dissociated NSCs were plated on a 24-well culture plate coated with Poly-D-lysine at a density of 5×10⁴ cells. StemPro® NSC serum-free medium containing bFGF and EGF was added and maintained throughout the imaging session without changing. An artificial scratch was created with a white 10-µl pipette tip at the center of the mono-layered cultured cells. NSCs migrating toward the scratch were monitored real-time for up to 4 consecutive days using JuLI™Stage (NanoEntek). Areas occupied by migrating cells within the scratch were automatically calculated by JuLI™STAT software.

Neurospheres derived from Igf1r (+/+) or (+/−) NSCs were seeded on a 24-well culture plate with 1 ml of StemPro® NSC serum-free medium containing bFGF and EGF. Approximately 50 to 100 neurospheres were present in each well. Culture medium was not changed during the entire imaging period to prevent perturbation of neurospheres floating in the medium. Time-lapse images were obtained using JuLI™Stage for 4 days to visualize spontaneous neurosphere movement and changes in cytoplasmic protrusions extending from neurospheres. To trace spontaneous motile paths of neurospheres, the center of a neurosphere was manually marked in every image acquired at 30-min intervals using the Manual Tracking plugin in Image J, and this software calculated the summed moving distances of the center of each neurosphere. To measure the extent of changes in filopodia-like cytoplasmic protrusions, the tip of the longest protrusion from each neurosphere traced for the moving paths was also manually marked, and the Manual Tracking plugin calculated the summed length of the changes in tip position. Three randomly selected fields of view were determined in each live imaging experiment. We traced 4 to 6 neurospheres in each imaging session that were clearly identified and delineated for at least 3-hour duration. Quantified data were collected from three independent live imaging experiments for each experimental condition. To examine the influence of IGF-1 and related signaling pathways on the neurosphere motility, IGF-1 (50 ng/ml, #791-MG, R&D Systems), phosphoinositide-3 kinase (PI3K) inhibitor (LY294002; 10 µM, #440202, CalbioChem), mitogen-activated protein kinase/extracellular signal-regulated kinase (MEK/ERK) inhibitor (U0126; 10 µM, #662005, Calbio-Chem), Rho-associated protein kinase (ROCK) inhibitor (Y27632; 10 µg/ml, #129830-38-2, Tocris), IGF-1R inhibitor (PPP, picropodophyllin; 100 µM, #SC-204008, Santa Cruz Biotechnology) were added in culture medium.

Error bars in all graphs represent mean±standard error of mean (SEM). Statistical analysis was performed using GraphPad Prism (version 5.0) and SPSS software (version 23). Unpaired Student's t test or one-way ANOVA was used for comparison of group means. Repeated-measures two-way ANOVA was used to compare mean differences matched at different time points. Two-way ANOVA followed by post hoc Bonferroni analysis was used to test statistical significance of genotype and TMT factors on behavioral outcomes and survival and migration of transplanted NSCs in the injured spinal cord.

To examine influence of IGF-1/IGF-1R signaling on transplanted NSCs in the lesioned spinal cord, we attempted to generate NSCs with deleted Igf1r gene. Since Igf1r-homozygous knockout mice die at birth [24,29], we initially crossed heterozygous male and female mice to produce embryos for primary NSC cultures. But the number of litter was very small, and the frequency of Igf1r-homozygous genotype was much lower in 14-day-old embryos than that expected by the Mendelian principle. Therefore, heterozygous males were crossed with wild-type females to obtain Igf1r-heterozygous (+/−) NSCs. Neurospheres generated from +/− NSCs exhibited similar size (Fig. 1A) and number (Fig. 1B) compared to those from wild-type (+/+) NSCs, indicating that only one allele of Igf1r gene is sufficient to maintain the stemness of NSCs.

We have previously observed that transplanted NSCs at the epicenter region in the spinal cord are under RNS- and ROS-induced cellular stresses [23]. Treatment of either +/+ or +/− NSCs with SIN-1 or H₂O₂, generating RNS or ROS, respectively, significantly reduced the number of viable NSCs (Fig. 1C, D). Addition of IGF-1 to the culture medium significantly prevented RNS- or ROS-induced NSC death, while IGF-1 treatment did not increase the number of NSCs without cellular stresses. The IGF-1-induced robust pro-survival effects were almost completely abrogated in +/− NSCs (Fig. 1C, D). These data suggest that deleting only one allele of Igf1r gene profoundly attenuates the responsiveness of NSCs to IGF-1 under cellular stresses.

We assessed whether IGF-1R signaling in transplanted NSCs could affect TMT-supported locomotor recovery of mice following contusion injury. Half the animals received +/+ NSC grafts and the other received +/− grafts, and half the animals in each graft group underwent TMT up to 8 weeks after transplantation. The mice receiving +/+ NSC grafts began to exhibit improved locomotor quality, as measured using BMS, compared with those receiving +/− grafts at 3 weeks after injury and thereafter (Fig. 2A). TMT effect on BMS was noted at 4 weeks and thereafter in mice with +/+ NSC grafts, showing the best recovery of BMS in mice that received +/+ grafts and underwent TMT. Mice that received +/− grafts also exhibited gradual improvement in locomotor function, but the extent of recovery was inferior to that of the +/+ graft group. Although TMT effect was also observed in mice with +/− NSC grafts, TMT-induced recovery was delayed and less pronounced in this group than that in the +/+ graft group. Repeated-measures two-way ANOVA revealed a significant interaction between treatment conditions and the time points after SCI (p<0.001). Post hoc Bonferroni analysis showed statistically significant difference between +/+ NSC grafts with TMT and +/− without TMT groups (Fig. 2A). Furthermore, the difference in BMS score improvement between mice with +/+ and +/− NSC grafts was statistically significant in animals subjected to TMT (Fig. 2A). In contrast, BMS scores in different genotypes were not significantly different in animals without TMT, suggesting that genotype-dependent improvement of BMS score was more evident in animals subjected to TMT. However, the differences in BMS score between the same genotype groups, whether it is +/+ or +/− group, with and without TMT were not statistically significant by post hoc Bonferroni test.

In the ladder walk test, mice that received +/+ NSC grafts tended to exhibit hind-paw placement errors less frequently than those with +/− grafts, regardless of TMT, at both 4- and 8-week time points (Fig. 2B, C). TMT reduced the number of errors in both genotype groups, but more evidently in +/+ graft group, resulting in the smallest number of errors in mice with +/+ grafts as well as TMT. At both time points, the effects of both genotype and TMT factors were statistically significant by two-way ANOVA (4-week genotype, F_{(1, 32)}=7.816, p=0.009; 4-week TMT, F_{(1, 32)}=9.811, p=0.004; 8-week genotype, F_{(1, 32)}=10.294, p=0.003; 8-week TMT, F_{(1, 32)}=14.586, p=0.001). Post hoc Bonferroni test did not reveal statistically significant TMT effect for either genotype at 4 weeks (Fig. 2B). At 8 weeks, however, TMT effect was significant only in mice with +/+ NSC grafts (p=0.045) (Fig. 2C), indicating that TMT significantly supported accuracy of hind-paw placement in the ladder walk test when Igf1r (+/+) NSCs were transplanted. Among the four parameters in the Catwalk analysis, neither stride length nor base of support was affected by genotype or TMT factors (Fig. 2D, E). Relative position, which is indicative of coordination between fore- and hind-paws, was markedly reduced by TMT in both genotype groups (Fig. 2F). Mice with +/+ NSC grafts and TMT together showed the lowest value, indicating the greatest recovery of fore- and hind-paw coordination in this group. Similar findings were noted upon measurement of the paw angle (Fig. 2G). TMT factor was statistically significant for both measures, as revealed by two-way ANOVA (relative position, F_{(1, 32)}=8.109, p=0.008; paw angle, F_{(1, 32)}=11.580, p=0.002). However, post hoc Bonferroni test did not reveal statistically significant TMT effect for either genotype in the measurements of relative position and paw angle (Fig. 2F, G). Although a trend of worse performance (higher values) in both measures was noted in mice with +/− NSC grafts, the effect of genotype was not statistically significant. Taken together, these behavioral data indicated that transplanting Igf1r (+/+) NSCs combined with TMT could achieve the highest locomotor recovery, and that TMT-supported functional improvement was significant in mice with +/+ NSC grafts in the ladder walk test.

We compared the survival of NSC grafts heterozygous for Igf1r to that of wild-type NSC grafts following transplantation into the injured spinal cord, and examined whether TMT affected the survival of either genotype NSCs. Surviving NSCs were identified by RFP expression, and then the number of RFP-positive cells was stereologically counted. Four weeks after transplantation, a considerable number of RFP-positive grafts were visualized not only around the epicenter but also in distant regions caudal or rostral to the epicenter in mice with +/+ NSC grafts (Fig. 3A). However, the extent of NSC survival was only approximately 10~15% of the total transplanted cells (5×10⁵) (Fig. 3D), indicating substantial demise of transplanted NSCs during the post-transplantation period. The survival of +/− NSC grafts was more profoundly decreased, with most of the surviving NSCs being present only at the epicenter areas (Fig. 3A). Survival of transplanted NSCs only slightly diminished by the 8-week time point, indicating that the loss of NSCs predominantly occurred early after transplantation. Being subjected to TMT or not did not lead to appreciable difference in the survival of either +/+ (70920±6861 vs. 76950±13490 cells, without or with TMT, respectively) or +/− (28170±8136 vs. 26790±5971 cells, without or with TMT, respectively) NSCs at 4 weeks (Fig. 3A, D). At 8 weeks, a similar pattern of NSC survival was noted, with a genotype-dependent decrease in the number of surviving NSCs (Fig. 3B, C). There was no noticeable beneficial effect of TMT on the survival of +/− NSC grafts (21760±3265 vs. 25562±7437 cells, without or with TMT, respectively), while it tended to increase the survival of +/+ NSCs (54620±11289 vs. 75047±9676 cells, without or with TMT, respectively) especially in distant regions caudal or rostral to the epicenter. Quantitative analysis revealed that only the genotype effect was statistically significant at both time points by two-way ANOVA (4-week, F_{(1, 16)}=26.087, p<0.001; 8-week, F_{(1, 32)}=23.623, p<0.001). Although TMT increased the survival of +/+ grafts by 37.4% at 8 weeks (Fig. 3D), TMT effect was not statistically significant (F_{(1, 32)}=2.045, p=0.162). However, post hoc Bonferroni test revealed that the genotype effect was statistically significant only between the groups subjected to TMT (Fig. 3D), suggesting that the genotype effect was more obvious when animals underwent TMT. Transplanted NSCs, either heterozygous or wild-type for Igf1r gene, differentiated into three neural cell types (Fig. 4A~E). There was no discernable influence of TMT on differentiation of NSCs of either genotype.

We found that different experimental conditions resulted in marked differences in the extent of NSC migration (Fig. 3). When we focused on the distance of the migrating NSCs from the epicenter, +/+ NSC grafts showed a tendency to migrate further than +/− grafts, either rostrally or caudally, at 8 weeks after transplantation (Fig. 5A~D). At this time point, mice that were subjected to TMT exhibited longer migration of +/+ NSCs (Fig. 5A, B). NSCs positioned at the leading edge of the grafts frequently showed elongation of cytoplasmic processes in a rostro-caudal direction. However, this influence of TMT was not obvious in mice that received +/− NSC grafts (Fig. 5C, D). Furthermore, elongated morphology was rarely observed in +/− NSCs. We obtained the migration index (the longest longitudinal migratory distance divided by the longest transverse diameter) for each animal from the longitudinal spinal cord section showing the longest migration. The effect of genotype on the migration index was already significant at 4 weeks by two-way ANOVA (F_{(1, 16)}=14.975, p=0.001). Post hoc Bonferroni test revealed that the genotype factor was significant in animals subjected to TMT, but not in those without TMT (Fig. 5E). However, TMT factor was not statistically significant at this time point by two-way ANOVA (F_{(1, 16)}=1.281, p=0.274). At 8 weeks, the migration index was markedly increased compared with 4 weeks only in animals with +/+ NSC grafts that were subjected to TMT. The influence of the genotype factor on the degree of migration was highly significant, as revealed by two-way ANOVA (F_{(1, 32)}=55.842, p<0.001). The TMT factor was also significant at this time point (F_{(1, 32)}=15.471, p<0.001). Post hoc Bonferroni test revealed that the difference in the extent of migration between genotypes was statistically significant in animals with or without TMT (Fig. 5E). Importantly, the migration difference caused by TMT was statistically significant only in animals with +/+ NSC grafts (Fig. 5E), indicating that TMT positively influenced the migration of NSCs that were +/+ for Igf1r. These data indicated that migration of NSC grafts is highly dependent on Igf1r genotype at both 4 and 8 weeks after transplantation, and that TMT significantly enhanced the migration only when NSCs were homozygous for Igf1r at the 8-week time point.

Given the crucial role of the Igf1r genotype of NSCs and TMT in determining the extent of NSC migration in the lesioned spinal cord, we directly compared in vitro migratory ability between +/+ and +/− NSCs. In the Boyden chamber assay, +/+ NSCs occasionally migrated to the bottom side of the membrane without a concentration gradient of any chemoattractant (Fig. 6A). When IGF-1 was added to the lower chamber, the number of migrating cells markedly increased. The concentration gradient of IGF-1 in the lower chamber was still effective in promoting migration of +/− NSCs, but the basal migration ability was low in NSCs with only one copy of Igf1r gene. Quantification showed that addition of IGF-1 significantly increased the number of migrating NSCs with both genotypes. However, the basal migratory ability of +/− NSCs was significantly lower than that of +/+ NSCs, and the number of cells migrating to the bottom side in response to IGF-1 was also significantly attenuated in +/− NSCs (Fig. 6C). After a scratch injury at the center of NSC cultures on a 24-well plate, +/+ NSCs started to migrate to the scratch site as early as 6 h, and more than 80% of the scratched area became occupied by migrating NSCs by 24 h (Fig. 6B, D). In contrast, +/− NSCs showed appreciable migratory behavior only approximately 24 h after the scratch, and only half the scratch area was occupied at 96 h, the last time point examined (Fig. 6B, D). These results collectively suggest that IGF-1 can act as a chemoattractant to NSCs, and IGF-1R signaling in NSCs is essential for their migratory behavior.

Previous studies have reported highly migratory behavior of NSCs in vivo [30,31], but it remains unclear whether NSCs are inherently motile, and if so, how their motility is regulated. Since basal migratory capacity of NSCs was observed in vitro (Fig. 6), we continuously monitored the behavior of cultured neurospheres using a live cell imaging apparatus. Surprisingly, wild-type neurospheres were highly motile (Supplementary video 1). The majority of neurospheres frequently changed their positions by a distance of up to several hundreds of micrometers (Fig. 7A), although not all neurospheres were motile. Individual NSCs occasionally migrated away from the original neurosphere; on the contrary, separate neurospheres occasionally merged together to form larger neurospheres (Supplementary video 1). The highly motile activity of neurospheres was accompanied by dynamic formation and/or retraction of cytoplasmic protrusions resembling filopodia or lamellipodia (Fig. 7B). The motile paths and the changes in cytoplasmic protrusions were quantified by tracking the marks at the center of neurospheres and tip of cytoplasmic protrusions, respectively (Fig. 7A~C). The spontaneous motility of neurospheres with dynamic cytoplasmic protrusions were quite stable during the 4-day culture duration (Fig. 7D, E), indicating that the neurosphere motility was not a transient phenomenon due to a stabilization process following the initial plating of cells. To examine potential signaling pathways involved in the spontaneous motility of neurospheres, cultured neurospheres were treated with various pharmacological inhibitors (Fig. 8). The extent of both neurosphere movement and cytoplasmic protrusions were potently attenuated only by LY294002, a PI3K inhibitor.

Next, we tested whether IGF-1 signaling could regulate neurosphere motility. When IGF-1 was added to the culture medium, not only was the movement of individual neurospheres enhanced but also the proportion of motile neurospheres was increased (Fig. 9A, B) (Supplementary video 2). At the same time, cytoplasmic protrusions were much more dynamic in the IGF-1-treated condition. On the contrary, pharmacological inhibition of IGF-1R using PPP attenuated the extent of neurosphere movement and cytoplasmic protrusions (Fig. 9C) (Supplementary video 3). Finally, neurospheres obtained from NSCs (+/−) for Igf1r gene exhibited very little motility (Fig. 9D) (Supplementary video 4). In particular, cytoplasmic protrusions were rarely observed during the entire culture duration. Quantification results showed that IGF-1 and IGF-1R inhibitor significantly titrated the extent of +/+ neurosphere movement (Fig. 9E). Neurospheres heterozygous for Igf1r exhibited severely suppressed movement. Furthermore, changes in cytoplasmic protrusions were reduced by more than 3-fold in +/− neurospheres (Fig. 9F). These results collectively indicate that IGF-1/IGF-1R signaling in NSCs critically regulates inherent motility that is highly likely to be the cellular basis of their migratory behavior in the injured spinal cord.