Long-term stability and computational analysis of migration patterns of L-MYC immortalized neural stem cells in the brain

Background Preclinical studies indicate that neural stem cells (NSCs) can limit or reverse central nervous system (CNS) damage through delivery of therapeutic agents for cell regeneration. Clinical translation of cell-based therapies raises concerns about long-term stability, differentiation and fate, and absence of tumorigenicity of these cells, as well as manufacturing time required to produce therapeutic cells in quantities sufficient for clinical use. Allogeneic NSC lines are in growing demand due to challenges inherent in using autologous stem cells, including production costs that limit availability to patients. Methods/Principal findings We demonstrate the long-term stability of L-MYC immortalized human NSCs (LM-NSC008) cells in vivo, including engraftment, migration, and absence of tumorigenicity in mouse brains for up to nine months. We also examined the distributions of engrafted LM-NSC008 cells within brain, and present computational techniques to analyze NSC migration characteristics in relation to intrinsic brain structures. Conclusions/Significance This computational analysis of NSC distributions following implantation provides proof-of-concept for the development of computational models that can be used clinically to predict NSC migration paths in patients. Previously, models of preferential migration of malignant tumor cells along white matter tracts have been used to predict their final distributions. We suggest that quantitative measures of tissue orientation and white matter tracts determined from MR images can be used in a diffusion tensor imaging tractography-like approach to describe the most likely migration routes and final distributions of NSCs administered in a clinical setting. Such a model could be very useful in choosing the optimal anatomical locations for NSC administration to patients to achieve maximum therapeutic effects.


Conclusions/Significance
This computational analysis of NSC distributions following implantation provides proof-ofconcept for the development of computational models that can be used clinically to predict NSC migration paths in patients. Previously, models of preferential migration of malignant tumor cells along white matter tracts have been used to predict their final distributions. We suggest that quantitative measures of tissue orientation and white matter tracts determined from MR images can be used in a diffusion tensor imaging tractography-like approach to describe the most likely migration routes and final distributions of NSCs administered in a clinical setting. Such a model could be very useful in choosing the optimal anatomical locations for NSC administration to patients to achieve maximum therapeutic effects. PLOS  Introduction Stem cell-based therapies for tumors and neurodegenerative diseases depend on efficient delivery of stem cells to the areas of damage. Neural stem cells (NSCs) are inherently pathotropic and suggested to migrate to sites of tumor and damage in the central nervous system (CNS) [1]. Therefore, NSCs can be exploited for cell replacement, regeneration, and therapeutic delivery strategies. NSCs are currently being evaluated in clinical trials for stroke, multiple sclerosis, amyotrophic lateral sclerosis, Parkinson's disease, and other CNS diseases [2][3][4][5][6]. In addition, we have previously shown that a genetically modified NSC line (HB1.F3.CD) can deliver anticancer agents selectively to invasive brain tumor sites [7][8][9][10]. Allogeneic NSC lines are in growing demand due to the limitations of adult autologous patient-derived NSCs. Although self-renewing NSCs are present in developing brain tissue, in vitro passages of these cells lead to decreased capacity for cellular self-renewal, decreased differentiation potential, and increased accumulation of chromosomal and functional instabilities [11][12][13][14]. Recently, we developed a new human NSC line, LM-NSC008, that was genetically engineered to stably express the L-MYC gene. This line has favorable physiological, multi-lineage differentiation and migratory properties [14]. The LM-NSC008 cell line overcomes several major hurdles to clinical translation of cell-based therapies by demonstrating long-term stability, lack of tumorigenicity, and ease of production in sufficient quantities for clinical trials. Further supporting this line's potential for use in clinical applications, LM-NSC008 cells demonstrate tropism to sites of brain tumors and/or injury when administered intranasally or intracranially to the mouse brain [14]. Ultimately, characterization of the long-term stability and migratory properties of allogeneic LM-NSC008 cells will have a relevance for use of this cell line for proof-ofconcept studies and may pave the way for successful translation of these cells to target neurodegenerative diseases. Clinical translation of NSC-based therapy has been hampered by inability to quantify or predict NSC migration to sites of tumor or injury. A method for predicting NSC migration is needed because the migration route as well as the location of tumor/injury may affect the final number of viable cells that reach the tumor/injury site. We expect the characterization of these migration paths and quantification of the NSC viability at the tumor/injury site will ultimately allow for disease or injury site-specific modification of NSC dose and route of administration. We have recently published manuscripts describing use of 3-dimensional reconstructions to determine the distribution and coverage of brain tumors by therapeutic NSCs in orthotopic xenograft models of glioma [15,16]. We have now used computational analyses of tissue structure, orientation, and anisotropy to characterize preferential routes of NSC migration, which may provide the ability to predict routes of migration and spatial distribution of LM-NSC008 cells within the brain. Successful development of a computational predictive model of LM-NSC008 migration in the brain would provide a quantitative method for analysis of repeat treatments for LM-NSC008 cells or other NSC lines, which could change the NSC treatment administration paradigm for patients with brain tumors or injuries.
IncuCyte live cell analysis imaging system demonstrated similar growth kinetics of LM-NSC008 cells at p5 and p45 passages. LM-NSC008 cells were plated on 24-well plates at a density of 2 x 10 4 cells/cm 2 (40,000 cells/well). Cells were grown for 10 days at 37 0 C, media (RHB-A) change was performed every 3 days and imaging of eGFP-labeled LM-NSC008 cells was acquired every 12 h using the IncuCyte S3 Live Cell Analysis System (Sartorius). Experimental data is represented as the mean ± SD of 2 independent assays performed in quadruplicate.

Genomic DNA and qRT-PCR analysis
LM-NSC008 cells were cultured and passaged in complete growth medium and harvested every 5th passage. Genomic DNA was isolated using the DNeasy kit (Qiagen) and PCR was performed to amplify the L-MYC gene. The PCR product was analyzed by electrophoresis on a 0.8% agarose gel and visualized by ethidium bromide staining. The L-myc-pMXs plasmid was used as a positive control and genomic DNA from untransduced NSC008 cells and no DNA template were used as the two negative controls. Total RNA was isolated using TRIzol LS reagent (Invitrogen) according to the manufacturer's protocol. RNA (100 ng) was reverse transcribed into cDNA using the iScript cDNA Synthesis Kit (Bio-Rad). Quantitative real-time PCR (qRT-PCR) was performed in 25 μl reactions using the iQ SYBR Green Supermix Kit (Bio-Rad) and L-MYC-specific primers (L-MYC: 5'AGAGGCAGTCTCTGGGTATT3'; 5'TG TGCTGATGGATGGAGATG3') on a Bio-Rad CFX96. Values obtained from qRT-PCR were normalized to GAPDH, and the relative expression of the L-MYC gene was calculated using the 2 -ΔΔCt method. The intensities of the L-MYC PCR bands were analyzed using ImageJ software, quantified, and represented as percentage of the total size of measured area.

Detection of expressed and secreted proteins
Protein array analysis was conducted using Cytokine Antibody Array V (RayBiotech, http:// www.raybiotech.com). LM-NSC008 cells (p5 and p45) were cultured under conditions described previously [14] for 7 days. Culture media (CM) was collected and cells were pelleted by centrifugation at 16000 g and lysed with lysis buffer (20mM TRIS, 150 mM NaCl, 2 mM EDTA, 0.5% Triton X-100). Antibody array membranes were exposed to 1 mL of CM and 1 mL of cell lysates, containing 500 μg of total protein in blocking buffer. Membranes were incubated at 4˚C overnight and chemiluminescence films were developed as per the manufacturer's recommendations.

In vivo animal studies
All animal studies were performed under an approved City of Hope Institutional Animal Care and Use Committee protocol (IACUC #12025). Male and female NOD scid IL2Rgamma null (NSG) mice (8-12 weeks old) were intracranially injected with 5 x 10 5 NSCs/4 μl of PBS per mouse (n = 10) into the right frontal lobe. Six (Group 1, n = 5 mice) and nine months (group 2, n = 5 mice) after NSC injection, mice were euthanized and their brains isolated and post fixed in 4% paraformaldehyde (PFA). Mice developed no symptoms that required euthanasia. However, mice were monitored daily for the standard AAALAC criteria for euthanasia (e.g., 20% weight loss, domed head, ataxia) as well as any distress or discomfort in accordance with the recommendations of the Panel of Euthanasia of the American Veterinary Medical Association. Mice were euthanized by CO 2 inhalation, with a gradual increase in the flow of CO 2 , allowing for observation of mice during euthanasia and minimizing distress.
After fixation, brains were rinsed with 1x PBS and transferred to 70% EtOH for dehydration (for 3-5 days) and later embedded into paraffin blocks (four mice per Groups 1 and 2).
The paraffin blocks were sectioned into 10 μm sections and stained (every 10th section) with anti-human Nestin antibody (Millipore; Cat #: MAB5326) and DiI lipophilic tracer (Invitrogen; D282). Two brains (one from each Groups 1 and 2) were fresh frozen and immersed into Optimal Cutting Temperature Compound (OCT) over dry ice to be cryosectioned into 10 μm sections for staining with Stem123 antibody specific for human GFAP (Cellartis; Cat #: Y40420).

IHC and DiI staining
Paraffin-embedded brain sections were deparaffinized in xylene and rehydrated with ethanol [17]. The brain sections were then processed for antigen retrieval with Proteinase K (Dako ready-to-use Cat #: S3020). The tissue sections were incubated in peroxidase quenching solution (0.3% H 2 O 2 made in 100% methanol) for 20 min at room temperature and then in blocking solution for 1h at room temperature (50% BlockAid, Invitrogen B10710; 50% Western Blocking Reagent, Roche Applied Sciences 11921673001; 1% Triton-100x). Sections were then stained with primary antibody in blocking solution and incubated overnight at 4˚C followed by four washes in PBS and reaction with biotinylated secondary antibody for 1 h as described previously (1:250 dilution, Vector BA-2001) [17]. Sections were washed in PBS, incubated in avidin-biotin complex (ABC) solution for 1 h at room temperature and 5 min in 3, 3 0 -Diaminobenzidine (DAB) substrate solution containing 0.25% H 2 O 2 . Then the brain sections were washed in PBS and mounted with Cytoseal 8 mounting media for bright field imaging (Richard-Allan Scientific). Staining with the Stem123 antibody followed the same protocol but on frozen sections fixed with 4% PFA. Stem123 primary antibody (Celartis, Cat #: Y40420) was used at a 1:1000 dilution and a secondary biotinylated antibody at 1:250 (Biotinylated Anti-Mouse IgG (H+L); Vector, Cat #: BA-2001). For DiI staining, the OCT-immersed brain sections were fixed with 4% PFA, dehydrated through a graded ethanol series, and stained with DiI lipophilic tracer for 15 min (DiIC18 (3)

Computational analysis of tissue anisotropy
Tissue anisotropy and orientation were quantified in DiI-stained ex vivo tissue using the OrientationJ plugin for FIJI [18] following methods outlined by Budde [19][20][21]. OrientationJ uses Fourier transform analysis to compute eigenvalues and eigenvectors of a structure tensor at each pixel in the DiI image. The eigenvalues (λ) are combined to produce measures of anisotropy (coherence), directionality (orientation), and other quantities that were not used in this analysis. Coherence (C = (λ maxλ min )/(λ max + λ min )) measures the degree to which there is a dominant eigenvalue of the structure tensor (see S2 Fig for illustration and discussion of these concepts). Coherence takes values from 0-1, with a value of 0 corresponding to isotropic and a value of 1 to anisotropic. Orientation is the angle of the dominant eigenvector measured from the positive abscissa axis. To generate the orientation and coherence maps, a cubic spline gradient method with a Gaussian window of 1 pixel standard deviation was used in the Orien-tationJ Analysis module of the OrientationJ plugin. The orientation of the white matter (WM) was compared to the orientation of NSCs with respect to each other. To calculate the orientation of NSCs the following procedure was used. The center of each NSC cluster defined by the nestin-or Stem123-stained image was dilated by 200 pixels and eroded by 100 pixels to create coalesced regions of NSC clusters yielding an NSC density map. A second degree polynomial was fit to each of these coalesced regions and a line tangent to this polynomial and closest to each NSC cluster center was calculated. The angle of this line with the abscissa indicates the orientation of NSCs (θ NSC ) for that cluster.

Characterization of LM-NSC008 cells in vitro and in vivo
Characterization of LM-NSC008 cells was conducted in naïve non-tumor bearing mouse brain 6 or 9 months post NSC administration (Fig 1). To observe the behavior and migration of intracranially-injected LM-NSC008 NSCs, we intracranially injected LM-NSC008 cells, lentivirally transduced to express enhanced green fluorescent protein (eGFP) and firefly luciferase genes (Ffluc) (5 x 10 5 cells/4 μl PBS) into the right frontal hemisphere of adult male and female NSG mice (n = 10). Six or nine months after NSC injection, mice were euthanized and their brains were harvested, fixed, sectioned, and stained for histological examination and immunohistochemistry (IHC) analysis. Human nuclear antigen (HNA) and eGFP protein expressions were used to visualize LM-NSC008 cells that migrated from the injection site in the mouse Long-term behavior and migration of LM-NSC008 NSCs PLOS ONE | https://doi.org/10.1371/journal.pone.0199967 August 2, 2018 brain ( Fig 1A). Analysis of co-expression of eGFP and HNA confirmed the human origin of LM-NSC008 cells in the mouse brain (Fig 1, insets 1a-1d). We also demonstrated that LM-NSC008 cells expressed other neuroectodermal-specific markers such as nestin and Pax6 at 6 and 9 months post administration (Fig 1B), implicating in the radial growth of axons (huNestin, green) (Fig 1B and 1C). Pax6 is uniformly expressed in the early neuroectoderm of human fetuses and in neural progenitors differentiated from hESCs and induced pluripotent stem cells (iPSCs). The LM-NSC008.eGFP cells (white arrows) were double positive for Pax6 (red)/eGFP (green), indicating multipotent differentiation of LM-NSC008 cells towards the neuronal lineage in non-tumor bearing naïve mouse brain at 9 months post administration ( Fig 1C). Next, we performed IHC for the human-specific glial marker (Stem123), visualizing LM-NSC008 cells in the corpus callosum (Fig 1D). Ki-67 staining of mouse brain sections obtained at 9 months post administration indicated no expression of Ki-67 protein, suggesting that LM-NSC008 cells most likely were not dividing at the 9 month timepoint.
For in vitro studies, LM-NSC008 cells were grown and passaged up to 50 times under hypoxic conditions (4% O 2 ) in a humidified incubator, as described previously [14]. In growth factor-supplemented stem cell medium, LM-NSC008 cells grew as a monolayer and showed no changes in growth rate up to 50 passages in vitro (S1A-S1C Fig). The continued presence of L-MYC in LM-NSC008 cells was confirmed by PCR analysis of genomic DNA at every fifth passage (p5-p50) (Fig 1F). L-MYC band intensities for every fifth passage of LM-NSC008 cells (p5 to p50) were compared using one-way ANOVA and showed no statistical difference among the bands when analyzed using the same sized area (P = 0.058). We also analyzed L-MYC gene expression at every fifth passage using real time PCR (RT-PCR) (data not shown). Our results indicate L-MYC gene expression is relatively stable in LM-NSC008 cells over the long-term in vitro passaging.
To assess protein secretion and expression by LM-NSC008 cells over passages in vitro, we performed protein array analyses using conditioned medium (CM, p5 and p45) and cell lysates (derived from pelleted cells, p5 and p45). LM-NSC008 cells from low (p5) and high (p45) passages were cultured for 5 days, after which CM was collected and cells were collected and pelleted. CM derived from low (p5) and high (p45) passages showed comparable cytokine secretion profiles (MCP-1, EGF, TIMP-1, TGF-beta 2, OPN, and IGFBP-2) (Fig 1G2 and  1G4). Similar results were observed in lysates derived from LM-NSC008 cells from low and high passages, suggesting the stability of protein expression in vitro (Fig 1G1 and 1G3).
We also demonstrated self-renewal and growth kinetics of LM-NSC008 cells at p5 and P45 in vitro (S1D Fig) using IncuCyte. Scale-up expansion of LM-NSC008 cells to a large cell banks (expansion of LM-NSC008 cells from 5 x 10 7 cells to 3 x 10 9 cells within 10 days) was demonstrated using a Quantum Cell Expansion System (S1E-S1G Fig) [22]. Collectively, we demonstrated in vivo migration and fate of LM-NSC008 in mouse brain and stability of these cells during in vitro passaging. These data support future development of cell-based therapies using LM-NSC008 cells for various neurodegenerative disorders.

Migration of LM-NSC008 cells at 6 and 9 months post-injection in naïve mouse brain
To observe changes in the distributions of LM-NSC008 cells in the brain over the months following implantation, we used an orthotopic human NSC xenograft model in which we injected LM-NSC008 cells (5 x 10 5 /4 μl PBS) into the right frontal brain hemispheres of adult male and female NOD scid IL2R gamma null (NSG) mice (n = 10). Mice were euthanized 6 or 9 months after LM-NSC008 injection, and their brains were harvested, fixed, sectioned, and stained for histological examination.
LM-NSC008 cells were present at the NSC injection site and migrating through the corpus callosum (Fig 2). Accumulation of LM-NSC008 cells was evident in the central corpus callosum by 6 months post-implantation (Fig 2A and 2B), and by 9 months post-implantation an increased number and higher density of NSCs was evident within white matter (WM) distant from the injection site (Fig 2E and 2F), including the anterior commissure (AC). Within the AC, NSCs aggregated along the white matter/grey matter (WM/GM) interfaces (Fig 2). 3D reconstructions of the brain using aligned and Stem123-stained histological sections allowed us to visualize LM-NSC008 cells in 3-dimensions and conduct further analysis using computational methods (Fig 2C, 2D, 2G and 2H). 3D reconstructions and histological sections that were used in computational analyses for all mice are shown in supplemental material (S4-S6 Figs).

Computational analysis of LM-NSC008 distribution
Migration of LM-NSC008 cells away from the injection site was quantified for three time points: 3 (n = 5), 6 (n = 4) and 9 (n = 3) months post-injection ( Fig 3A). To quantify migration of LM-NSC008 cells away from the injection site, we used brain sections stained with hu-nestin and Stem123 antibodies to identify LM-NSC008 cells and lipophilic cyanine dye 1,10-dioctadecyl-3,3,3030-tetramethylindocarbocyanine perchlorate (DiI) to identify myelin associated with white matter tracts. DiI-stained sections were aligned with hu-nestin-or Stem123-stained sections using Reconstruct software (version 1.1.0.1) [23]. Linear distances of LM-NSC008 cells from the injection site (excluding cells within a 1000 pixel [1444 μm] radius from the injection site) as a proportion of the total number of LM-NSC008 cells for each LM-NSC008 cluster was represented as a cumulative probability distribution (CDP). Median distances at 3, 6, and 9 months post-injection were 437 μm, 1030 μm, and 902 μm, respectively. CDPs were also calculated separately for LM-NSC008 cells in white matter (WM; Fig 3B) and grey matter (GM; Fig 3C). We note that NSCs aggregated in the AC ipsilateral to the injection site at the 9 month time point traveled a longer distance than the linear distance measured from the injection site. Roughly 66%, 72%, and 59% of NSCs were found within the WM at 3, 6, and 9 months post-injection, respectively (Fig 3D).
We observed that NSCs aggregated at WM/GM interfaces along the boundary of the AC. To quantify the degree to which NSCs aggregated along the WM/GM interfaces, we calculated the distance between LM-NSC008 cells and the closest WM/GM interface for each brain section and computed the CPDs for the NSCs in both WM ( Fig 3E) and GM (Fig 3F). The distances between of NSCs and the nearest WM/GM interface at the 6 and 9-month time points were smaller than those for the 3-month time point, indicating an increased proportion of NSCs closer to WM/GM interfaces. The median distances between NSCs and the nearest WM/GM interface at the 3, 6, and 9 month time points were 263 μm, 118 μm, and 87 μm, respectively (combined WM and GM data).

Analysis of LM-NSC008 migration and alignment with intrinsic brain structures
The alignment of LM-NSC008 cells with intrinsic brain structures was evaluated by correlating the orientation of the LM-NSC008 cells with the orientation of the brain tissue. First, tissue orientation was quantified on a section of DiI-stained tissue (Fig 4A) using structure tensor analysis with the OrientationJ [24] plugin for FIJI (Fig 4B and 4C). The eigenvector associated with the largest eigenvalue of the structure tensor was used to calculate the dominant direction of the tissue at each pixel by calculating the angle of the principle eigenvector with respect to the abscissa axis (θ WM ) (Fig 4C). NSCs were identified by positive nestin or Stem123 staining ( Fig 4E) on a tissue section adjacent to the DiI-stained section. Clusters of NSCs (Fig 4F) were used to determine the orientation of the LM-NSC008 cells with respect to each other (Fig 4G). The orientation of the LM-NSC008 cells (θ NSC ) was calculated and was found to positively correlate with the orientation of the DiI-defined WM (θ WM ) (Fig 4H; slope = 0.65, r 2 = 0.22). The linear regression was weighted by the total number of NSC clusters in each coalesced region analyzed. The distribution of LM-NSC008 cells in the brain combined with the correlation of the orientation of the cells with the intrinsic structures of the brain suggests that NSCs preferentially migrate along WM and align their major axis with the structure of the brain tissue.  Using a computational model, we performed simulations of NSC migration in the brain to study potential NSC migration paths. We initialized 500 simulations at random points within a circular region of the injection site, located near the corpus callosum (Fig 4D). The orientation vectors at each point on the DiI image were calculated using OrientationJ with 5 pixel Gaussian kernel (to calculate stable θ WM ). These calculated θ WM were used to simulate NSC migration in the WM. In the GM, NSC migration angles were randomly sampled from a uniform distribution of 0-180 degrees relative to the direction of the migration vector of the Long-term behavior and migration of LM-NSC008 NSCs previous time step. The paths were truncated if the NSC path intersected itself. Thus, we formulated a predictive model of NSC migration in the brain, consistent with observed NSC distributions at the 3, 6, and 9 month post-injection time points. A potential limitation of our approach is that we considered only NSC migration within the brain parenchyma, and did not account for leptomeningeal migration through the cerebrospinal fluid or via intravascular dispersion. Another limitation of the analysis is the two-dimensional analysis of an inherently three-dimensional process.

Discussion
The LM-NSC008 cell line demonstrated long-term stability and lack of tumorigenicity in vitro and in vivo. Propagation of LM-NSC008 cells using Quantum Cell Expansion bioreactor (Terumo BCT) demonstrated ease of production in sufficient quantities for clinical trials (S1 Fig). These data support the further development of immortalized LM-NSC008 cells for allogeneic use in the treatment of brain tumors and injury and other neurodegenerative diseases.
In the absence of any pathological injury, LM-NSC008 cells migrated along the corpus callosum (CC) when injected into right frontal lobe. Migrating LM-NSC008 cells localized to the CC (especially the central CC) by 3 months after injection [14]. Analysis of NSC distribution in brains 9 months post-injection showed NSCs aggregated at the interface of the WM/GM in the CC and AC, which was consistent with a nonlinear anisotropic migration predicted by mathematical models of cellular movement with differential rates of migration in the tissue [25].
The brain responds to tumor or injury by activating a variety of defense mechanisms aimed at repairing neuronal damage [26][27][28]. These mechanisms include enhanced synthesis and secretion of neurotrophins and release of pro-and anti-inflammatory molecules that trigger mobilization and migration of endogenous NSCs to the sites of brain injury. Similar cues are involved in attraction of exogenous NSCs to the site of injury [29]. Three independent processes, determine the success of cell-based therapies: 1) survival of the administered cells, 2) trafficking of the cells to the injured/tumor area; and 3) integration of the cells into the host's neuronal circuits [29,30]. Because transplanted NSCs can migrate and specifically home into the sites of injury, they have potential therapeutic value in the treatment of diverse neuronal disorders [31]. We developed a method of quantitative analysis of NSC migration that should be useful in analyzing migration of endogenous, exogenous and tumor/injury specific or endogenously cued migration and interpretation of NSC fate in the brain.
The presented computational analysis of NSC migration and distribution provides proofof-concept for the development of a more mechanistic computational model that can be used to predict NSC migration paths. Although the calculation of tissue orientation (eigenvectors) may appear complicated at first, the task is easily accomplished using the OrientationJ plugin of the open source image analysis software FIJI/ImageJ. Mathematical models of preferential migration along WM tracts have been used to predict distributions of malignant tumor cells, and may also be used to predict NSC migration routes [32][33][34]. We suggest that quantitative measures of tissue orientation and WM tracts determined from magnetic resonance images [35] can be used in a diffusion tensor imaging tractography-like approach to describe the most likely routes of NSC migration. The presence of chemo-attractants or repulsive signals can easily be incorporated into the computational framework to predict migration rates and routes of NSCs to sites of tumor or TBI. Such a model could greatly enhance the translational potential of delivery of NSCs to treat brain tumors and other brain pathologies by allowing us to identify the most efficient targets for NSC therapy based on the most likely route of migration to the target site. In addition, it could ultimately enable choosing the optimal location for NSC administration to a patient to achieve maximum therapeutic effect.

S1 Fig. Morphology of LM-NSC008 cells in culture and in naïve non-tumor bearing mouse brain. (A-C)
Images of LM-NSC008 cells in culture at passages 5, 10, and 45. Scale bar, 100 μM. (D) LM-NSC008 cells were plated in 24-well plates at a density of 2 x 10 4 cells/cm 2 (40,000 cells/per well). Cells were grown for 10 days and imaged every 12 h, using IncuCyte S3 Live Cell Analysis. Media was changed every 3 days. Experimental data is represented as mean ± SD of 2 independent assays performed in quadruplicate.