OUP user menu

Neural stem cells LewisX + CXCR4 + modify disease progression in an amyotrophic lateral sclerosis model

Stefania Corti, Federica Locatelli, Dimitra Papadimitriou, Roberto Del Bo, Monica Nizzardo, Martina Nardini, Chiara Donadoni, Sabrina Salani, Francesco Fortunato, Sandra Strazzer, Nereo Bresolin, Giacomo P. Comi
DOI: http://dx.doi.org/10.1093/brain/awm043 1289-1305 First published online: 17 April 2007


Amyotrophic lateral sclerosis (ALS) is a fatal neurological disease characterized by the degeneration of the motor neurons. We tested whether treatment of superoxide dismutase (SOD1)-G93A transgenic mouse, a model of ALS, with a neural stem cell subpopulation double positive for Lewis X and the chemokine receptor CXCR4 (LeX+CXCR4+) can modify the disease's progression. In vitro, after exposure to morphogenetic stimuli, LeX+CXCR4+ cells generate cholinergic motor neuron-like cells upon differentiation. LeX+CXCR4+ cells deriving from mice expressing Green Fluorescent Protein in all tissues or only in motor neurons, after a period of priming in vitro, were grafted into spinal cord of SOD1-G93A mice.

Transplanted transgenic mice exhibited a delayed disease onset and progression, and survived significantly longer than non-treated animals by 23 days. Examination of the spinal cord revealed integration of donor-derived cells that differentiated mostly in neurons and in a lower proportion in motor neuron-like cells. Quantification of motor neurons of the spinal cord suggests a significant neuroprotection by LeX+CXCR4+ cells. Both VEGF- and IGF1-dependent pathways were significantly modulated in transplanted animals compared to controls, suggesting a role of these neurotrophins in MN protection.

Our results support the therapeutic potential of neural stem cell fractions through both neurogenesis and growth factors release in motor neuron disorders.

  • neural stem cell
  • transplantation
  • motor neuron
  • amyotrophic lateral sclerosis


Amyotrophic lateral sclerosis (ALS) is a fatal neurological disorder involving the selective degeneration of motor neurons (MNs) in the brain and spinal cord leading to rapidly progressive paralysis. At present no effective treatment exists.

Although the precise pathogenesis of ALS remains to be elucidated, mutations in Cu/Zn superoxide dismutase (SOD1) account for about 20% of familial ALS cases (Rosen et al., 1993).

Transgenic mice over-expressing human mutant SOD1 develop MN degeneration and an ALS-like phenotype providing a useful animal model for this disease (Gurney et al., 1994).

Recent data from chimeric mice deriving from normal and SOD1 mutant-expressing cells and obtained by injection of wild-type (wt) embryonic stem (ES) cells into SOD1 blastocysts demonstrated the non-cell autonomous toxicity of ALS-causing SOD1 mutants and the influence of environmental cells on the MNs (Clement et al., 2003). In addition, this study pointed out the role of cell replacement therapy in ALS, since wt non-neuronal cells delay MN degeneration and significantly extend survival of SOD1 mice (Clement et al., 2003), thus contributing to the comprehension of the MN disease pathogenesis. These data were recently confirmed by the use of mice carrying a deletable mutant SOD1 gene. Diminishing the mutant SOD1 levels in microglia cells slowed disease progression, confirming the possibility of non-cell-autonomous death of MNs (Boillee et al., 2006).

Stem-cell transplantation is a potential therapeutic strategy not only via cell replacement, but also by modification of the extracellular motor neuronal environment, through a trophic and neuroprotective effect, as well as immunomodulation strategies.

A variety of cell sources have been considered for cell therapy including neural stem cells (NSCs) (Lindavall et al., 2004), ES-derived neurons (Trounson, 2006) and somatic stem cells such as haematopoietic stem cells (Corti et al., 2004a). Transplantation of haematopoietic cells from murine bone marrow (Corti et al., 2004b) or from human cord blood cells (Ende et al, 2000; Garbuzova-Davis et al., 2003) have been used in SOD1 mice increasing their life span, probably through a neuroprotective effect rather than neurogenesis.

Recent research has pointed out the possible use of ES as a source of MNs. ES of both murine and human origin can differentiate into MNs, establish functional neuromuscular junctions with myotubes and acquire physiological properties typical of MNs when exposed to Sonic Hedgehog (Shh) and Retinoic Acid (RA) (Wichterle et al., 2002; Miles et al., 2004; Li et al., 2005).

ES-derived MNs transplanted in the spinal cords of adult rats with viral-induced MN injury survive and generate MNs in the anterior horns with ∼2–3% donor-derived motor axons within the ventral roots of each animal (Harper et al., 2004).

A variety of neuronal precursor/stem cells (primary cell culture, immortalized cell lines, genetically modified cells) have been used as cell sources for transplantation in rodent models of spinal cord diseases (Lindvall et al., 2004; Klein and Svendsen, 2005). In these studies the NSCs were reported to mainly differentiate into glial phenotypes after grafting into non-neurogenic regions of uninjured adult central nervous system (CNS) (Taupin and Gage, 2002; Cao et al., 2002) or injured spinal cord (Cao et al., 2001; Vroemen et al., 2003; Hofstetter et al., 2005). However, human CNS stem cells grown as neurospheres may differentiate quite well in neurons when transplanted in injured rat spinal cord and improve their function, suggesting that the origin of cells and the experimental protocol may influence the fate of transplanted cells (Cummings et al., 2005).

Furthermore, the in vitro commitment of NSCs with a pre-differentiation treatment may partially overcome the glial environmental cues in the spinal cord giving rise to neurons (Wu et al., 2002) and also to cholinergic neurons capable of connecting to the periphery at least in a specific model (Gao et al., 2005). In addition to human NSCs, neurospheres deriving from spinal cord or from olfactory bulb after exposure to a cocktail of growth factors or after forced expression of transcription factors like HB9 and Ngn2 can also express some motor neuronal features, even if a variable degree of MN commitment has been achieved in different experiments (Shihabuddin et al., 1997; Liu and Martin, 2004; Brejot et al., 2006; Corti et al., 2006; Zhang et al., 2006).

To date, limited experience with neural stem or precursor cells has been acquired in the ALS animal model. Human neural progenitor cells (hNPC), isolated from the foetal cortex and modified using lentivirus to secrete glial cell line-derived neurotrophic factor (GDNF), produce some beneficial effects on the course of MN disease when transplanted in the spinal cord of SOD1-G93A animals (Klein et al., 2005). The transplantation of human embryonic spinal cord stem cells resulting in a later disease onset and slower progression in SOD1 mice has recently been described (Yan et al., 2006).

We have previously isolated a primitive neural stem cell subset, double positive for the cell surface marker Lewis X (LeX) and the chemokine receptor CXCR4 (Le+CX+), that possesses CNS homing potential and extensive neuronal repopulating capacity. Le+CX+ cells are self-renewing and multipotent. In vivo, Le+CX+ cells display widespread incorporation and differentiate into complex neuronal phenotypes such as cortical and hippocampal pyramidal neurons (Corti et al., 2005).

In the present study we investigate the potential of the Le+CX+ neural stem cell population to modify disease progression and generate new neurons in the SOD1-G93A mice.

Material and methods

Animal models

Transgenic mice of the strain B6.Cg-Tg(SOD1-G93A)1Gur/J, which carries a high copy number of the mutant allele human SOD1 containing the Gly93 ->Ala (G93A) substitute, were used. Progeny for experimental analysis was obtained from breeding pairs between SOD1-G93A transgenic and C57BL/6 wt mice. Transgenic mice were identified by PCR using a described method (Gurney et al., 1994).

Transgenic TgN(ACTbEGFP)1Osb mice expressing an enhanced GFP cDNA under the control of a chicken β-actin promoter and cytomegalovirus enhancer were used as NSC donors. Their transgene is widely expressed in all tissues, with the exception of erythrocytes and hair (Okabe et al., 1997). The mHB9-GFP1b transgenic mice, expressing eGFP DNA under the control of the mouse HB9 promoter, in the cell bodies of spinal MNs in E9.5-P10 mice, allowing the immediate detection of the acquired motor neuronal phenotype (Wichterle et al., 2002) were also used.

All transgenic animals were purchased from Jackson Laboratory (Bar Harbor, ME, http://www.jax.org). All animal experiments were performed according to institutional guidelines in compliance with national (Italian Law no. 116 published at the Gazzetta Ufficiale, suppl 40, February 18, 1992, and Law no. 8, Gazzetta Ufficiale, July 14, 1994) and international legislation (EEC Council Directive 86/609, OJ L358, 1 December 12, 1987; Guide for the Care and Use of Laboratory Animals, U.S. National Research Council, 1996).

NSC isolation

NSCs were isolated from the adult brain of C57BL/6 mice (6–8 weeks), β-actin-GFP and HB9-GFP mice dissecting the subventricular zone (n = 10 mice for preparation), as described previously (Doetsch et al., 1999; Tropepe et al., 1999).

Single cell suspension was seeded at a density of 100 000 cells/ml in neurobasal medium (GIBCO Invitrogen, CA), containing B-27, N2 (Invitrogen), EGF (20 ng/ml), FGF (20 ng/ml; Sigma-Aldrich), penicillin (100 U)/streptomycin (100 µg/ml; Invitrogen). Cells were grown in uncoated T75 plastic flasks (NUNC), as free-floating clusters (neurospheres). The cultures were passaged every 5–7 days as described previously (Tropepe et al., 1999). Cells used for separation were passaged three to five times.

Cell isolation

For cell isolation, single cell suspension from neurospheres was labelled with LeX (SSEA-1) microbeads (Milteniy Biotec, Germany) and purified by immunomagnetic selection according to the manufacturer's instructions. Positive cells were then selected for double positivity for LeX and CXCR4 by FACS. Cells were labelled with FITC conjugated anti-LeX antibody (mouse, called also anti SSEA1, 1:200; BD) and R-phycoerythrin (R-PE) conjugated CXCR4 antibody (rat, 1:100; BD) for 30 min at 4°C and then rinsed twice with serum-free media by centrifugation at 4°C. In the case of GFP cells, we used unlabelled anti-LeX (1:200; BD) for 30 min followed by Cy5 goat Anti-Mouse Ig (Jackson Immunoresearch; 20 min) and CXCR4 (R-PE).

Isotype-matched mouse immunoglobulin (BD) and cells labelled only with the secondary antibody served as controls. Flow cytometric sorting was conducted using a FACS Vantage SE (BD). Sorting of antibody-labelled cells was performed using FACS gates set with unlabelled cells. LeCX positive and negative cells were collected separately in plating medium. Twenty independent FACS analyses were performed.

Cell culture and differentiation of Le+CX+ cells

For cell expansion and clonal culture, Le+CX+ sorted cells were plated in the growth medium described earlier. For in vitro priming (Wu et al., 2002), neurospheres were cultured in Neurobasal plus N2, 0.1 mM 2-mercaptoethanol, 20 ng/ml bFGF, 1 µg/ml laminin, 5 µg/ml heparin, 10 ng/ml neural growth factor (NGF) (Invitrogen), 10 µM forskolin, 0–1000 nM Shh (R&D Systems, Minneapolis, MN) and RA (1 µM) for 48 h or 5 days. This protocol was an optimization of previously published protocol to differentiate NSC into cholinergic neurons: Neurobasal plus N2, 20 ng/ml bFGF, 1 µg/ml laminin, 5 µg/ml heparin, 10 ng/ml NGF (Invitrogen) and 10 ng/ml Shh (R&D Systems, Minneapolis, MN) (Corti et al., 2005, 2006). Improved long-term survival of neurons could be obtained by the addition of 2-mercaptoethanol (Grill et al., 1993; Ishii et al, 1993).

Forskolin, a molecule that increases intracellular cAMP, can stimulate axonal elongation (Roisen et al., 1972). RA and Shh at different concentrations were previously used to promote ES differentiation into MNs (Wichterle et al., 2002; Harper et al., 2004; Li et al., 2005).

After priming, brain-derived neurotrophic factor (BDNF), GDNF, ciliary neurotrophic factor (CNTF), IGF1 and Neurotrophin 3 (NT3) (10 ng/ml, R&D Systems, Minneapolis, MN) were added to the medium. GDNF, BDNF, CNTF and NT3 (10 ng/ml) were previously used by Wichterle et al. (2002) to promote terminal motor neuronal differentiation of murine ES.

IGF1 was used in association with BDNF, GDNF and IGF1 (10 ng/ml; PeproTech Inc.) to differentiate human ES in motor neurons (Li et al., 2005).

For co-culture with myoblasts, Le+CX+ cells primed into MNs were seeded on C2C12 myoblasts (American Type Culture Collection) induced to differentiate in Muscle differentiation medium (Promocell, Heidelberg, Germany).

Immunocytochemistry on cell culture

Cultured cells were fixed in 4% paraformaldehyde (PFA, 10 min) at room temperature. After rinsing with PBS, and pre-incubation in a mixture of 5% normal serum and 0.25% Triton X-100 in PBS, the cultures were incubated with the primary antibodies (see later) overnight at 4°C. The following proteins were evaluated: nestin (mouse monoclonal, 1:200; Chemicon, Temecula, CA), vimentin (mouse monoclonal, 1:200; Chemicon), Sox2 (rabbit, 1:200; Chemicon), Musashi (rabbit, 1:200; Chemicon), Sox1 (rabbit polyclonal, 1:200; Chemicon), Pax 6 (mouse monoclonal, 1:200; Chemicon), Pax7 (goat polyclonal, 1:100; Santa Cruz, CA), En1 (rabbit, polyclonal, 1:200; Chemicon), otx2 (rabbit, polyclonal 1:200; Chemicon), Irx3 (rabbit polyclonal 1:100; Santa Cruz), Olig2 (rabbit polyclonal 1:500; Chemicon), Nkx2.2 (rabbit polyclonal 1:200; Chemicon), Nkx6.1 (goat polyclonal 1:100; Santa Cruz), HOXC8 (mouse 1:200; Covance), HOXC6 (goat polyclonal 1:100; Santa Cruz), beta III-tubulin (TuJ-1) (mouse monoclonal, 1:200; Chemicon), NF M and H phosphorylated (mouse monoclonal, 1:200; Chemicon), NeuN (mouse monoclonal, 1:100; Chemicon), mouse monoclonal anti- MAP2 (1:100 dilution; Sigma-Aldrich, MO), rabbit anti-ChAT (1:100; Chemicon), rabbit anti-Islet-1 (1:200; Chemicon), rabbit 27 anti-HB9 (1:200; Chemicon), mouse Cy3 conjugated GFAP (1:400 dilution; Sigma-Aldrich), oligodendrocyte marker (O4) (mouse monoclonal, 1:100; Chemicon) and Alexa 488 rabbit polyclonal antibodies recognizing GFP (1:400; Molecular Probes, Eugene, OR); rhodamine-conjugated bungarotoxin was purchased from Molecular Probes (1:1000).

After repeated rinses in PBS, the primary unconjugated antibodies were further incubated with FITC and RPE or TRITC conjugated secondary antibodies (1:100; DAKO, Carpinteria, CA) (1 h, dark, room temperature) in PBS, then rinsed in PBS and coverslipped. Controls with omission of primary antibodies were made, with no detection of positive signals. For quantitative analyses of Le+CX+ phenotypes after in vitro differentiation, 10 fields were randomly chosen for each sample. The percentage of any given phenotype in a sample was obtained by averaging proportions of a specific cell type in each of the 10 fields. At least four samples were counted for each treatment group.

Real-time RT-PCR

For the evaluation of neuronal transcripts, total RNA was extracted from Le+CX+ cells cultured in neuronal growth and differentiation medium. 3T3 cells were used as negative controls.

Total RNA was prepared using the EUROzol isolation kit (EuroClone Ltd, UK). RT was carried out as follows: two micrograms of total RNA from each sample were reverse transcribed using random primers by the Ready-To-GoTM ‘You-Prime First-Strand Beads’ kit (Amersham Biosciences, Piscataway, NJ), according to the manufacturer's recommendations. One microlitre of the generated cDNA was used as template for each reaction in real-time quantitative PCR analysis, using Assays-on-Demand Gene Expression Products in an ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Foster City, CA).

The following gene expression assays were performed: OLIG2: Mm01210556_m1; HOXC6: Mm01307713_m1; NKX6-1: Mm00454962_m1; SOX1: Mm00486299_s1; HOXC8: Mm00439369_m1; PAX6: Mm00443072_m1; nestin: Mm00450205_m1; TuJ1: Mm00727586_s1; GFAP: Mm00546086_m1.

Eucaryotic 18S ribosomal RNA (Hs99999901_s1) was used as endogenous control to normalize all the above target genes; to determine relative quantification, the comparative threshold (Ct) cycle method was used. The fold of change in gene expression profile was referred to unseparated embryonic neurospheres. All Assays-on-Demand Gene Expression Products consist of target assays to amplify and detect expression of specific RNA sequences.

Preparation of cells for grafting

Before cell transplantation, isolated Le+CX+ cells were cultured (for priming) in Neurobasal plus N2, 0.1 mM 2-mercaptoethanol, 20 ng/ml bFGF, 1 µg/ml laminin, 5 µg/ml heparin, 10 ng/ml NGF (Invitrogen), 10 µM forskolin, 1000 nM Shh (R&D Systems, Minneapolis, MN) and RA (1 µM) for 5 days. On the day of transplant, the primed cells were harvested with trypsin (0.05% trypsin/EDTA) and centrifuged in DMEM/FBS (1000 rpm/7 min). Following a wash with sterile phosphate-buffered saline (PBS), the cells were counted using Trypan Blue to evaluate viability (Trypan Blue exclusion assay), centrifuged and resuspended in sterile saline solution. The cell suspension was placed on ice.

Le+-CX+ NSC transplantation

Seventy-day-old transgenic SOD1-G93A mice were transplanted with Le+CX+ NSCs (20 000 cells) or only with vehicle (saline solution). The mice were anesthetized and then an incision was made in the skin overlying the lumbar enlargement of the spinal cord (Garbuzova-Davis et al., 2002). The muscle and connective tissue were dissected, through the lamina of the L1 vertebra (corresponding to the L4–L5 segment of the spinal cord), the dura and spinal cord were exposed, and a laminectomy was performed. A glass injection micropipette (100 µm tip diameter) was bilaterally inserted into the anterior horns (1–2 mm below the dura in depth). A slow infusion of 2 µl of cell suspension (1 × 104 cells per horn) was injected into the spinal cord over 3–5 min for horn. After injection the needle was removed and the surgical site sutured. Control animals received 2 µl of saline solution injections per horn in a similar fashion as the cell-transplanted animals. SOD1-G93A mice were divided into cell-treated (n = 24: n = 12 males, n = 12 females) and non-treated (n = 24: n = 12 males, n = 12 females) groups that were evaluated up to the end-stage for rotarod analysis, survival record and histological evaluation of donor cell phenotype. The study was designed so that littermates were distributed equally into the transplanted and untransplanted groups.

For MN and axon count, three groups of animals (each composed of a total of 12 mice: transplanted SOD1, untransplanted SOD1 mice and wt) were sacrificed at 110 days of age and were analysed respectively for histological quantification (for each group: n = 6) and for growth factors evaluation with ELISA (for each group: n = 6).

Assessment of motor function and survival

Media- and cell-injected SOD1-G93A mice were monitored daily following transplantation for clinical signs of disease onset. The motor performance of all mice was assessed weekly by using an accelerating rotarod device (4–40 r.p.m. Rota-Rod 7650; Ugo Basile, Comerio, Italy), starting from the week before transplantation. The time during which mice remained on the rotarod was registered. Each mouse performed three consecutive trials and the longest latency to fall was recorded. The investigators performing the behavioural assessment were blind to the treatment.

The rotarod test was analysed by analysis of variance (ANOVA) followed by a Tukey post hoc analysis for multiple comparison. The disease onset was defined as the time at which a 10% decline in the mouse rotarod performance is observed. Log-rank test were used for onset comparisons. The disease progression was defined by the interval between the onset and end-stage and was analysed by ANOVA. Mortality was scored as the age at which the mouse was unable to right itself within 30 s when placed on its back in a supine position as previously described (Li et al., 2000). Kaplan–Meier survival analysis and log-rank test were used for survival comparisons.

Tissue analysis

The animals were sacrificed, perfused and fixed with 4% PFA in PBS (pH 7.4). The spinal cord was isolated, immersed in PFA solution for 1 h, then in sucrose 20% solution in PBS (pH 7.4) overnight and frozen in Tissue Tek OCT compound (Sakura), with liquid nitrogen. The tissues were cryosectioned and mounted on gelatinized glass slides. Twenty micrometres were collected every 10th section. All sections were blocked with 1% FCS in PBS and permeabilized with 0.25% Triton X-100.

Sections were processed for multiple markers to determine the cellular phenotype of GFP-labelled cells. Primary antibodies were added overnight at 4°C at dilutions of 1:200 for Neu-N (mouse monoclonal antibody, Chemicon), 1:200 for NF (mouse monoclonal antibody, Chemicon), 1:200 for TuJ1 (mouse monoclonal antibody, Chemicon), 1:200 for MAP-2 (mouse monoclonal antibody, Sigma-Aldrich), 1:200 for nestin (mouse monoclonal, Chemicon), 1:200 for vimentin (mouse monoclonal, Chemicon), 1:100 for rabbit anti-ChAT (Chemicon), rabbit anti-Islet-1 (1:200; Chemicon), rabbit anti-HB9 (1:200; Chemicon), 1:500 for rabbit anti-tyrosine hydroxylase (TH, Chemicon), 1:100 for anti glutamate decarboxylase 67 (GAD 67) (mouse monoclonal, Chemicon), 1:100 goat-Doublecortin 30 (DCX C-18, Santa Cruz), 1:200 for O4 (mouse monoclonal, Chemicon), 1:200 for GFAP (mouse monoclonal Cy3 conjugate, Sigma), 1:500 IGF-1Rβ (rabbit, Santa Cruz, CA), 1:200 IGFBP5 (rabbit, Santa Cruz).

Primary antibodies identifying protein on blood immune cells were used to address the type and intensity of host-versus-graft cellular response: CD4 and CD8, (rat anti mouse, 1:200, BD); anti CD49b [natural killer cells (NK), rat anti mouse 1:100, BD], MAC-2 (activated microglia, rat anti-mouse 1:100, Cedarlane).

For secondary antibodies: donkey, goat, mouse, rabbit or rat conjugated with FITC, R-PE, CY3 or biotin, 1:200 (Jackson Immunoresearch and DAKO) were used for 1 h at room temperature, when unconjugated primary antibody was used.

Anti-GFP antibody rabbit serum Alexa 488 (1:400 dilution; Molecular Probes) was used to reveal GFP positivity in double immunostaining.

Co-expression of GFP/tissue specific markers was evaluated by conventional fluorescence microscope (Zeiss Axiophot, Germany) and by laser confocal scanning (Leica TCS SP2 AOBS, Germany) microscopic analysis.

To obtain an unbiased stereological estimation of GFP positive cells, optical dissectors and random sampling were used. For donor cells quantification, a systematic random series of every 10th coronal section (20 µm) was obtained throughout the entire spinal cord (a mean of 25 sections for animal).

Numerical density of neurons was then estimated using the optical dissector method (Gundersen et al., 1988; Messina et al., 2000). Optical dissectors sized 100 × 70 × 14 µm were randomly sampled and the number of positive cells in each dissector was quantified. The density was calculated dividing the total number of GFP cells by the total volume of optical dissectors. The total volume of tissue per specimen (Vcord) containing labelled neurons was calculated using the Cavalieri method. This total volume of tissue, multiplied by the number of neurons per µm3, gave the total number of neurons per specimen (N = NvxVcord) (Messina et al., 2000). We evaluated the number of donor-derived motor neurons based on the positivity for ChAT and GFP immunoreactivity, the position in the lateroventral horns and the cell dimension (>25 µm) for every section (L1/L6) (Grondard et al., 2005).

MN and axon count

Paraffin serial sections (12.5 µm) of lumbar spinal cord were processed for Nissl-staining quantification of MN numbers on the light microscope as described (Jablonka et al., 2000).

For axon count, the tissue was dissected, immersed in 2.5% glutaraldehyde overnight and then postfixed in 2% osmium tetroxide. Samples were then dehydrated in ethanol and embedded in Epon. Semi-thin transverse sections (1 µm) were stained with toluidine blue. L4 roots were examined for axon counting on the optic microscope.

Enzyme-linked immunosorbent assay (ELISA) in vitro and in vivo

For in vitro studies, IGF1 and VEGF levels were measured in cell culture supernatants of Le+CX+ cells (six independent experiments for each cytokine) with a commercially available enzyme-linked immunosorbent assay (ELISA) system (R&D Systems) according to procedures recommended by the manufacturer. ELISA kits for VEGF and IGF1 were used to detect the cytokines in the lumbar spinal cord (n = 18: n = 6 transplanted SOD1-G93A, n = 6 untransplanted SOD1-G93A, n = 6 wild-type mice) according to the manufacturer's directions for ELISA assay. The data were analysed with Student's t-test.


Isolation and characterization of CNS-derived LewisX+/CXCR4+ (Le+CX+) cells

We set out to determine whether cells with MN features could be obtained from a NSC population expressing the cell surface marker Lewis X (LeX) and the chemokine receptor CXCR4.

We isolated Le+CX+ cells from adult murine neurospheres by magnetic cell sorting for LeX followed by double fluorescence activated cell sorting (FACS) for LeX and CXCR4 (Supplemnetary Fig. 1A and D–F). Positive fractions ranged from 70 to 98% of purity (Supplementary Fig. 1B and C).

As we previously described, Le+CX+ cells share the property of stem cells and are self-renewing and multipotent (Corti et al., 2005). The frequency of stem cells, tested as the ability to generate neurospheres, is 1:7 Le+CX+ and 1:570 in Le −CX− in adult ones (P < 0.001). These cells expressed NSC antigens like Sox2 (86.4 ± 6.6%), Musashi (54.8 ± 4.6%) and nestin (92.2 ± 4.5%) (Supplementary Fig. 1G–I). Newly generated neurospheres from Le+CX+ cells can be dissociated to single cell that gives rise to new spheres.

The multipotency of the expanded neurospheres derived from Le+CX+ cells was confirmed after in vitro differentiation, resulting in cells expressing markers for neurons (Supplementary Fig. 1J), astroglial cells (Supplementary Fig. 1K) or oligodendrocytes (Supplementary Fig. 1L). The neurons, expressing the microtubule associated protein 2 (MAP2) represented 38.4 ± 6.4% of the total cell number while 24.2 ± 6.4% cells were glial fibrillary acid positive (GFAP), and 4.2 ± 1.4% of cells expressed O4 antigen, indicating an astrocytic and oligodendroglial phenotype, respectively.

Differentiation of CNS-derived Le+CX+ cells into MN-like cells

We have previously described that Le+CX+ cells can be induced after the exposure to growth factors and morphogens to acquire a cholinergic and pyramidal cell phenotype in vivo (Corti et al., 2005); we wondered whether these cells may be induced to acquire a MN phenotype as well.

Le+CX+ cells were grown in the presence of RA with Shh, associated to laminin, heparin, forskolin b-FGF and NGF (priming) (Wu et al., 2002; Corti et al., 2005, 2006). After 5 days, the cells were induced to differentiate in the presence of 2% FBS in DMEM/F12, supplemented with N2 and the following neurotrophic factors, BDNF, GDNF, CNTF, IGF1 and NT3, which have been demonstrated to increase MN differentiation and survival (Wichterle et al, 2002; Li et al., 2005).

We analysed the phenotype of RA–Shh exposed Le+CX+ with respect to the positional identity along the rostrocaudal axis (Fig. 1A). Shh's concentration ranged from 0 to 1000 nM. The data shown in the results and figures refer to a concentration of 1000 nM.

Fig. 1

Acquisition of a caudal spinal cord phenotype by Le+CX+ cells exposed to RA and Shh. (A) Schematic diagram showing the caudalization pattern promoted by the RA and the ventralization process stimulated by the Shh signalling that determine the MN fate. (B–E) At the beginning of the differentiation, the cells expressed high level of Sox1+, indicative of a primitive NSC phenotype (B), then they acquired a rostral progenitor phenotype demonstrated by the expression of OTX2 (C) excluding HoxC6 (D) and HB9 (E). (F–I) After differentiation, these rostral neural progenitors acquired a spinal positional identity in response to RA and expressed caudal markers like Hoxc6 (H). Subsequently, spinal progenitor cells showed an MN identity in response to the ventralizing action of Shh and expressed HB9 (I). These cells were negative for Sox1 (F) and Otx2 (G). (J–Q) Initially the cells expressed Otx2 and En1 (J–K), that is indicative of an early midbrain identity while afterwards they expressed Hoxc8 (L) and HB9 (M), which are markers of the spinal cord phenotype as demonstrated by immunocytochemistry and real-time RT-PCR (O–Q). (N) Quantitative analysis of transcription factor expression at T0 and after 10 days of differentiation (T10) expressed as percentage of positive cells. Scale bar: B, C, F, G: 80 µm; D, E, H, I: 150 µm; J–M: 150 µm.

At the beginning of the differentiation process, many Le+CX+ cells showed a primitive neuroectodermal phenotype by the expression of SOX1, followed by the acquisition of a rostral phenotype as demonstrated by the expression of Otx2 and En1 antigens, which are associated with early midbrain identity (Davis and Joyner, 1988; Mallamaci et al., 1996) and the contemporaneous exclusion of Hoxc6 and Hoxc8, markers of spinal cord identity (Belting et al., 1998; Liu et al., 2001) (Fig. 1B–E, J and K).

Le+CX+ cells, after exposure to RA and Shh, progressively downregulated Otx2 and En1 and expressed Hoxc6, and Hoxc8, as demonstrated by immunocytochemistry and real-time RT-PCR, suggesting the acquisition of a spinal cord phenotype (Fig. 1F–I, L–Q).

To determine whether the Le+CX+ positive cells can differentiate into MN-like cells, we analysed the expression of homeodomain (HD) and basic helix-loop-helix (bHLH) transcription factors that characterize different subsets of neural progenitor cells positioned along the dorsoventral and mediolateral axes of the developing spinal cord (Fig. 2A and B) (Briscoe and Ericson, 2001).

Fig. 2

Ventralization and lateralization of LeCX cells. (A–B) Expression of transcription factors along dorsalventral (A) and mediolateral (B) axis of developing spinal cord. (C) Quantitative analysis of HD and bHLH proteins expression at T0 (white bars) and after 10 days of differentiation (T10, black bars) expressed as percentage of positive cells. (D–E): Shh-activated transcriptional pathway of spinal MNs-like cells that expressed low levels of Pax6 (D) and upregulate Nkx6.1 (E), as demonstrated by real-time RT-PCR. (F–K) At the beginning of the differentiation, Le+CX+ cells were positive for PAX7, PAX6 and Irx3 while they were negative for Nkx6.1 and Olig2 and low positive for Nkx2.2. (L–Q) Differentiated spinal MN progenitors derived from Le+CX+ decreased the expression of Pax6, exclude Pax7 and Irx3, upregulated Nkx6.1 and Olig2 and, at lower degree, Nkx2.2. Scale bars: FH: 100 µm; I–Q: 150 µm.

In embryonic development, spinal MN progenitors acquire a progressive ventrolateral position and express Pax6, Nkx6.1 and Olig2 while they downregulate Pax7, Irx3 and Nkx2.2.

Shh exposure activated transcriptional pathway of spinal MNs-like cells—derived from Le+CX+ fraction—that expressed Olig2, Nkx6.1 and discrete levels of Pax6, as demonstrated by immunocytochemistry and real-time RT-PCR (Fig. 2C–Q).

To directly identify and monitor Le+CX+ cell-derived MNs, we used as cell donors the transgenic mice that express GFP under the control of HB9 promoter only in MNs (Supplementary Fig. 2).

HB9-GFP+ MNs expressed high levels of GFP in the cell bodies, in the axons and dendrites in a parallel manner to endogenous HB9 (Supplementary Fig. 2A). Furthermore they co-expressed Islet-1 (Supplementary Fig. 2B).

The number of HB9-GFP MNs was increased upon exposure to Shh in a concentration-dependent manner. GFP/HB9+ MNs represented 22.5 ± 4.6% of the total number of cells exposed to 1000 nM Shh (Supplementary Fig. 2C).

Then we evaluated whether HB9-GFP-derived Le+CX+ cells presented features that are specific of bona fide MNs.

Their morphological aspects were similar to MNs, presenting a larger cell soma and elaborate neuritic extensions and they expressed choline acetyltransferase (ChAT), suggesting their cholinergic neuronal phenotype (Supplementary Fig. 2D and E).

To address the potential of MN-committed Le+CX+ cells to form neuromuscular junctions, we established a co-culture with skeletal muscle-differentiated murine myoblasts. Le+CX+-derived neurons extended long axons that connect with myotubes. The proportion of HB9 + cells that developed neuromuscular junctions in our co-culture assay was 15.6 ± 3.7%.

Staining of these co-cultures with α-bungarotoxin, which binds to acetylcholine receptors (AChR), demonstrated AChR clustered at MN axon–muscle cell junctions, suggesting that Le+CX+-derived MNs are able to establish neuromuscular synapses (Supplementary Fig. 2F and G).

Le+CX+ cells transplantation delays disease progression and increases survival in SOD1-G93A mice

To examine the ability of Le+CX+ cells to differentiate appropriately in neurons within the spinal cord and modify ALS phenotype, we transplanted 20 000 Le+CX+ cells into the spinal cord of SOD1-G93A mice at an early-symptomatic disease phase (70 days). We grafted cells that were ‘primed’ for 5 days in culture, as described for in vitro experiments, to promote the acquisition of MN differentiation. Preliminary experiments regarding the use of different cell concentrations are provided as Supplementary Table 1. We used as cell source transgenic mice expressing the gene reporter GFP into all tissues (Okabe et al., 1997) and HB9-GFP mice expressing GFP only in MNs (Wichterle et al., 2002).

Untransplanted transgenic SOD1-G93A, littermates of treated SOD1 mice, that received only the vehicle were used as control. The study was designed so that siblings were distributed equally in the treated and control groups and were equally divided between male and female.

Transplantation of Le + CX + cells led to a significant increase of survival from 142.83 ± 8.67 (non-treated mice n = 24) to 165.75 ± 9.15 days in Le+CX+-treated mice (n = 24; P < 0.0001; χ2 = 45.06) in SOD1-G93A transplanted mice, compared to vehicle-treated group.

GFP-Le+CX+-transplanted mice (n = 12) and HB9-GFP-Le+CX+-mice (n = 12) showed a prolonged mean survival time of 22–23 days compared to untreated SOD1 littermates (GFP—group mean 165 ± 9.5 versus SOD1-G93A-untransplanted mice 142.9 ± 8.25 days; P < 0.0001; χ2 = 20.61; HB9-group mean 166.5 ± 9.08 versus SOD1-G93A-untransplanted mice 142.75 ± 9.44; P < 0.0001; χ2 = 22.71) (Fig. 3B).

Fig. 3

Transplantation of Le+CX+ cells modifies disease progression of SOD1 mice. (A) To evaluate the effect of NSC transplantation on the SOD1 disease we performed neuromuscular evaluation. Transplantation of NSC improved significantly motor performance in SOD1 mice as demonstrated by rotarod test. Transplanted SOD1-G93A mice with β-actin-GFP Le+CX+ cells: blue line; transplanted SOD1-G93A mice with HB9-GFP Le+CX+ cells: pink line; orange and light blue lines represent their relative untreated controls. (B) Survival was significantly (P < 0.0001) extended for transplanted mice compared to untreated mice by 23 days as demonstrated by the Kaplan–Meier curve (B). Transplanted SOD1-G93A mice with β-actin-GFP Le+CX+ cells: blue line; transplanted SOD1-G93A mice with HB9-GFP Le+CX+ cells: pink line; orange and light blue lines represent their relative untreated controls. (C–H) Spinal cord and ventral root analysis. (C–E) Photomicrographs show Nissl stained sections through the lumbar spinal cord (paraffin serial sections) from wild-type mice (C) SOD1-treated (D) and untreated (E) mice at 110 days of age. (F–H) Light microscope images of L4 ventral root semithin sections stained with toluidine blue from wild-type mice (F), transplanted SOD1 mice (G) and untreated SOD1 mice (H) at 110 days of age. (I) Motor neuronal counting (n = 6 for each group) in the lumbar spinal cord of treated and untreated SOD1 mice and wild-type (data represent mean ± SD of motor neuron number per section), at 110 days of age. The number of large neurons in the ventral part of lumbar spinal cord sections was determined by stereological analysis. Significance was evaluated by using ANOVA. The evaluation revealed significantly increased numbers of surviving neurons in G93A mice treated mice (P < 0.001). (J) Quantification of axons in wild-type, mutant and transplanted mice (data represent mean + SD) at 110 days of age (n = 6 for each group). Data represent mean ± SD of axons per section. Significance was evaluated by using ANOVA. Scale bars: C–E: 300 µm; F–H: 50 µm.

The longest life span in each condition was 155 days for untreated SOD1-G93A mice, 182 days for transgenic animals transplanted with GFP-Le+CX+ cells and 183 days for transgenic animals transplanted with HB9-GFP-Le+CX+ cells.

Neuromuscular function was evaluated on transplanted (n = 24) and untreated (n = 24) animals by rotarod test. Between 112 and 119 days of age, untreated animals showed a marked decrease in performance, whereas the treated animals presented the greatest deficits 21 days later (Fig. 3A). Thus the decline of motor performance (onset) was significantly delayed by 21 days in treated mice compared to untreated mice (P < 0.0001). Instead, the mean progression time (from onset to the end disease stage) ranged between 32 and 33 days in transplanted mice which results not significantly different from untreated mice (mean progression 30 days).

Histological analysis of the transplanted SOD1-G93A spinal cord

At the end-stage of the disease the animals were sacrificed and histological analysis was performed to examine the engraftment and differentiative fate of the transplanted Le+CX+ cells in the SOD1-G93A spinal cord. All transplanted mice showed the presence of GFP-donor cells in the anterior horns of lumbar spinal cord (Fig. 4). GFP-Le+CX+ cells were observed at least 5 mm away from the implantation site in both the rostral and caudal directions.

Fig. 4

Le+CX+ cells transplanted in SOD1 spinal cord survive and generate MN-like cells. For in vivo experiments, we transplanted 20 000 Le+CX+ positive cells, after an in vivo induction in MNs into the spinal cord of SOD1 mice. To trace the fate of transplanted cells we used LeX + CX+ cells deriving from β-actin GFP transgenic mice that express the gene reporter in all cells and HB9-GFP mice that express GFP only in MNs. (A–B) After transplantation of β-actin-GFP Le+CX+ cells into SOD1 mice, we demonstrated the presence of integrated GFP donor cells in the gray matter of the spinal cord (A–B). (C–D) GFP+ cells (green signal), deriving from HB9-GFP donors, were detected in the anterior horn of the spinal cord, demonstrating the acquisition of MN phenotype. Scale bars: A: 300 µm; B, C, D: 150 µm.

Unbiased stereological quantification with optical dissectors and random sampling demonstrated that the average total β-actin-GFP cells was 5,365 ± 76, while the number of HB9-GFP cells—expressing a putative MN phenotype—was 1236 ± 44.

We examined the acquired phenotype of GFP-Le+CX+ cells performing immunohistochemical analysis for neuroectodermal markers followed by confocal microscopy on the spinal cord sections of transplanted animals (Fig. 5). Transplanted cells differentiated in a large proportion in neurons as demonstrated by the positive staining for several neuronal specific antigens including Neurofilament (NF), MAP2 and Nuclear Neural-specific Antigen (NeuN). MAP-2 positive β-actin-GFP neurons accounted for 44.6 ± 4.6% of all GFP cells, while all HB9 positive cells expressed neuronal antigens.

Fig. 5

Phenotype of donor cells evaluated by immunohistochemistry for neuroectodermal antigens. (A–I) Confocal microscopy showed that GFP donor-derived neurons (green: A, D, G) co-expressed neuronal-specific proteins such as NF (B), MAP2 (E) and NeuN (H) (C, F, I merge). (J–L) A fraction of GFP neurons presented motor neuronal features as demonstrated by double immunofluorescence staining of GFP under the HB9 promoter (green, J) and cholinergic neurotransmitter (K) (L: merge). (M–O) HB9-GFP axons (green signal, M) of transplanted SOD1 mice were detected within ventral roots, labelled with NF (red signal, N), (merge yellow signal, O), suggesting some ability of donor cells to extend their axons appropriately. Scale bars: A–I: 150 µm; J–L: 100 µm; M–O: 150 µm.

Immunohistochemical analysis for glial antigens (GFAP and O4) in β-actin-GFP cells showed the presence of 26.2 ± 5.2% astrocytes and 4.4 ± 1.3% oligodendrocytes.

We also examined whether Le+CX+ cells presented some specific motor neuronal features in addition to the expression of the MN specific HB9-GFP. Immunohistological analysis revealed that 20.4 ± 6.4% of β-actin GFP-cells and 76.6 ± 7.8% of HB9-GFP cells expressed ChAT, demonstrating the acquisition of a cholinergic phenotype. Furthermore, these cells morphologically resembled bona fide MNs.

Regarding HB9-GFP+ cells transplantation, the stereological count processed between L1/L6 and based on ChAT and GFP immunoreactivity, the position within the anterior horns and the cell dimension (>25 µm) resulted in 3.8 ± 1.2 donor neurons (ChAT + HB9+) and 18.46 ± 2.2 endogenous MNs (ChAT + HB9+) for section. Regarding GFP neurons, we detected 3.9 ± 1.9 donor neurons (ChAT + GFP+) and 18.92 ± 2.1 endogenous MNs (ChAT + GFP+) for every section. Based on these data, donor neurons represent the 20% of total MNs.

GFP Le+CX+-derived cells extended long processes in the rostral–caudal projection and, more significantly, horizontally in the white matter and in the ventral roots (a mean of 33 ± 12 GFP positive axons per animal, 4.8 ± 1.7% of all axons in the anterior root), suggesting that they have the capacity to elongate their axons towards the periphery (Fig. 5M–O). The GFP+ axons length detected is estimated to be 3–5 mm distal to the root entry into the spinal cord.

Graft rejection was studied by immunohistochemistry with antibodies recognizing markers of blood immune cells (i.e. lymphocytes and NK cells) or activated microglia. Blood cells were detected using antibodies against CD4 and CD8 surface antigens of T cells and against a surface epitope present in NK cells, while activated microglial cells were labelled with an antibody against MAC-2.

CD4+, CD8+, NK and microglial cell recruitment to grafting sites was not observed (Supplementary Fig. 3).

Le+CX+ transplantation protects from MN loss SOD1-G93A mice

To determine whether Le+CX+ transplantation provides protection against MN loss, we evaluated the MNs in spinal cord sections and ventral spinal nerve roots from a quantitative point of view (Fig. 3C–J).

At day 110 of life, MN loss was significantly reduced by Le+CX+ transplantation, as compared to vehicle-treated SOD1-G93A mice (P < 0.001).

At the final end-stage of the disease both treated and untreated mice presented a marked MN loss.

Furthermore, transplanted animals presented a preservation of axonal density in the L4 ventral root, while untransplanted mice showed a substantial reduction in axons.

The quantitative assessment revealed that almost half of L4 motor axons were lost in untransplanted mice (P < 0.0001), while in transplanted SOD1-G93A mice this loss was significantly reduced (treated versus untreated P < 0.0001), preserving over 75% of the mean number of axons in adult L4 ventral nerve roots.

At disease end-stage, the quantitative evaluation of L4 anterior roots showed a marked loss in axon number both in untransplanted and transplanted mice.

Transplanted Le+CX+ cells produce VEGF and IGF1 in recipient SOD1-G93A spinal cord

To investigate the mechanisms accounting for the observed improvement in SOD1-G93A mice after transplantation, we evaluated whether Le+CX+ cells can produce neurotrophic factors with a putative neuroprotective role on MNs.

We determined by ELISA the production of VEGF and IGF1 in Le+CX+ cell culture before transplantation (Fig. 6A and B).

Fig. 6

Protective role of Le+CX+ of MN through VEGF and IGF1 production. (A–B) Le+CX+ cells expressed VEGF and IGF1 as demonstrated by ELISA analysis compared to unselected cells (P < 0.001). The growth factors were dosed in the supernatants. (C–D) The expression levels of VEGF and IGF1 were quantified by ELISA in SOD1 transplanted spinal cord compared to untreated SOD1 and wild-type spinal cord demonstrating that Le+CX+ transplantation increases their level (P < 0.001). (E–L) IGFBP5 and IGF1-R immunohistochemistry and quantification in SOD1 transplanted, untransplanted and wild-type spinal cord. IGFBP5 was higher in SOD1 than in controls. (E–G) IGFBP5 staining in red, nuclei are conterstained with DAPI (blue signal). (H–J) IGF1-R staining in red, nuclei are conterstained with DAPI (blue signal). NSC transplanted spinal cord presented a slight decrease of IGFBP5 expression level (K) and a significant downregulation of IGF1R (L) (P < 0.001). Scale bar: E–J: 300 µm.

The analysis showed a significantly higher concentration of VEGF and IGF1 in Le+CX+ cell culture medium compared to that detected in medium of unselected cells derived from neurospheres (P < 0.001).

On the basis of these findings we investigated the expression of IGF1 and VEGF in transplanted SOD1-G93A spinal cord compared to untreated and wild-type spinal cord by ELISA.

Analysis of spinal cord at 110 days of age indicated that Le+CX+ transplantation significantly increased the levels of VEGF and IGF1 compared to their level in vehicle-injected SOD1-G93A controls (P < 0.001) (Fig. 6C and D).

We evaluated also the numbers of MNs immuno-positive for insulin-like growth factor binding protein-5 (IGFBP5) and IGF-1Receptor β (IGF-1Rβ) in wild-type, in SOD1-G93A transplanted and in untreated mice.

The percentage of IGF-1Rβ and IGFBP5 MNs per total living MNs was increased in untreated compared to wild-type mice, in line with previous observations (Wilczak et al., 2003).

IGFPB5's upregulation preventing binding of IGF-1 to IGF-1R can promote the IGF-1R upregulation due to free IGF-1 reduction.

Transplanted spinal cord presented a slight decrease of IGFBP5 expression level and a significant downregulation of IGF1-R (P < 0.001), suggesting that free IGF1 deriving from stem cells can act on IGF1-R reducing their expression on MN surface.


Our study demonstrates that transplantation of brain-derived adult NSC subpopulation double positive for LeX and CXCR4 antigens induces the generation of MN-like cells into the spinal cord of SOD1-G93A mice, protects the endogenous MNs and ameliorates the neurological phenotype, supporting the potentiality of the NSCs as a therapeutic intervention for MN diseases.

These results extend our previous findings on the multipotentiality of LeX and CXCR4 positive brain-derived NSCs, showing that after exposure to Shh and RA they can differentiate into MN-like cells.

Several lines of evidence support this conclusion. In vitro Le+CX+ cells, when exposed to growth factors and morphogens, generate MN-like cells through a molecular pathway that recapitulates the steps of embryonic MN generation in vivo, as demonstrated by the progressive expression of transcription factors specific of caudal spinal neuronal progenitors (HOXC6 and HOXC8) and lateral spinal neuronal progenitors (Nkx6.1 and Olig2), whose expression precedes MN-specific factors such as HB9 and Islet1. Furthermore, Le+CX+ cells when derived from HB9-GFP mice, after differentiation in MNs, co-express the gene reporter appropriately together with endogenous HB9, Isl1 and cholinergic neurotransmitters. Finally, these MN-like cells possess the capacity to form neuromuscular junctions, a fundamental characteristic of bona fide MNs.

In vivo, transplanted Le+CX+ cells are capable of differentiating into MN-like cells characterized by a distinct cholinergic neurotransmitter profile and, even if to a minor extent, they can extend their axons appropriately into the ventral roots of ALS animals. The newly generated neurons appear to be well integrated into the host ventral horn environment and persist up to 2–3 months in vivo. Up to now, the follow-up period of wild-type mice has been extended to 6 months after transplantation. At this time point, we have detected surviving transplant-derived (GFP+) motor neurons within the gray matter of the host spinal cord.

The adult mammalian spinal cord is a non-neurogenic site representing a non-permissive environment for the generation of new neurons (Shihabuddin et al., 2000). Therefore, we transplanted Le+CX+ cells that were primed to a MN fate before the engraftment, overcoming the glial signals in the host spinal cord, giving rise both to neurons and cholinergic MN-like cells. It has been previously described that human NSCs, after receiving a pre-differentiation in vitro, can give rise to cholinergic neurons in the spinal cord (Wu et al., 2002) capable of connecting to the periphery at least in particular lesional models (Gao et al., 2005).

Neurospheres from spinal cord or from olfactory bulb, after exposure to specific signals or after forced expression of transcription factors like HB9 and Ngn2, can also express some motor neuronal features, even if a complete MN fate has not been obtained in some of the experiments (Shihabuddin et al., 1997; Liu and Martin, 2004; Brejot et al., 2006; Corti et al., 2006; Zhang et al., 2006). Based on our results, we surmise that the isolation of a defined NSC subpopulation combined with the optimization of a chemically defined differentiation protocol allow the acquisition of complex neuronal phenotypes like the pyramidal one (Corti et al., 2005) and motor neuronal identity.

The transplantation protocol uses 20 000 selected and primed cells, because after various preliminary experiments we determined this cell density as the more effective and well-tolerated cell concentration that could be applied to the SOD1 mice lumbar spinal cord. Previous experiments of stem-cell transplantation in motor neuron disease rat models used successfully a cell dose of 60 000 cells for direct lumbar spinal cord injection (Harper et al., 2004; Deshpande et al., 2006).

To achieve a good correspondence with L4–L5, injections were targeted to the portion of spinal cord immediately underneath the L1 vertebra. We chose to place the transplants at this level because motor neurons in this region innervate the hind limbs, which are first affected in this ALS model. The SOD1-G93A mice develop initial signs of neuromuscular deficits like tremor of the legs and loss of extension reflex of the hind paws. The same injection site has previously been used, for the same reasons, in other cell transplantation experiments in SOD1 mice (Garbuzova-Davis et al., 2002; Yan et al., 2006). A more widespread cell engraftment could possibly optimise survival results. The grafted cells act directly on lumbar spinal cord and more diffusely on the other spinal cord segments by a paracrine effect that probably becomes lower by increasing the distance from the graft.

We performed the cell transplantation at 70 days of age. There is histological evidence of MN subcellular degeneration from as early as 35 days suggesting that actual disease onset occurs earlier (Mourelatos et al., 1996). Furthermore, it has also been described that microvacuoles in the cytoplasm, with marked swelling of the mitochondria are present since 2 weeks of age (Bendotti et al., 2001). As described by other groups (Barnéoud et al., 1997; Bendotti et al., 2001), SOD1 mice develop initial neuromuscular signs and loss of extension reflex of the hind paws at 70 days of age. For these reasons we consider the 70 days as an early symptomatic phase.

Although no immunosuppression was used, the CNS tissues analysed did not exhibit any sign of immunoreaction, such as microglial activation or leucocyte infiltration. Several other stem-cell transplantations have been performed in mouse CNS without the need of an immunosuppressive regimen (Lee et al., 2005; Alvarez-Dolado et al., 2006; Hendricks et al., 2006; Li et al., 2006). Indeed, the donor and recipient mice share an identical genetic background differing only for the expression of the reporter gene and the mutant SOD1. The transplantation procedure was well tolerated from all animals. We did not observe any tumour formation or other side effects. To our knowledge, no neoplastic growth has been reported after NSC transplantation, in contrast to the possible teratoma formation described after ES transplantation.

In order to restore motor function, donor MNs need to connect with their proper muscle targets. In vitro Le+CX+ cell-derived MNs formed functional connections with muscle fibres identified by the presence of axon-induced acetylcholine receptor aggregation, while in vivo a fraction of donor MNs extended their axons through the white matter and into the ventral roots. These data are consistent with those previously observed with ES-derived MNs both from mouse and human cells (Harper et al., 2004; Li et al., 2005). Furthermore, it has been recently described that ES-derived MNs transplanted in paralysed adult rats can extend their axons towards the periphery, forming neuromuscular junctions, at least in particular experimental conditions such as in the presence of signals that overcome myelin-mediated repulsion (Deshpande et al., 2006).

Transplanted SOD1-G93A mice showed an amelioration of MN phenotype as demonstrated by neuromuscular function test (rotarod) and an increased survival. Le+CX+ transplantation performed at early symptomatic phase was significantly effective in prolonging survival compared to the untreated control group. These observations correlated with the neuropathological analysis showing a significant amelioration of MN loss at 110 days of age as compared to vehicle treatment.

The effect of cell transplantation has postponed the motor neuron loss. It has been recently suggested that probably two pathogenetic pathways get involved in the disease onset or progression rate. Boillee et al. (2006) have described that the onset of disease may be more related to the expression of SOD1 in motor neurons, whereas duration of disease may be related to expression of SOD1 in microglia. In our experiments, Le+CX+ seemed to influence the disease onset and not the rate of progression. Thus, likely the life-span enhancement is mainly due to the fact that cell transplantation modifies the mechanisms regulating the onset and not the progression.

We hypothesized that this beneficial effect was due to multiple events induced by stem-cell transplantation: the integration of new donor cells and the protective environmental change on endogenous MNs by trophic support. Le+CX+ cells can integrate within the spinal cord ventral horns, express motor neuron markers and develop extensive neuritic outgrowth, therefore providing an additional cell population to the primary site of the neuropathological lesion.

The demonstration of VEGF and IGF1 production by Le+CX+ cells in vitro and their increased level in treated spinal cords are consistent with the neuroprotective effects exerted by these cells.

In vitro Le+CX+ expressed similar levels of IGF and VEGF, while in vivo we detected a lower level of VEGF compared to the level of IGF1 both in wild type and SOD1 mice spinal cord before and after treatment. The in vivo data are in line with those observed by other groups (Nygren et al., 2002; Kaspar et al., 2005; Storkebaum et al., 2005; Bilic et al., 2006), although other data were obtained using different experimental conditions.

The major increase observed in the IGF1 respect to the VEGF may be attributed to a more significant production of IGF rather than VEGF by cells in vivo, or to an indirect growth factor stimulation within the host's tissue.

IGF1 and VEGF have been suggested to play a significant role in ALS from both pathogenic and therapeutic points of view. These neurotrophic factors regulate the growth and survival of certain populations of neurons in the central and peripheral nervous systems and exert a protective effect against MN degeneration.

Experimental studies showed that overexpression of IGF1 or VEGF significantly prolonged survival and motor function in a mouse model of ALS (Kaspar et al., 2005; Vande Velde and Cleveland, 2005). We demonstrated also that immuno-positive MNs for IGF-1Rβ and IGFBP5 were increased in untreated SOD1-G93A mice compared to wt mice. IGFBP5 expression level slightly decreased while IGF1R was significantly downregulated in transplanted mice. We hypothesized that free IGF1 production from stem cells reduces the presence of IGF1-R, inducing a neuroprotective effect. Recently it was shown that both IGF-1R and IGFBPs were increased on MNs of ALS patients (Wilczak et al., 2003). We confirmed an IGF1R upregulation also in SOD1-G93A murine spinal cord. Indeed, we previously described the upregulation of IGFBP5 in the nmd mice, an animal model of SMARD 1, an infantile MN disorder (Corti et al., 2006). Wilczak et al. (2003) reported that, although the IGF-1 expression in the spinal cords of ALS patients was normal, the level of free IGF-1 was 53% lower than in control patients. Moreover, they demonstrated the increase of IGF binding proteins (IGFBP) 2, 5 and 6 in ALS spinal cords, suggesting their binding to IGF-1 with high affinity and thus preventing the binding of IGF-1 to IGF-1R (Wilczak et al, 2003). Combined with these previous results, the present experiment suggests that increased level of IGF-1 by NSCs may reconstitute a normal IGF1 signalling.

The expression of mutant SOD1 genes in transgenic mice generated a useful model whose general features resemble ALS in humans. However, unfortunately, no animal model reproduces all the salient features of ALS, particularly the involvement of corticospinal and corticobulbar tracts.

In conclusion, we have demonstrated that transplantation with a NSC cell subpopulation prolongs survival in SOD1-G93A mice, suggesting that this class of stem cells may be efficient in MN diseases both by integration of new donor cells and other neuroprotective mechanisms.

Basic advances in understanding stem cells suggest their use as a possible therapeutic strategy, but many biological hurdles (such as the definition of the better cell type and methods of transplantation in humans) need to be overcome in order to prove that stem cells can be a realistic therapeutic strategy in neurodegenerative diseases.

Supplymentary material

Supplymentary data are available at Brain Online.


The financial support of the following research grants to G.P.C. is gratefully acknowledged: Italian Ministry of Health, Ricerca Finalizzata 2004 ‘Studio di protocolli di terapia cellulo-mediata nelle patologie neurodegenerative e nelle distrofie muscolari’; MIUR (Ministero Istruzione Università e Ricerca Scientifica) Italian Ministery FIRB 2002 ‘Animali geneticamente modificati per lo studio di patologie neurodegenerative’; Istituto Superiore di Sanità (ISS) 2006 ‘Studio di popolazioni staminali somatiche nello sviluppo di una terapia cellulare per la SLA’. Special thanks go to ‘Associazione Amici del Centro Dino Ferrari’ for their support.


  • Abbreviations:
    amyotrophic lateral sclerosis
    neural stem cells
    central nervous system
    neural growth factor


View Abstract