Journal Logo

Original Articles: Cell Therapy and Islet Transplantation

Human Neural Stem Cell Grafts Ameliorate Motor Neuron Disease in SOD-1 Transgenic Rats

Xu, Leyan1; Yan, Jun1; Chen, David1; Welsh, Annie M.1; Hazel, Thomas2; Johe, Karl3; Hatfield, Glen1; Koliatsos, Vassilis E.1,3,4,5,6

Author Information
doi: 10.1097/


The potential of neural stem cells (NSCs) as therapy for motor neuron disease (MND) will depend on their ability to survive, differentiate, and become integrated when grafted into spinal cords undergoing chronic degenerative changes. However, neurodegeneration represents a particularly challenging biological environment and cell death signals present in established neurodegenerative disease (1–3) may be incompatible with graft survival. In addition, the adult spinal cord is viewed as lacking cells and/or signals required for regeneration (4) and the majority of NSC grafting studies have shown poor or restricted differentiation (5, 6). Recent findings using human NSCs have rekindled some optimism, but these data are limited to intact animals or animals with spinal cord injuries (7). In this report, we use a transgenic (Tg) model of motor neuron disease (SOD1 G93A rats), which has been extensively characterized (2, 8) and represents a comprehensive model for the neuropathology and clinical symptoms of an especially aggressive form of amyotrophic lateral sclerosis (ALS). Our findings show that partial grafts of human NSCs into the lumbar segments of spinal cord survive, undergo extensive neuronal differentiation, and promote motor neuron survival and function in these animals. Furthermore, our data indicate that exogenous cells with therapeutic properties can survive in a degenerative environment in the spinal cord and that rodent models are quite suitable for the preclinical evaluation of human NSCs as therapeutic tools for MND.


SOD1 G93A Breeding

SOD1 G93A male rats supplied by Dr. David S. Howland, University of Washington, Seattle, were bred to four Tac: N(SD) female rats from Taconic (Germantown, NY). Rats were bred for one week in pairs of one male: one female. Offspring were weaned and genotyped at 21 days of age and positive transgenic pups were identified for treatment. Colony was propagated by back-breeding male pups of the same litter with the original female breeders to reduce phenotypic variance.

Derivation and Propagation of Human NSCs

Human NSCs were prepared from the cervical-upper thoracic cord of a single 8-week human fetus after an elective abortion. Tissues were donated by the mother in a manner fully compliant with the guidelines of the National Institutes of Health and the U.S. Food and Drug Administration. Spinal cord tissues cleared of meninges and dorsal root ganglia were dissociated into a single cell suspension by mechanical dissociation in serum-free, modified N2 medium and serially expanded in monolayer (9). Growth medium was changed every other day, and, on alternate days, 10 ng/ml of basic fibroblast growth factor (bFGF) was added to the culture. The first passage was conducted at 16 days postplating, a time point at which the culture was composed mostly of dividing NSCs and postmitotic neurons. Dividing cells were harvested by brief treatment with trypsin followed by dissociation and replated in new precoated plates. Cells were harvested at ∼75% confluence, which occurred within five or six days. This process was repeated for up to 20 passages. Cells from various passages were frozen in the growth medium plus 10% DMSO in liquid nitrogen. Upon thawing, recovery rate was 80–95%.

Passage 10–12 cells were used in this study. Five to seven days prior to surgery, one cryopreserved vial of the appropriate passage was thawed, washed, and cultured again. Cells were harvested by brief enzymatic treatment as described above, washed in buffered saline, couriered to the surgery site on wet ice, and used within 24 hr. Viability of cells on ice was typically greater than 80% within this 24-hour period. Immediately prior to grafting, the overwhelming majority of human NSCs expressed nestin, whereas ∼5% were immunoreactive for PSA-NCAM. Less than 1% expressed the neuronal markers type III D- tubulin epitope J1 (TUJ1) and MAP2 or the astroglial marker glial fibrillary acidic protein (GFAP).

Surgical Procedures

All surgical procedures were carried out according to protocols approved by the Animal Care and Use Committee of the Johns Hopkins Medical Institutions using gas anesthesia (enflurane: oxygen: nitrous oxide=1: 33: 66) and aseptic methods.

Live or dead NSCs were grafted into the lumbar protuberance (L4 & L5) of 62 day- old SOD1 G93A rats (220–300 g; mixed gender; n=45) mounted on a Kopf spinal stereotaxic unit under microscopic guidance. Twenty-seven subjects grafted with live (n=16) or dead (n=11) cells were used in motor testing followed by anatomical analyses. The remaining animals (n=18) were used for motor neuron survival, enzyme-linked immunosorbent assay (ELISA), Western blot, and real-time reverse transcriptase polymerase chain reaction (RT-PCR) studies equally divided between subjects grafted with live and subjects grafted with dead cells.

Dead cells were prepared by 3× freezing in liquid nitrogen (−70°F) and then thawing in room temperature. To confirm cell death of NSCs, 10 μl of the cell suspension was mixed with 40 μl of Hank’s balanced salt solution and 50 μl of stock Trypan Blue solution was added to the mixture. Solution was left for five min at room temperature and then a drop of it was placed on a hemacytometer and studied with light microscopy. All cells prepared as above were seen to take up the Trypan Blue stain, evidence that they were destroyed.

Cell suspensions were delivered under aseptic conditions via eight injections aimed at the ventral horn on both sides (1 μl with 5×104 NSC per injection site, four injection sites per side) with pulled-beveled glass micro-pipettes connected to 10 μl Hamilton microsyringes via silastic tubing. All rats received FK-506 (1 mg/kg i.p. daily) till euthanized to prevent immune rejection. Rats were killed with perfusion-fixation when their Basso, Beattie, and Bresnahan (BBB) score (see below) was <3.

Animal Testing

Rats were tested for motor strength and weight twice weekly. Motor strength tests included the BBB locomotor rating scale (10, 11) and the inclined plane scale (12). For BBB scoring, animals were tested for four to five min in an open field. For inclined plane scoring, rats were placed on the inclined mat and plane angle was adjusted to the highest point at which the animal could retain position for 5 seconds; this angle was then recorded as the subject’s inclined plane score. BBB and inclined plane scores were analyzed by repeated-measures analysis of variance (ANOVA) followed by Fisher LSD post-hoc test to assess differences between live-and dead-cell groups. Disease onset was defined as the point at which body weight was found decreased for the second consecutive time, i.e., a sensitive and very objective measure for determining disease onset in murine models of ALS (13). Course of illness as an effect of graft type (live- or dead-cell graft) was analyzed by comparing age at disease onset and age of death between the two groups (with student’s t test) as well as with Kaplan-Meier survival analysis followed by log-rank testing.

Histology, ICC, and Microscopy

Tissues were prepared from animals perfused with 4% freshly depolymerized, neutral-buffered paraformaldehyde based on protocols approved by the Animal Care and Use Committee of the Johns Hopkins Medical Institutions. The thoraco-lumbar spinal cord segments with attached roots and lumbar nerves were further fixed by immersion in the same fixative for an additional four hr after removing the dura. Blocks containing the entire grafted area plus 1 mm border above and below were cryoprotected and frozen for further processing. L3-S1 roots were processed separately as whole-mount preparations or after separating rootlets with heat-coagulated tips of glass pipettes. Blocks were sectioned (35 μm) at the transverse or saggital plane.

NSC survival and differentiation was studied with dual-label immunofluorescence that combined, in most cases, human nuclear protein antibody (Hnu) with another cellular marker and was performed essentially as described (6). Primary antibodies were from Chemicon International (Temecula, CA) except otherwise noted and included: mouse anti-HNu protein antibody (1:800); rabbit anti-TUJ1 (1:400; Research Diagnostics Inc., Flanders, NJ); rabbit anti-GFAP (1:400; Dako Carpinteria, CA); rabbit anti-human nestin (1:200); mouse anti-NF70 (1:100, human and porcine specific); goat anti-human glial cell line-derived neurotrophic factor (GDNF) (1:400, R&D Systems Inc. MN); mouse anti-SYN (1:200, human and hamster specific); mouse anti-BSN (1:400, rat and mouse specific; Stressgen, Victoria, BC); goat anti-ChAT IgG (1:100); Guinea pig anti-vesicular Glutamate Transporter (VGLUT) 1/2 (1:10,000, rat specific); and guinea pig anti-vesicular acetylcholine transporter (VAChT) (1:500). Dual immunofluorescence utilized an indirect protocol combining donkey anti-mouse and donkey anti-rabbit or donkey anti-goat IgG labeled with Cy3 or Cy2, respectively (1:200; Jackson ImmunoResearch, West Grove, PA). Sections were incubated in linking antibodies for 2–4 hr at room temperature and were then counterstained with the fluorescent DNA dye DAPI, coverslipped with DPX, and studied with epifluorescence or confocal microscopy. Triple immunofluorescence was used for the colocalization of HNu, Bassoon, and TUJ1; this procedure involved first a dual immunofluorescence step with donkey anti-rabbit IgG-Cy2 and donkey anti-mouse IgG-Cy3 and then an additional incubation in biotinylated HNu antibody followed by treatment with streptavidin-AMCA (Jackson). Digital images were optimized for brightness/contrast and resolution (600 ppi) with the aid of Adobe PhotoShop 6.0 software (Adobe Systems, San Jose, CA).

Cell Counts for Differentiated NSCs and Stereological Counts for Surviving α-Motor Neurons

To study NSC differentiation, we used a nonstereological method of counting the total number of HNu (+) cells, as well as cells dually labeled with HNu and a protein marker of neural cell differentiation on randomly selected high-power (100×) fields from our immunofluorescent preparations. One field in each of 6 sections spaced ∼1 mm apart through the graft area was used from each animal. Numbers of HNu (+) and double-labeled profiles were pooled from all six fields counted from each case and grouped per experimental protocol. Average numbers of single- and double-labeled cells were generated for the dead-or live-cell group (n=6 per group).

To assess motor neuron survival in rats grafted with live or dead cells (n=3 per group), we used tissues from animals sacrificed at 128 days of age, i.e. a midpoint in the course of disease for subjects that received dead-cell grafts. Every sixth section in the L3–S1 region from each animal was sampled as per stereological requirements (6) and stained with cresyl violet (Nissl staining) for motor neuron counting. α-motor neurons, identified as multipolar cells with distinct nucleus and a soma diameter >35 μm (14) were counted with the optical fractionator probe using the Stereo Investigator V hardware and software (MicroBrightField Inc., Williston, VT) as described (6). Differences between animals grafted with live versus dead cells were analyzed with a Student’s t test.

ELISA for Motor Neurotrophic Factors: GDNF and Brain-Derived Neurotrophic Factor (BDNF)

Cerebrospinal fluid (CSF) was sampled from the fourth ventricle of animals (n=3 per dead-or live-cell group) under gas anesthesia. Animals were then euthanized by decapitation and tissue samples containing grafting sites and areas adjacent to those were dissected from transverse 1 mm-thick spinal cord slices. CSF or cord samples were processed as described (15) and levels of GDNF and BDNF protein were measured with the E-Max ImmunoAssay system (Promega, Madison, WI). Reaction product was revealed with HRP-conjugated chicken IgY antibody and tetra-methyl-benzidine (TMB) as chromogen. TMB absorbance was read with a ThermoMax microplate reader at 450 nm. Variance in concentrations among samples from live-cell grafts, areas adjacent to grafts and dead-cell grafts was analyzed with one-way ANOVA followed by a Tukey’s Multiple Comparison posthoc test. Difference in CFS concentration of GDNF and BDNF between animals grafted with live versus dead cells was analyzed with student’s t test.

Western Blotting for GDNF and BDNF

Protein samples from CSF or spinal cord prepared as for ELISA were electrophoresed on 12% NuPAGE precast gels (Invitrogen, Carlsbad, CA) and transferred on to nitrocellulose membranes (BA-S 85; Schleicher & Schuell, Keene, NH). Blots were blocked in TBS (pH 7.4) containing 5% donkey serum, and then incubated in GDNF and BDNF antibodies (1:500; overnight, 4°C), followed by HRP-linked donkey anti-goat IgG (for GDNF) and anti-rabbit IgG (for BDNF) (1:2000; Jackson) (1 hr, room temperature). All antibodies were diluted in TBS containing 5% donkey serum. Blots were developed with the SuperSignal Chemiluminescent Substrate (Pierce) and exposed to Kodak-XAR film (Eastman Kodak, Rochester, NY). Blots were then striped and re-blotted with β-actin antibody (1:500, Sigma), followed by HRP-linked donkey anti-mouse IgG (1:10000, Jackson). Immunoreactive bands were analyzed with Bio-Rad Quantity One software (Bio-Rad Laboratories, Hercules, CA). Band density ratios (GDNF or BDNF: β-actin) were calculated per animal and group means were entered for statistical analysis as in the case of ELISA experiments.

Real-time RT-PCR for BDNF, GDNF, and Vascular Endothelial Growth Factor (VEGF)

Cultured NSCs prior to grafting and homogenized spinal cord tissues (experimental n=4; control n=3) were subjected to RNA extraction as described (6). Template cDNA was reverse transcribed with the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA) and SYBR green-based real-time PCR was performed with the iCycler (Bio-Rad). Human- and rat-specific primers were synthesized by Integrated DNA Technologies (Coralville, IA). SDHA and RPL13a served, respectively, as the human- and rat-specific reference sequences used for normalization. PCR reactions, run in triplicate for each sample, contained diluted cDNA (1:1; 1 μl), 10 μM of sense and antisense primer stocks (0.5 μl each), iQ SYBR green Supermix (Bio-Rad) (12.5 μl) and of nuclease-free water (10.5 μl) in 25 μl volume. PCR cycling conditions were 95°C for five min, 35 cycles of 95°C for 30 sec, 60°C for 30 sec, and 72°C for 30 sec. Reactions without template served as negative controls. Reactions without RT containing 25 ng RNA in 1 μl served as RT (−) controls. Melting curve analysis was carried out by heating the amplicon from 60 to 95°C in 71 0.5°C increments. Efficiency curves were created for each amplicon using five duplicate twofold dilutions, corresponding to 6.25–100 ng of the initial total RNA input into the RT reaction. Data were analyzed using the Bio-Rad Gene Expression Macro version 1.1 for Microsoft Excel. Normalized gene expression ratios were calculated as described (16).

Levels of rat- and human-specific gene expression in spinal cords with live grafts were expressed as fold changes from baseline expression in spinal cords with dead NSCs and from live NSCs prior to grafting. Standard deviations for normalized expression values were calculated as described (17).


Survival, Neuronal Differentiation and Structural Integration of Human NSCs in the Spinal Cord of SOD1 G93A Rats

Human NSCs were identified by their HNu protein signature and their phenotypic fates were tracked with dual immunocytochemistry (ICC) for HNu and epitopes specific for neural precursor, neuronal, and glial cells. At the end of experiments in SOD1 G93A rats, human NSCs showed robust engraftment and excellent long-term survival (Fig. 1A). A majority of HNu(+) cells (70.4%) had differentiated into the neuronal lineage, based on their colocalization of TUJ1 (Fig. 1B–D). Approximately one-fifth (19.2%) of HNu (+) cells colocalized nestin, and very few (1.3%) HNu (+) cells were positive for GFAP.

Survival and neuronal differentiation of human NSCs in the spinal cord of SOD1 G93A rats. (A) This section was taken from a representative graft site of an end-stage animal and stained with HNu ICC (red). A large number of HNu (+) cells are shown to survive at the original graft site in ventral horn/ventral root exit zone (arrows). (B–D) These fields are photographed from a section dually stained with HNu (red) and the neuronal marker TUJ1 (green) and show that a majority of HNu (+) cells in this ventral L5 graft are also TUJ1 (+). (C) represents an enlargement of the framed area in (B). (D) is a confocal image taken from the framed area in (C). (D) was optically resectioned perpendicular to the z plane to confirm the colocalization of TUJ1 (+) cytoplasm and HNu (+) nucleus at the same plane. Blue emission channel for DAPI counterstain is merged only in (D). Size bars: A, 150μm; B, 100μm; C, 20μm; D, 10μm.

The capacity of human NSCs to integrate within the host circuitry was tested with perikaryal markers for graft/host cells and markers selective for either host or graft axons/terminals. A large number of choline acetyltransferase (ChAT) (+) host motor neurons were found to be contacted by human synaptophysin (+) synaptic boutons of graft origin (Fig. 2A–B). Conversely, a large number of rat-derived synaptic boutons was seen in close proximity to graft-derived neurons in preparations stained with either a mixture of antibodies for VGLUT1/2 (present in host, but not in graft terminals), HNu and TUJ1 or a combination of antibodies for the presynaptic protein Bassoon (that recognizes rat and mouse but not human epitopes), HNu, and TUJ1. Synaptic contacts of rat nerve terminals on graft-derived neurons were not as common as human nerve terminals on host motor neurons (Fig. 2C–D).

Structural integration of human NSCs in the spinal cord of SOD1 G93A rats. Images depict patterns of graft-to-host innervation (A–B), host-to-graft innervation (C–D), and pathways of graft-derived axons forming within the host spinal cord (E–H). (A–B) These images were taken from a section that was dually stained for ChAT (green) and human-specific synaptophysin (red) ICC at lower (A) and higher (B, confocal) magnifications. Confocal image in panel B was taken from the framed area in (A). Section in (B) was counterstained with DAPI (blue). As shown for many host motor neurons in (A) and with a confocal level of detail in (B), cell bodies and proximal dendrites of host α-motor neurons are contacted by a large number of graft-derived boutons. In (B), several of these synaptic contacts are confirmed with x- and y-plane reconstruction (arrows). (C–D) These photographs were taken from two sections processed with triple immunofluorescence for HNu (blue), TUJ1 (green), and the rat form of the presynaptic protein Bassoon (red). Section in (C) was photographed under epifluorescence. Panel D represents a confocal detail from an adjacent section. Note a general overlap of host-derived terminals (red) with a TUJ1 differentiated graft in (C) (arrows). When examined in greater detail, i.e. with confocal microscopy (D), only few graft-derived neurons are seen to receive synaptic contacts by host cells (asterisk in main frame, arrow in x-plane reconstruction). (E–E′) This digital photograph, taken from a horizontal section stained for HNu and human NF-70 ICC (both red) and counterstained with ChAT ICC (green) was processed for the acquisition of red emission (E) or merged red and green emission (E′). A large number of graft-derived axons are shown in (E) to leave the graft on top and course preferentially along the left border of this field (arrows). The superimposition of green ChAT immunostaining in cells/processes in (E′) serves to indicate the position of host motor neurons in the ventral horn (asterisks) and to define the position of graft-derived pathway in the white matter of the ventral funiculus. (F) In all cases with live grafts, there was a prominent visualization of human axons and synapses around ependymal cells. Although a de novo projection of NSC-derived neurons to ependyma could not be ruled out, the majority of these processes resulted from migration of NSCs to ependymal sites. This image illustrates human synaptophysin (+) synapses apposed to DAPI-stained ependymal cells. (G) This teased preparation from ventral L5 root was stained en block with human NF-70 and shows sparse graft-derived axons coursing near the surface (arrows). Size bars: A, 50 μm; B, 10 μm; C, 10 μm; D, 20 μm; E, 100 μm; F, 10 μm.

Besides projecting locally to innervate host motor neurons, human NF70 (+) axons from differentiated NSCs formed bundles that coursed in the ventral funiculus and crossed over at least one to two spinal segments. The preference of these axonal bundles for white versus gray matter was evident in preparations stained for markers that revealed the boundaries of ventral horn (Fig. 2E). A dense plexus of human NF70 (+) axons and human synaptophysin (+) terminals was consistently observed in juxtaposition to ependymal cells (Fig. 2F), likely resulting from targeted migration of differentiated NSCs rather than the formation of de novo graft-to-ependyma projection. Only a few human NF70 (+) axons were found in teased or en block preparations of ventral roots (Fig. 2G).

NSC Grafts into the Lumbar Cord of SOD1 G93A Rats Prolong Lifespan and Delay Motor Neuron Death and Disease Onset and Progression

Animals grafted with live NSCs showed increased survival by both Kaplan-Meier (Fig. 3A) and end-point analysis; the average lifespan for animals grafted with dead NSCs was 138 days, whereas rats grafted with live NSCs lived for 149 days (Fig. 3C). Average time-to-disease-onset was 115 days for animals that received dead cells and 122 days for animals that were grafted with live NSCs (Fig. 3D). Time plots of BBB open field and inclined plane test scores show a significantly slower progression in muscle weakness in animals grafted with live cells compared to animals that received dead NSCs (Fig. 3B).

Effects of human NSC treatment on severity of motor neuron disease in G93A SOD1 rats shown with progression (A–B) as well as endpoint (C–E) analysis of clinical and pathological measures in cases with live-cell (L, red) and dead-cell (control, C) grafts (blue). (A) Kaplan-Meier plot showing a significant separation between experimental (n=16) and control animals (n=11) throughout the course of observation (P=0.0003). (B) Separation in the two principal measures of muscle weakness (BBB and inclined plane scores) between the two groups (p=0.00168 and 0.00125, respectively). End-point analysis of survival (C), time-to-disease-onset (D), and motor neuron numbers based on Nissl staining (E) in experimental (n=3) and control rats (n=3). Bar diagram in (C) shows a significant 11-day difference in life span between the two groups (P=0.0005), and diagram in (D) indicates a significant 7-day difference in time-to-disease-onset between the two groups (p=0.0001). (E) depicts a difference of 3,212 cells in the lumbar protuberance between live and dead NSC groups (p=0.01). Inset at the bottom of (E) illustrates the difference in motor neuron survival between a representative experimental (upper) and control (lower) rat at 128 days of age; arrows indicate the lateral motor neuron group. Size bars: 150 μm.

The effect of NSCs on motor neuron survival in the lumbar protuberance (L3–S1) of Tg rats was examined in a small group of animals that received live or dead NSCs and were sacrificed at 128 days of age. Stereologically estimated numbers of α-motor neurons were 6,418 for animals that received live NSCs and 3,206 for rats grafted with dead NSCs, i.e. there were twice as many neurons in the lumbar protuberance of experimental compared to control animals of the same age (Fig. 3E).

Molecular Correlates of Clinical and Biological Effects of Human NSCs in SOD1 G93A Rats: Evidence for the Expression and Release of Neurotrophins and Trophic Cytokines, with Emphasis on GDNF

To explore potential mechanisms of neuroprotection afforded by human NSCs on degenerating motor neurons, we studied the expression and release of BDNF and GDNF, two peptides with classical trophic effects on mammalian motor neurons (18–21), by Western blotting and ELISA using both spinal cord preparations and CSF samples.

In the case of GDNF, ELISA shows a threefold increase in the release of this trophic peptide in the spinal cord (Fig. 4A, left) and a fivefold increase in GDNF secreted in the CSF (Fig. 4A, right) in animals with live NSCs. Western blotting confirms the ELISA pattern of increase and shows a normalized GDNF density of 0.860 in live-cell grafts and 0.708 in dead-cell grafts (Fig. 4B). In the case of BDNF, there was an eightfold increase in BDNF concentration in the spinal cord (Fig. 4C, left) and a fourfold increase in the CSF (Fig. 4C, right) of animals grafted with live cells. BDNF was not detectable by Western blotting in our hands.

Expression and release of GDNF and BDNF in the spinal cord of NSC-grafted SOD1 G93A rats. (A) GDNF concentrations in the parenchyma and CSF of rats grafted with live cells (red and orange bars, n=3) and animals grafted with dead cells (blue bars, n=3) by ELISA. Red bars represent concentrations through the graft site; orange bars depict concentrations in tissues one segment above or below. Average GDNF concentration was 0.912 pg/μg at the graft site and 0.819 pg/μg one spinal segment away in animals grafted with live cells; in animals that received dead cells, average graft-site concentration was 0.368 pg/μg (left). In the CSF, GDNF concentration was 0.027 pg/μl in experimental and 0.006 pg/μl in control animals (right). Variance in parenchymal concentrations among groups is significant and caused by a large difference between red or orange and blue groups. Difference in CSF concentrations between experimental (live-cell, L) and control (dead-cell, C) groups are also significant by t test. (B) GDNF Western blotting, serving as confirmation of ELISA, detects a 16 kDa protein (left) and shows a higher normalized GDNF concentration in animals grafted with live NSCs. (C) BDNF concentrations in the parenchyma and CSF of experimental rats (red and orange bars, n=3) and controls (blue bars, n=3) by ELISA. BDNF concentration was 0.086 pg/μg at the graft site and 0.054 pg/μg one segment away in animals grafted with live cells. In rats grafted with dead cells, graft-site concentration was 0.010 pg/μg. In the CSF, BDNF concentration was 0.041 pg/μl in animals with live cells and 0.010 pg/μl in animals with dead cells. Variance in these values is significant because of large differences between live- and dead-cell grafts, but also between graft sites and sites adjacent to them in animals with live NSCs (left). Differences between experimental and control CSF concentrations are also significant (right). (D) BDNF and GDNF real-time RT-PCR demonstrates that spinal cord tissues with human NSC grafts express higher levels of human, but lower or unchanged levels of rat BDNF and GDNF mRNA. Melting curve and agarose gel analysis detected single melt peaks and specific bands for each of the eight rat and human-specific primers (data not shown). PCR efficiency values were between 92 and 105% for all amplicons. RT minus, NTC, rat cDNA (for human-specific primers), and human NSC cDNA (for rat-specific primers) control reactions did not amplify any specific product. The normalized expression ratio of human BDNF and GDNF in grafts is ∼8-fold and 9-fold higher, respectively, compared to levels of expression in NSCs prior to grafting. In contrast, host tissues with live NSC grafts express ∼3.5 times lower BDNF, whereas GDNF expression does not change compared to control tissues grafted with dead NSCs. *P<0.05. **P<0.01.

The ELISA data suggest a more widespread secretion of GDNF compared to BDNF in animals grafted with live NSCs, especially in the CSF. To explore the source of the excess trophic peptide, i.e. differentiate between graft (human) and host (rat) origin, we performed real-time RT-PCR analysis for GDNF and BDNF mRNA expression on samples from live- and dead-cell graft sites using human- and- rat-specific primers (Fig. 4D). In addition, we studied the expression of VEGF, an angiogenic trophic peptide with evident trophic effects in mammalian motor neurons (22). Normalized human GDNF and BDNF mRNA expression in spinal cords grafted with live cells was several-fold higher than that in NSCs prior to grafting; we found an eightfold change for human BDNF and a ninefold increase for human GDNF. Rat BDNF mRNA expression in spinal tissues with live NSC grafts decreased by 3.5-fold, whereas rat GDNF expression remained essentially unchanged when compared to tissues with dead NSC grafts. Human VEGF mRNA expression was found to be ∼2 times higher in grafted tissues compared to a pregrafting sample of NSCs, and rat VEGF mRNA expression decreased ∼3 times compared to levels in spinal cords with dead-cell grafts.

The graft origin of GDNF, i.e. the trophic peptide with the greatest upregulation and widespread secretion in subjects with live grafts, was further supported with GDNF ICC. In animals with live grafts, the vast majority of grafted HNu (+) cells were found to express immunoreactive GDNF (Fig. 5A–D). GDNF immunoreactivity was shown to be localized within NSC-derived terminals, such as in the dense terminal fields around ependymal cells (Fig. 5E). Moreover, spinal cords with live grafts exhibited a high density of GDNF- immunoreactive boutons attached to cell bodies and proximal dendrites of host motor neurons. These boutons were especially prominent in animals with early motor neuron disease (BBB score 19 or higher) and were less prominent in animals with advanced disease (BBB score 6 or lower) (Fig. 6A). In contrast to terminals near ependymal cells, the vast majority of NSC terminals contacting host motor neurons did not contain immunoreactive GDNF (Fig. 6B–C). On the other hand, GDNF (+) terminals on motor neuron cell bodies were found to colocalize ChAT and VAChT, i.e. classical markers of cholinergic nerve terminals, in early disease stages (Fig. 6D–E). The size, shape, and neurotransmitter identity of these GDNF (+) terminals identifies them as cholinergic C-boutons, i.e. terminals of segmental spinal origin that participate in local motor circuits (23). In concert, although most graft-derived differentiated cells express GDNF and the excess GDNF expression in grafted cords is of human origin, the terminals of graft-derived cells on host motor neurons rarely contain GDNF immunoreactivity. The increased GDNF immunoreactivity within synapses belonging to host neurons in an environment with substantially increased human and unchanged rat GDNF expression raises the suspicion of a transsynaptic transfer of GDNF from graft cell terminals onto host motor neurons and/or interneurons (24).

Localization of GDNF immunoreactivity in cell bodies (A-D) and terminals (E) of NSC-derived neurons in the spinal cord of SOD1 G93A rats. (A–B) These dually stained preparations for HNu (red) and GDNF (green) illustrate the abundance of GDNF immunoreactivity within the cytoplasm of grafted NSCs under low-power epifluorescence. (A) was photographed in multiple emission acquisition channels such as to allow for the visualization of green only (left), red only (middle), and merged red and green (right) epifluorescence. (B) was taken from a section adjacent to the one in (A) that was stained for HNu and rabbit IgG immunoreactivity to control for GDNF antibody background; image in (B) is the product of merged green and red channels. Note the delineation of the graft area by intense GDNF immunoreactivity in (A) and the abolition of specific immunoreactivity when anti-GDNF is replaced with pre-immune rabbit IgG (B). (C–D) These two merged-emission photographs serve to illustrate in greater detail the cytoplasmic GDNF immunoreactivity of NSC grafts. Red and green emissions represent HNu and GDNF immunoreactivity, as in (A–B). Confocal image in (D) is taken from the framed area of the epifluorescent image in (C). Confocal image was optically resectioned at the x and y planes to confirm GDNF and HNu colocalization. (E–F) These images were taken through the central canal region of sections dually stained for human synaptophysin (red) and GDNF (green) and were photographed under epifluorescent (E) or confocal (F) conditions for the acquisition of merged emission of red and green fluorescence and blue emission from DAPI staining. Note the precise colocalization of human synaptophysin, serving as a graft-specific synaptic marker, and GDNF immunoreactivity in the dense terminal field near ependymal cells (asterisks). Colocalization is especially evident with confocal microscopy in (F). (Index double-labeled synapses are indicated with arrows.) Size bars: A, B 150 μm; C, 50 μm; D, 10 μm; E, 10 μm.
Localization of GDNF immunoreactivity in synapses associated with host motor neurons in SOD1 G93A rats at different stages of disease progression. (A) These three panels are representative illustrations of GDNF immunoreactivity in motor neurons from animals grafted with live (left and right) and dead (center) NSCs at early and late-stage disease. Note the high density of GDNF (+) boutons, especially on the cell bodies and proximal dendrites of host motor neurons in early disease (left, arrows). These terminals are smaller with advanced disease (right, arrows). There is also granular GDNF immunoreactivity within the cytoplasm, best appreciated in confocal images (C′). (B–C′) Confocal images of a host motor neuron stained with GDNF (green) and human synaptophysin (red; to selectively label NSC-derived terminals). (B), corresponding to early motor neuron disease, shows the rare localization of GDNF within human synaptophysin (+) synapses (arrow). In late disease (C–C′), most immunoreactive GDNF appears in vesicular structures within host motor neurons (arrows in C′). (C′) is an enlargement of the framed area in (C). (D–E) Confocal images of a representative host motor neuron from sections dually stained with GDNF and ChAT or GDNF and VAChT. ChAT and VAChT are cholinergic markers, the former labeling both cholinergic cell bodies and terminals and the latter labeling local cholinergic terminals. GDNF immunoreactivity is detected with green and ChAT and VAChT with red emission. The cholinergic cell body is contacted by multiple ChAT or VAChT and GDNF (+) large terminals with the appearance of cholinergic C-boutons (arrows). X and y plane reconstruction verifies the colocalization of ChAT or VAChT and GDNF on these terminals (arrows). Size bars: A, 20μm; B, 10 μm; C, 10 μm; D, 10 μm; E, 10 μm.


Our findings show that NSCs can survive and differentiate into neurons in the degenerative spinal cord environment of SOD1 G93A rats. In addition, these grafts, whose immune rejection can be effectively prevented with FK-506, afford both clinical and biological benefits that are powerful and significant. The potency of this effect can be best appreciated if one considers the fact that, for a disseminated illness like ALS, lumbar cord grafting is a partial approach that omits other vital portions of the segmental motor apparatus, i.e. the cervical motor neuron column responsible for respiratory movements. However, we cannot rule out that the release of BDNF and GDNF into the CSF might have had a broader effect on host motor neurons throughout the cord. Wider grafting procedures, i.e. including both lumbar and cervical regions have significantly higher mortality, but we have had recent success with a limited number of grafting sites involving both regions (L. Xu, J. Yan, and V.E. Koliatsos, personal observations).

The apparent resistance of grafted NCSs to the ongoing degenerative process in the ventral horn of G93A SOD1 rats is especially promising. The survival and extensive differentiation of NSCs reported here is a strong indication that the inflammatory/excitotoxic signaling involving motor neurons harboring G93A SOD1 (1–3) has no evident toxicity on cells without adverse genetic properties. This factor alone raises optimism for future cellular strategies using grafts to restore motor function in degenerative MND.

The magnitude of clinical and biological effects reported here is in the neighborhood of effects reported in other studies on experimental therapeutics of motor neuron disease in SOD1 transgenic rodents that focus on prevention of cell death and promotion of motor neuron survival. These trials have tested a number of small organic compounds and trophic factors and have included: copper chelators (25); compounds targeting microglial-inflammatory and glutamatergic processes (26, 27); selective (28, 29) as well as nonselective (29) cyclooxygenase-2 inhibitors alone or in combinations with other compounds; GDNF retrogradely delivered to motor neurons from muscle with adeno-associated virus or via motor neuron transfer with lentivirus (30–32); and VEGF transferred to motor neurons via lentiviral vector injected into various muscles (33) or via direct delivery (34, 35). With the exception of a recent study using intraventricular VEGF (35), the vast majority of these studies have involved SOD1 Tg mice (36, 37) in which MND tends to be of later onset and milder progression than in the SOD1 Tg rats used here (2, 8). The only other example of NSC treatment of SOD1 Tg rodents known to us is a study in which investigators grafted GDNF-engineered, cortically derived human neural progenitors in the spinal cord of G93A SOD1 rats and found that these cells survived well and expressed high levels of immunoreactive GDNF; most of these cells remained at a nestin (+) precursor state and showed no apparent structural integration or clinical benefits (11).

Our experiments have also begun to identify some cellular and molecular mechanisms mediating the therapeutic effects of NSCs in Tg MND. Although NSCs engage in reciprocal connections with host motor neurons, they do not appear to exert therapeutic effects by replacing degenerated neuromuscular units. Based on our findings on the expression and release of trophic peptides, at least a portion of the effects of NSCs on degenerating motor neurons may be their delivery, via classical mechanisms including transsynaptic transfer, of neurotrophins and especially trophic cytokines to degenerating host motor neurons. In other words, the surprising establishment of synaptic appositions between grafted human NSCs and rat motor neurons, also observed in a recent study between human NSC grafts and mouse motor neurons (38), may play a significant therapeutic role beyond the obvious advantage of strengthening/amplifying local circuits. The significance of NSC-derived GDNF for the survival of motor neurons has also been stressed in another in vitro study showing a protective effect of this trophic cytokine on rat spinal cord explants exposed to an excitotoxic protocol (39).

Until further studies elaborate ways in which NSC- derived neurons can be induced to project into ventral roots and innervate muscle in animals with degenerative MND, the apparent advantage of NSCs over existing pharmacological strategies may be that trophic peptides secreted by NSCs are delivered to degenerating motor neurons under conditions of optimal bioavailability. The success of the experiments reported here raise further hopes for using rodent models in the preclinical evaluation of stem cells for cell replacement in degenerative MND.


This work was supported by National Institutes of Health grant RO1 NS045140-03, the Muscular Dystrophy Association, and the Robert Packard Center for ALS Research at Johns Hopkins.


1. Rothstein JD, Martin LJ, Kuncl RW. Decreased glutamate transport by the brain and spinal cord in amyotrophic lateral sclerosis. N Engl J Med 1992; 326: 1464.
2. Howland DS, Liu J, She Y, et al. Focal loss of the glutamate transporter EAAT2 in a transgenic rat model of SOD1 mutant-mediated amyotrophic lateral sclerosis (ALS). Proc Natl Acad Sci U S A 2002; 99(3): 1604.
3. Turner BJ, Atkin JD, Farg MA, et al. Impaired extracellular secretion of mutant superoxide dismutase 1 associates with neurotoxicity in familial amyotrophic lateral sclerosis. J Neurosci 2005; 25(1): 108.
4. Park KI, Ourednik J, Ourednik V, et al. Global gene and cell replacement strategies via stem cells. Gene Ther 2002; 9(10): 613.
5. Cao QL, Howard RM, Dennison JB, Whittemore SR. Differentiation of engrafted neuronal-restricted precursor cells is inhibited in the traumatically injured spinal cord. Experimental Neurology 2002; 177(2): 349.
6. Yan J, Welsh AM, Bora SH, et al. Differentiation and tropic/trophic effects of exogenous neural precursors in the adult spinal cord. J Comp Neurol 2004; 480(1): 101.
7. Yan J, Welsh AM, Hazel T, et al. Grafted human neural stem (NS) cells differentiate into neurons, migrate long distance and project axons in spinal cord and the roots of adult rats. Soc Neurosci Abstr 2003; 29: 150.19.
8. Nagai M, Aoki M, Miyoshi I, et al. Rats expressing human cytosolic copper-zinc superoxide dismutase transgenes with amyotrophic lateral sclerosis: associated mutations develop motor neuron disease. J Neurosci 2001; 21(23): 9246.
9. Johe KK, Hazel TG, Muller T, et al. Single factors direct the differentiation of stem cells from the fetal and adult central nervous system. Genes Dev 1996; 10(24): 3129.
10. Basso DM, Beattie MS, Bresnahan JC. A sensitive and reliable locomotor rating scale for open field testing in rats. J Neurotrauma 1995; 12(1): 1.
11. Klein SM, Behrstock S, McHugh J, et al. GDNF Delivery Using Human Neural Progenitor Cells in a Rat Model of ALS. Hum Gene Ther 2005; 16(4): 509.
12. Rivlin AS, Tator CH. Objective clinical assessment of motor function after experimental spinal cord injury in the rat. J Neurosurg 1977; 47(4): 577.
13. Weydt P, Hong SY, Kliot M, Moller T. Assessing disease onset and progression in the SOD1 mouse model of ALS. Neuroreport 2003; 14(7): 1051.
14. Swett JE, Wikholm RP, Blanks RH, et al. Motoneurons of the rat sciatic nerve. Exp Neurol 1986; 93(1): 227.
15. Sheng JG, Bora SH, Xu G, et al. Lipopolysaccharide-induced- neuroinflammation increases intracellular accumulation of amyloid precursor protein and amyloid beta peptide in APPswe transgenic mice. Neurobiology of Disease 2003; 14(1): 133.
16. Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res 2001; 29(9).
17. Vandesompele J, De Preter K, Pattyn F, et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol 2002; 3(7): 0034.
18. Yan Q, Elliott J, Snider WD. Brain-derived neurotrophic factor rescues spinal motor neurons from axotomy-induced cell death. Nature 1992; 360: 753.
19. Yan Q, Matheson C, Lopez OT. In vivo neurotrophic effects of GDNF on neonatal and adult facial motor neurons. Nature 1995; 373: 341.
20. Henderson CE, Phillips HS, Pollock RA, et al. GDNF: a potent survival factor for motoneurons present in peripheral nerve and muscle. Science 1994; 266: 1062.
21. Koliatsos VE, Clatterbuck RE, Winslow JW, et al. Evidence that brain-derived neurotrophic factor is a trophic factor for motor neurons in vivo. Neuron 1993; 10(3): 359.
22. Van Den BL, Storkebaum E, Vleminckx V, et al. Effects of vascular endothelial growth factor (VEGF) on motor neuron degeneration. Neurobiol Dis 2004; 17(1): 21.
23. Hellstrom J, Oliveira ALR, Meister B, Cullheim S. Large cholinergic nerve terminals on subsets of motoneurons and their relation to muscarinic receptor type 2. J Comp Neurol 2003; 460(4): 476.
24. Rind HB, Butowt R, von Bartheld CS. Synaptic targeting of retrogradely transported trophic factors in motoneurons: Comparison of glial cell line-derived neurotrophic factor, brain-derived neurotrophic factor, and cardiotrophin-1 with tetanus toxin. J Neurosci 2005; 25(3): 539.
25. Hottinger AF, Fine EG, Gurney ME, et al. The copper chelator d-penicillamine delays onset of disease and extends survival in a transgenic mouse model of familial amyotrophic lateral sclerosis. Eur J Neurosci 1997; 9: 1548.
26. Kriz J, Nguyen MD, Julien JP. Minocycline slows disease progression in a mouse model of amyotrophic lateral sclerosis. Neurobiol Dis 2002; 10(3): 268.
27. Kriz J, Gowing G, Julien JP. Efficient three-drug cocktail for disease induced by mutant superoxide dismutase. Ann Neurol 2003; 53(4): 429.
28. Klivenyi P, Kiaei M, Gardian G, et al. Additive neuroprotective effects of creatine and cyclooxygenase 2 inhibitors in a transgenic mouse model of amyotrophic lateral sclerosis. J Neurochem 2004; 88(3): 576.
29. Kiaei M, Kipiani K, Petri S, et al. Integrative role of cPLA with COX-2 and the effect of non-steriodal anti-inflammatory drugs in a transgenic mouse model of amyotrophic lateral sclerosis. J Neurochem 2005; 93(2): 403.
30. Acsadi G, Anguelov RA, Yang H, et al. Increased survival and function of SOD1 mice after glial cell-derived neurotrophic factor gene therapy. Hum Gene Ther 2002; 13(9): 1047.
31. Wang LJ, Lu YY, Muramatsu S, et al. Neuroprotective effects of glial cell line-derived neurotrophic factor mediated by an adeno-associated virus vector in a transgenic animal model of amyotrophic lateral sclerosis. J Neurosci 2002; 22(16): 6920.
32. Guillot S, Azzouz M, Deglon N, et al. Local GDNF expression mediated by lentiviral vector protects facial nerve motoneurons but not spinal motoneurons in SOD1(G93A) transgenic mice. Neurobiol Dis 2004; 16(1): 139.
33. Azzouz M, Ralph GS, Storkebaum E, et al. VEGF delivery with retrogradely transported lentivector prolongs survival in a mouse ALS model. Nature 2004; 429(6990): 413.
34. Zheng C, Nennesmo I, Fadeel B, Henter JI. Vascular endothelial growth factor prolongs survival in a transgenic mouse model of ALS. Ann Neurol 2004; 56(4): 564.
35. Storkebaum E, Lambrechts D, Dewerchin M, et al. Treatment of motoneuron degeneration by intracerebroventricular delivery of VEGF in a rat model of ALS. Nat Neurosci 2005; 8(1): 85.
36. Gurney ME, Pu H, Chiu AY, et al. Motor neuron degeneration in mice that express a human Cu, Zn superoxide dismutase mutation. Science 1994; 264: 1772.
37. Wong PC, Pardo CA, Borchelt DR, et al. An adverse property of a familial ALS-linked SOD1 mutation causes motor neuron disease characterized by vacuolar degeneration of mitochondria. Neuron 1995; 14: 1105.
38. Cummings BJ, Uchida N, Tamaki SJ, et al. Human neural stem cells differentiate and promote locomotor recovery in spinal cord-injured mice. Proc Natl Acad Sci U S A 2005; 102(39): 14069.
39. Llado J, Haenggeli C, Maragakis NJ, et al. Neural stem cells protect against glutamate-induced excitotoxicity and promote survival of injured motor neurons through the secretion of neurotrophic factors. Mol Cell Neurosci 2004; 27(3): 322.

Amyotrophic lateral sclerosis; Motor neuron; Spinal cord; Regeneration; Transplantation

© 2006 Lippincott Williams & Wilkins, Inc.