Neuroprotective and Angiogenic Effects of Bone Marrow Transplantation Combined With Granulocyte Colony-Stimulating Factor in a Mouse Model of Amyotrophic Lateral Sclerosis

Abstract

Bone marrow (BM) cells from amyotrophic lateral sclerosis (ALS) patients show significantly reduced expression of several neurotrophic factors. Monotherapy with either wild-type (WT) BM transplantation (BMT) or granulocyte colony-stimulating factor (GCSF) has only a small clinical therapeutic effect in an ALS mouse model, due to the phenomenon of neuroprotection. In this study, we investigated the clinical benefits of combination therapy using BMT with WT BM cells, plus GCSF after disease onset in ALS mice [transgenic mice expressing human Cu/Zn superoxide dismutase (SOD1) bearing a G93A mutation]. Combined treatment with BMT and GCSF delayed disease progression and prolonged the survival of G93A mice, whereas BMT or GCSF treatment alone did not. Histological study of the ventral horns of lumbar cords from G93A mice treated with BMT and GCSF showed a reduction in motor neuron loss coupled with induced neuronal precursor cell proliferation, increased expression of neurotrophic factors (glial cell line-derived neurotrophic factor, brain-derived neurotrophic factor, vascular endothelial growth factor and angiogenin), and neovascularization compared with controls (vehicle only). Compared with G93A microglial cells, most BM-derived WT cells differentiated into microglial cells and strongly expressed neurotrophic factors, combined BMT and GCSF treatment led to the replacement of G93A microglial cells with BM-derived WT cells. These results indicate combined treatment with BMT and GCSF has potential neuroprotective and angiogenic effects in ALS mice, induced by the replacement of G93A microglial cells with BM-derived WT cells. Furthermore, this is the first report showing the effects of combined BMT and GCSF treatment on blood vessels in ALS.

INTRODUCTION

Amyotrophic lateral sclerosis (ALS) is a progressive, fatal neurodegenerative disease characterized by the selective loss of central and peripheral motor neurons. Many hypotheses have been put forward to explain the cause of this disease, including glutamate toxicity (9,20), oxidative stress (13), axonal transport deficiency (18,31,50), protein misfolding and aggregation (8,28), and mitochondrial dysfunction (33,38,39). In approximately 15–20% of familial ALS cases, a variety of dominant missense mutations, or small deletions, in the Cu/Zn superoxide dismutase (SOD1) gene have been identified (7,17,51). Several lines of transgenic (Tg) mice expressing a mutant human SOD1 gene have been established, and these act as valuable models for human ALS (24,56). Nonneuronal cells are also thought to be involved in the pathogenesis of ALS (12,49). The expression of mutant SOD1 in astroglial and microglial cells contributes to the progression of motor neuron degeneration (5,42,57). In contrast, wild-type (WT) nonneuronal cells extend the survival of mutant SOD1 transgenic mice (12).

Bone marrow (BM) cells from ALS patients significantly reduce the migration and expression of several neurotrophic factors (11). BM transplantation (BMT), using WT BM cells, has clinical therapeutic effects and slows motor neuron loss in G93A SOD1 Tg mice (15). The highest efficiency of neural stem cell delivery to the central nervous system is found in symptomatic ALS mice (40), and BM cells are potential neural stem cells (29,58). Although WT BM cells mainly differentiate into microglial cells in both the brain and spinal cord of ALS mice (3,15), BMT using WT BM cells enhances the expression of neurotrophic factors and suppresses the production of cytotoxic factors (4). This suggests that the main therapeutic benefit is due to neuroprotection, rather than neurogenesis. Granulocyte colony-stimulating factor (GCSF), an hematopoietic growth factor, is known to have both neuroprotective and neuroregenerative effects (52,53) and administration of GCSF shows clinical therapeutic effects in G93A mice (48). However, in previous studies, BMT and GCSF treatment of G93A mice was started before disease onset (15,30), and the therapeutic effects in ALS patients are unclear (44).

In a murine stroke model, combined treatment with BMT and GCSF showed significant neuroprotective and neuroregenerative effects compared with BMT alone (30). Combined treatment also leads to significant proliferation of BM-derived cells and their differentiation into mature neurons within the brain (30). Combined treatment with BMT (using WT BM cells) and GCSF can be expected to further increase the number of WT BM-derived cells within the central nervous system of ALS mice and has potentially significant clinical benefits compared with BMT or GCSF treatment alone. Therefore, we planned to start combined BMT and GCSF treatment of ALS mice after disease onset. Previous studies in ALS mice focused only on clinical experiments, the neuroprotective effects on motor neurons, amelioration of gliosis, and neurogenesis. A recent report focused on disruption of the blood–spinal cord barrier in ALS mice (21,22,60); however, the effects of treatment on such vascular changes are still unclear. Here, we show that treatment of ALS mice with a combination of BMT and GCSF provides therapeutic benefit in terms of both neuroprotection and angiogenesis.

MATERIALS AND METHODS

Animal Models

All experimental procedures were carried out according to the guidelines of the Animal Care and Use Committee of the Graduate School of Medicine, Dentistry and Pharmaceutical Sciences of Okayama University. Transgenic (Tg) mice with the G93A human SOD1 mutation (G1H/+) were obtained from Jackson Laboratories (Bar Harbor, ME, USA) (24) and maintained as hemizygotes by mating Tg males with C57BL/6J females. Green fluorescent protein (GFP) Tg mice [C57BL/6 TgN (act-EGFP) OsbC14-Y01-FM131], a kind gift from Dr. Okabe (47), were used as BM donors. G93A Tg mice experience disease onset at around 91 days of age (48); therefore, 91-day-old G93A Tg mice were used in this study. Mice were divided into four experimental groups: G93A Tg mice treated with vehicle (n = 13; 6 male and 7 female), G93A Tg mice treated with BMT alone (n = 16; 9 male and 7 female), G93A Tg mice treated with GCSF alone (Chugai Pharmaceutical, Tokyo, Japan; n = 18; 10 male and 8 female), and G93A Tg mice treated with BMT + GCSF (n = 22; 12 male and 10 female). All groups were used to assess clinical scores and survival rates. Other G93A Tg mice, treated with vehicle (n = 5), BMT alone (n = 5), GCSF alone (n = 6), or BMT and GCSF (n = 6), and six non-Tg control WT B6SJL littermates were used to assess blood counts and spleen weights and for histological studies.

BMT

Bone marrow transplantation was carried out as previously described (55). Briefly, 91-day-old G93A Tg recipient mice received two doses (both 5 Gy) of lethal whole-body irradiation with a 6.5-h interval to minimize gastrointestinal toxicity. Immediately after irradiation, they were transplanted with BM cells harvested from the femurs of 9–12-week-old GFP donors. Donor bone marrow cells were resuspended in Hank’s balanced salt solution (HBSS) and 1 × 10⁷ BM cells (suspended in 200 μl of HBSS) were injected into the tail vein of the recipients. The recipient mice were maintained in a specific pathogen-free environment (isolated singly in plastic cages under the conventional facility in the experimental animal center of Okayama University) after BMT. The mice received normal chow and hyperchlorinated drinking water for the first 3 weeks after BMT, and they received normal chow and distilled water after the 3-week period.

Administration of GCSF and 5-Bromodeoxyuridine Labeling

Recombinant human GCSF was dissolved in Milli-Q water (Millipore, Bedford, MA, USA). Subcutaneous administration of GCSF (0.6 mg/kg) or vehicle (saline) was started the day after BMT. G93A Tg mice were injected every 2 days over a period of 18 days (a total of 10 injections).

The cell proliferation marker, 5-bromodeoxyuridine (BrdU; Sigma, St. Louis, MO, USA), was dissolved in saline. Intraperitoneal injections of BrdU (50 mg/kg) were also started the day after BMT. G93A Tg mice were injected every 2 days over a period of 18 days (a total of 10 injections) along with administration of GCSF. Using this method, proliferating cells in the spinal cord could be labeled.

Clinical Scores

For behavioral studies, rotarod tests were performed twice per week starting at 90 days of age (1,43,45). Rotarod testing began with mice trying to stay on a rod rotating at 1 rpm; the speed was then increased by 1 rpm every 10 s. The length of time that mice stayed on the rod (up to a maximum of 5 min) was recorded as an indicator of grasping power. Three trials were performed and the best result was recorded. A recovery period of at least 10 min was allowed between trials. Using rotarod analysis, the clinical midpoint and end stage were defined as <50% or <10% of the maximum recorded time, respectively. The time point at which mice were unable to roll over within 20 s of being pushed onto their side was recorded as “time of death” by an investigator blinded to the treatment conditions. Mice were euthanatized by deep anesthesia with an intraperitoneal infusion of sodium pentobarbital (10 mg; Abbott Laboratories, Abbott Park, IL, USA) at time of death. Disease duration was defined as 91 days of age to time of death.

Blood Cell Counts and Spleen Weight

For the assessment of blood counts, spleen weights, and histological studies, six non-Tg control WT B6SJL littermates and five or six G93A Tg mice treated with vehicle, BMT, GCSF, or BMT with GCSF at 130 days of age were deeply anesthetized with an intraperitoneal infusion of sodium pentobarbital (10 mg; Abbott Laboratories) and heparinized blood samples were obtained via cardiac puncture. The following parameters were measured using a multiparameter auto-cell counter (MEK-6450; Nihon Kohden, Tokyo, Japan): erythrocytes, hemoglobin, hematocrit, leukocytes, and platelets. After blood sampling, all mice were perfused transcardially with heparinized saline. Spleens were removed from the WT and G93A Tg mice and weighed.

Motor Neuron Count

The lumbar cord spanning L4–L5 and the cervical cord spanning C5–C6 was removed from the WT and G93A Tg mice, fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4), and frozen after cryoprotection using a series of phosphate-buffered sucrose solutions of increasing concentration (10%, 20%, and 30%). One hundred consecutive transverse sections (10 μm in thickness) of the L4 and C5 segments were cut using a cryostat, and the motor neurons in L4 and C5 were counted in every fifth transverse section (20 sections in total) from each lumbar and cervical cord after staining with cresyl violet (Nissl stain) (43). All cells in both ventral horns (VH) below a lateral line across the spinal cord from the central canal were microscopically video-captured, and only cells with a diameter greater than 20 μm with clear nucleoli were counted by an investigator blinded to the treatment conditions (37). The average of 20 sections was taken as the number of motor neurons in each mouse.

Immunofluorescence Analysis

To determine whether BM-transplanted GFP cells differentiated into neuronal precursors, mature neuronal, astroglial, or microglial cells within the lumbar cords of G93A Tg mice, we performed double immunofluorescence studies using GFP together with highly polysialylated neural cell adhesion molecule (PSA-NCAM), microtubule-associated protein 2 (MAP2), glial fibrillary acidic protein (GFAP), or ionized calcium-binding adapter molecule-1 (Iba1) and PSA-NCAM together with doublecortin (DCX). PSA-NCAM and DCX, MAP2, GFAP, and Iba1 are markers for neuronal precursors, differentiating neuronal cells, astroglial cells, and microglial cells, respectively. Five transverse sections from each lumbar cord were incubated overnight at 4°C with rabbit polyclonal anti-GFP [1:1000; Medical and Biological Laboratories (MBL), Nagoya, Japan] and an antibody for each cellular marker: mouse monoclonal anti-PSA-NCAM (1:200; Chemicon International, Temecula, CA, USA), mouse monoclonal anti-MAP2 (1:500; Chemicon), mouse monoclonal anti-GFAP (1:200; Chemicon), or goat polyclonal anti-Iba1 (1:500; Abcam, Cambridge, UK). Also, lumbar cord sections from G93A mice treated with vehicle or GCSF alone were incubated overnight at 4°C with a mixture of antibodies to rabbit polyclonal anti-DCX (1:1000; Abcam) and the same anti-PSA-NCAM antibody. After rinsing in PBS, the sections were incubated with secondary antibodies conjugated to Alexa 488 and 555 (Molecular Probes, Eugene, OR) for1hatroom temperature. The sections were then mounted using Vectashield mounting medium containing DAPI (Vector Laboratories, Burlingame, CA, USA). The specificity of immunostaining was confirmed by omission of the primary antibody. Sections for double immunofluorescence studies were scanned using a confocal microscope equipped with argon and HeNe1 lasers (LSM-510; Zeiss, Jena, Germany). To quantify the GFP-labeled cells, the number of positively stained cells within five transverse sections from each lumbar cord was counted per animal. We noted the total number and the regional differences between the VH, and the dorsal horn (DH) above a lateral line across the spinal cord from the central canal, and white matter (WM). In the same manner, PSA-NCAM-, MAP2-, GFAP-, and Iba1-labeled cells that were also positive for GFP were counted.

To investigate the neuroprotective effects of combined BMT and GCSF treatment, we performed single immunofluorescence studies using glial cell line-derived neurotrophic factor (GDNF), brain-derived neurotrophic factor (BDNF), vascular endothelial growth factor (VEGF), and angiogenin (ANG). Five transverse sections from each lumbar cord were incubated with antibodies to each neurotrophic factor: mouse monoclonal anti-GDNF (1:200; Santa Cruz Biotechnology, Santa Cruz, CA, USA), goat polyclonal anti-BDNF (1:200; Santa Cruz), goat polyclonal anti-VEGF (1:500; Santa Cruz), or mouse monoclonal anti-ANG (1:200; Santa Cruz), respectively, and subsequently detected using the avidin-biotin complex method. Immunoreactivity was visualized using 3,3′-diaminobenzidine tetrahydrochloride. Additionally, lumbar cord sections from G93A mice treated with BMT alone or with BMT and GCSF were incubated overnight at 4°C with a mixture of antibodies to the neurotrophic factors and the same anti-GFP antibody. In contrast, lumbar cord sections from G93A mice treated with vehicle or GCSF alone were incubated overnight at 4°C with a mixture of antibodies to the neurotrophic factors and rabbit polyclonal anti-neurofilament (NF) (1:500; Millipore), anti-Iba1 (1:200; Wako, Osaka, Japan), or anti-GFAP (1:200; Dako, Glostrup, Denmark). After rinsing in PBS, the sections were incubated with secondary antibodies conjugated to Alexa 488 and 555 (Molecular Probes) for 1 h at room temperature. For the semiquantitative evaluation of the expression of neuroprotective factors in the VH, only the VH within sections stained with a single antibody for each neurotrophic factor was microscopically video-captured. The staining intensity was measured using image processing software (Scion Image, Scion Corporation, Frederick, MD, USA). Furthermore, GDNF-, VEGF-, and ANG-immunopositive large neuronal cells (diameter greater than 20 μm) in the VH were microscopically video-captured and counted by an investigator blinded to the treatment conditions.

To investigate vascular changes, we performed single immunofluorescence studies using CD31, a marker for vascular endothelial cells. Five transverse sections from each lumbar cord were incubated with a rat monoclonal anti-CD31 antibody (1:50; Becton Dickinson, Franklin Lakes, NJ, USA) and detected using avidin-biotin. The density of CD31-positive vascular endothelium in the VH was measured using image processing software (Scion Image, Scion Corporation). To investigate the angiogenic effects of BMT + GCSF, we performed double immunofluorescence studies using CD31 and GFP and triple immunofluorescence studies using CD31, GFP, and BrdU. Five transverse sections from each lumbar cord were incubated with a mixture of the same anti-CD31 and GFP antibodies used for the double immunofluorescence studies. For triple immunofluorescence studies, five transverse sections from each lumbar cord were first incubated in 2 N HCl at 37°C for 20 min to denature the DNA, and then rinsed in 0.1 M boric acid (pH 8.5) at 25°C for 10 min. The sections were then incubated overnight at 4°C with a mixture of anti-CD31, anti-GFP, and mouse monoclonal anti-BrdU antibodies (1:100; Sigma-Aldrich, St. Louis, MO, USA). After rinsing in PBS, the sections were incubated with secondary antibodies conjugated to Alexa 405, 488, and 555 (Molecular Probes) for 1 h at room temperature. Sections for triple immunofluorescence studies were scanned using a confocal microscope equipped with LD405, 473 and 559 lasers (FV10i; Olympus, Tokyo, Japan).

Statistical Analysis

Data were expressed as means ± SD. Statistical comparisons of histological data with respect to the number of GFP and PSA-NCAM double-positive cells between the BMT and combined BMT and GCSF treatment groups were performed using t-tests. Statistical comparisons of clinical and other histological data were performed using one-way ANOVA. Data from the rotarod analysis were compared using repeated-measures ANOVA, followed by a Tukey-Kramer post hoc comparison. Kaplan-Meier survival analysis and the log-rank test were used for survival and comparison of clinical midpoints and end stages. Statistical significance was set at p < 0.05.

RESULTS

Combination of BMT and GCSF Prolongs Survival in G93A SOD1 Tg Mice

To examine the clinical benefits of BMT with GCSF treatment after disease onset, 91-day-old G93A mice were treated with vehicle, BMT only, GCSF only, or BMT and GCSF. The survival time of G93A mice treated with either BMT or GCSF alone (142.5 ± 8.5 days, n = 16; 143.7 ± 8.2 days, n = 18, respectively) was similar to that of mice treated with vehicle (140.6 ± 3.8 days, n = 13) (Fig. 1a). However, the survival time of mice treated with BMT and GCSF was 148.2 ± 9.0 days (n = 22), representing an increased survival time of 5.4% compared with vehicle-treated mice (χ² = 11.73, p = 0.0006) (Fig. 1a). The disease duration in mice treated with BMT and GCSF (57.2 ± 9.0 days) was prolonged by 15.3% (*p < 0.05) compared with vehicle-treated mice (49.6 ± 3.8 days). Mice treated with either BMT or GCSF alone showed only small increases in disease duration (51.5 ± 8.5 and 52.7 ± 8.2 days, respectively) (Fig. 1b).

Figure 1.

Disease progression and survival in G93A SOD1 (human Cu/Zn superoxide dismutase bearing a G93A mutation) transgenic (Tg) mice. (a) The cumulative probability of survival for G93A mice treated with vehicle (n = 13), bone marrow transplantation (BMT; n = 16), granulocyte colony-stimulating factor (GCSF; n = 18), or BMT and GCSF (n = 22) is shown in the Kaplan-Meier curve. (b) BMT and GCSF treatment prolongs disease duration in G93A mice compared with vehicle. Data represent means ± SD. *p < 0.05 (one-way ANOVA followed by Tukey-Kramer post hoc comparison). (c) BMT and GCSF treatment improves motor performance in the rotarod test (not significant). (d) The cumulative probability of a drop in time (<10% of the max time) recorded in the initial rotarod analysis is shown by a Kaplan-Meier curve.

Combined BMT With GCSF Treatment Improves Motor Performance in G93A SOD1 Tg Mice

Mice treated with vehicle alone showed a progressive decline in motor performance in the rotarod test (Fig. 1c). Combined BMT and GCSF treatment resulted in improved motor performance (not significant), whereas mice treated with BMT or GCSF alone showed little improvement (Fig. 1c).

To evaluate the effect of treatment on disease progression, the clinical midpoint and end stage were defined as <50% and <10%, respectively, of the “maximum” time recorded in the initial rotarod test. Combined BMT and GCSF treatment significantly prolonged both the clinical midpoint and end stage (125.3 ± 8.1 days, χ² = 7.477, p = 0.0062; and 138.6 ± 6.3 days, χ² = 10.96, p = 0.0009, respectively) compared with vehicle treatment (117.5 ± 6.6 and 130.3 ± 6.0 days, respectively), whereas treatment with BMT alone (119.6 ± 7.0 and 132.8 ± 7.4 days, respectively) or GCSF alone (119.4 ± 9.4 and 134.4 ± 8.4 days, respectively) had little effect (Fig. 1d).

BMT, GCSF, and Combined BMT + GCSF Treatment Ameliorate Motor Neuron Loss in the Lumbar and Cervical Cord of G93A SOD1 Tg Mice

Blood samples were taken from five or six G93A mice in each treatment group and also from six WT mice at 130 days of age. Mice were then sacrificed and their spleens and spinal cords removed. Complete blood cell counts showed a decreased number of leukocytes in G93A mice treated with vehicle compared with that in WT mice. However, this increased after BMT and GCSF treatment (Table 1). The weight of the spleens from G93A mice treated with vehicle was significantly less than that of WT mice (*p < 0.05), but increased after combined BMT and GCSF treatment (Table 1).

Table 1.

Hematologic Parameters and Spleen Weight of WT and G93A SOD1 Tg Mice

		G93A
	WT (n = 6)	Vehicle (n = 5)	BMT (n = 5)	GCSF (n = 6)	BMT + GCSF (n = 6)
Leukocytes (cells × 102/mm3)	16.2 ± 9.3	9.6 ± 5.2	8.0 ± 4.1	10.2 ± 7.7	17.7 ± 14.7
Erythrocytes (cells × 105/mm3)	87.9 ± 6.0	87.4 ± 10.5	65.9 ± 13.0	82.6 ± 18.7	80.5 ± 18.3
Hemoglobin (g/dl)	12.4 ± 1.0	12.2 ± 1.5	9.5 ± 2.0	11.8 ± 2.8	11.5 ± 2.3
Hematocrit (%)	42.1 ± 2.6	41.6 ± 5.1	32.6 ± 6.7	40.1 ± 9.1	39.8 ± 8.3
Platelet (cells × 104/mm3)	66.1 ± 30.6	68.2 ± 13.1	53.3 ± 12.9	51.4 ± 20.3	47.2 ± 28.0
Spleen weight (mg)	99.7 ± 15.9	56.4 ± 16.2*	65.4 ± 17.1	69.5 ± 21.4	77.8 ± 26.9

Data represent mean ± SD. WT, wild-type; SOD1, Cu/Zn superoxide dismutase; Tg, transgenic; BMT, bone marrow transplantation; GCSF, granulocyte colony-stimulating factor.

^* p < 0.05 compared with WT mice.

We counted the number of surviving Nissl-stained motor neurons in the L4 and C5 spinal cord sections from these mice. The number of motor neurons in the lumbar and cervical VH was significantly higher in mice treated with either BMT or GCSF alone compared with those treated with vehicle (**p < 0.01; Fig. 2a, b). However, the number of motor neurons in the lumbar and cervical VH of mice treated with BMT and GCSF was significantly higher than that in mice treated with vehicle, BMT, or GCSF alone (**p < 0.01; Fig. 2a, b). The number of motor neurons in the lumbar VH of mice treated with BMT and GCSF was 1.88-fold higher than in mice treated with vehicle and the number of motor neurons in the cervical VH of mice treated with BMT and GCSF was 1.60-fold higher than in those with vehicle. These results suggest that combined BMT and GCSF treatment is more effective in reducing motor neuron loss in the lumbar cord than it is in the cervical cord.

Figure 2.

Motor neuron counts in the L4 lumbar and C5 cervical cords of G93A SOD1 Tg mice. (a) There is an increased number of surviving Nissl-stained motor neurons in mice treated with BMT and GCSF compared with that in vehicle-, BMT-, or GCSF-treated mice. Data represent means ± SD (**p < 0.01, one-way ANOVA followed by Tukey-Kramer post hoc comparison). (b) Representative photomicrographs of Nissl-stained motor neurons in the lumbar ventral horn (VH) of wild-type (WT) or G93A mice. Scale bar: 50 μm.

Combined BMT With GCSF Treatment Activates the Proliferation of Neuronal Precursor Cells in the VH of G93A SOD1 Tg Mice

To evaluate gliosis, neurogenesis, and the characteristics and differentiation status of BM-derived GFP cells in the lumbar cords of the G93A mice within each treatment group, we immunostained L4 spinal cord sections with anti-Iba1, anti-GFAP, anti-PSA-NCAM, anti-DCX, and anti-MAP2 antibodies (cellular markers for microglial, astroglial, neuronal precursor and differentiating neuronal cells, respectively) and a mixture containing anti-GFP together with anti-Iba1, anti-GFAP, anti-PSA-NCAM or anti-MAP2, and anti-PSA-NCAM plus anti-DCX.

The number of Iba1-positive cells in the VH of mice treated with GCSF alone or with BMT + GCSF increased compared with that in mice treated with vehicle or BMT alone, although the total number of Iba1-positive cells, or the number of Iba1-positive cells in the DH or WM, did not change in any of the treatment groups (*p < 0.05, **p < 0.01; Fig. 3a). Upon regional analysis, the number of Iba1-positive cells in the VH and WM was larger than that in the DH in all treatment groups. The total number of GFAP-positive cells did not change, but the number of GFAP-positive cells within the VH and WM was greater than that in the DH in all treatment groups (**p < 0.01; Fig. 3b). The total number of PSA-NCAM-positive cells and the number of PSA-NCAM-positive cells within the VH of mice treated with BMT and GCSF increased compared with that in mice treated with vehicle (**p < 0.01; Fig. 3c). The number of PSA-NCAM-positive cells within the VH was larger than that in the DH or WM in all treatment groups.

Figure 3.

Neuronal precursor cell proliferation in the VH of G93A SOD1 Tg mice. (a–c) Total number of ionized calcium-binding adapter molecule-1 (Iba1), glial fibrillary acidic protein (GFAP), or highly polysialylated neural cell adhesion molecule (PSA-NCAM)-positive cells and regional differences between the VH, dorsal horn (DH) and white matter (WM) in the L4 lumbar cords from G93A mice (*p < 0.05, **p < 0.01, one-way ANOVA followed by Tukey-Kramer post hoc comparison). (d–f) Double immunofluorescence analysis of green fluorescent protein (GFP; green) and Iba1, or GFAP and PSA-NCAM (red) expression in the VH of G93A mice treated with BMT and GCSF. The smaller photomicrographs show the double-positive cells (d, f). Scale bar on larger photomicrographs: 50 μm; scale bar on smaller photomicrographs: 20 µm. (g–i) Total number of GFP-positive cells and Iba1 (one-way ANOVA followed by Tukey-Kramer post hoc comparison) or PSA-NCAM cells that were also positive for GFP (t-test), and the regional differences between the VH, DH, and WM in the L4 lumbar cords from BMT-treated and BMT with GCSF-treated G93A mice (*p < 0.05). (j) Double immunofluorescence analysis of doublecortin (DCX; green) and PSA-NCAM (red) expression in the VH of G93A mice treated GCSF. Scale bar: 20 µm. See Results section for details.

The number of GFP-positive cells in the VH of mice treated with BMT and GCSF increased compared with that in mice treated with BMT, and was larger than that in the DH in both groups (*p < 0.05; Fig. 3d–g). Double-labeling with antibodies to GFP together with Iba1, GFAP, PSA-NCAM, or MAP2 and PSA-NCAM together with DCX showed that many GFP-positive cells within the VH also expressed Iba1 (BMT alone = 73.7%; BMT + GCSF = 74.4%) and low levels of PSA-NCAM (BMT = 0.8%; BMT + GCSF = 1.8%) (Fig. 3h, i). Most of the PSA-NCAM-positive cells in the lumbar spinal cords of G93A mice expressed DCX (Fig. 3j). None of the GFP cells in the lumbar spinal cords of G93A mice treated with BMT alone or BMT and GCSF expressed GFAP or MAP2 (Fig. 3e). Upon regional analysis of the VH, the number of GFP and Iba1 double-positive cells in the BMT and GCSF group was significantly larger than that in the BMT group (*p < 0.05; Fig. 3h). Furthermore, the proportion of GFP and Iba1 double-positive cells within the Iba1-positive cell population was larger in the BMT with GCSF group (30.3%) than in the BMT group (10.7%). The number of GFP and PSA-NCAM double-positive cells in the BMT and GCSF group was also significantly greater than that in the BMT group (*p < 0.05; Fig. 3i). The proportion of GFP and PSA-NCAM double-positive cells within the PSA-NCAM-positive cell population was a little larger in the BMT with GCSF group (3.8%) than in the BMT group (1.4%). These results suggest that many of the GFP cells were microglial cells, and a few were neuronal precursor cells; however, none had differentiated into astroglial cells or mature neurons. GCSF treatment stimulated BM-derived GFP cells to proliferate and differentiate into microglial or neuronal precursor cells in the VH. Although 30.3% of endogenous G93A microglial cells in the VH were replaced with BM-derived GFP cells after GCSF stimulation, the majority of neuronal precursor cells (96.2%) in the VH were not BM-derived GFP cells, but endogenous cells from the G93A mice.

Combined BMT and GCSF Treatment Increases the Expression of Neurotrophic Factors in the VH of G93A SOD1 Tg Mice

To investigate the neuroprotective effect of BMT and GCSF treatment in the lumbar VH of G93A mice, we evaluated the expression of neurotrophic factors within the lumbar VH using immunofluorescence. Immunohistochemical analysis of GDNF showed weak immunoreactivity of neuronal cells in the VH of WT mice compared with G93A mice treated with vehicle (Fig. 4a, b). In G93A mice treated with vehicle, the immunoreactivity of neuronal and glial cells was strong, and was further enhanced after BMT and GCSF treatment. BDNF showed strong immunoreactivity throughout the whole gray matter, and even stronger immunoreactivity in neuronal cells within the VH of WT mice compared with G93A mice treated with vehicle (Fig. 4c, d). Immunoreactivity in the gray matter was weak (and even weaker in neuronal cells) in the VH of G93A mice treated with vehicle. However, both BMT and GCSF treatment enhanced the immunoreactivity of the whole gray matter, and BMT with GCSF treatment enhanced it significantly. However, little enhancement was seen in the immunoreactivity of neuronal cells. Immunohistochemical analysis of VEGF expression in the VH of WT mice showed only weak immunoreactivity of neuronal cells compared with G93A mice treated with vehicle (Fig. 4e, f). In the VH of G93A mice treated with vehicle, the immunoreactivity of neuronal and glial cells was strong, and was further enhanced after treatment with BMT and GCSF. Moreover, both GCSF and BMT + GCSF treatment enhanced the immunoreactivity of the whole gray matter. Immunohistochemical analysis of ANG within the VH of WT mice showed weak immunoreactivity of neuronal cells compared with G93A mice treated with vehicle (Fig. 4g, h). In the VH of G93A mice treated with vehicle, the immunoreactivity of both neuronal and glial cells was strong, and was further enhanced after treatment with BMT and GCSF.

Figure 4.

Expression of neurotrophic factors in the VH of G93A SOD1 Tg mice. (a, c, e, g) Immunohistological analysis of glial cell line-derived neurotrophic factor (GDNF), brain-derived neurotrophic factor (BDNF), vascular endothelial growth factor (VEGF), and angiogenin (ANG) expression in the VH of WT or G93A mice. The smaller photomicrographs show GDNF-, VEGF-, and ANG-immunopositive motor neurons. Scale bar on larger photomicrographs: 50 µm, scale bar on smaller photomicrographs: 20 µm. (b, d, f, h) Increased expression of GDNF, BDNF, VEGF, and ANG in the VH of mice treated with BMT + GCSF compared with vehicle. ##p < 0.01 and #p < 0.05 compared with vehicle-treated G93A mice. ††p < 0.01 compared with BMT-treated G93A mice. ‡‡p < 0.01 compared with GCSF-treated G93A mice. §§p < 0.01 compared with BMT and GCSF-treated G93A mice. *p < 0.05, **p < 0.01 (between the different treatment groups; one-way ANOVA followed by Tukey-Kramer post hoc comparison). (i) Increased numbers of GDNF-, VEGF-, and ANG-immunopositive motor neurons in mice treated with BMT + GCSF compared with those in vehicle-, BMT-, or GCSF-treated mice. **p < 0.01 (one-way ANOVA followed by Tukey-Kramer post hoc comparison). (j) Increased ratio of GDNF-, VEGF-, and ANG-immunopositive motor neurons to motor neurons in mice treated with BMT and GCSF compared with that in vehicle-treated mice. *p < 0.05, **p < 0.01 (one-way ANOVA followed by Tukey-Kramer post hoc comparison). See Results section for details.

To evaluate the neuroprotective effect on motor neurons, we counted the GDNF-, VEGF-, and ANG-immunopositive large neuronal cells (diameter greater than 20 μm) in the VH of G93A mice (Fig. 4a, e, g; smaller photomicrographs). The number of GDNF-, VEGF-, and ANG-immunopositive large neuronal cells within the VH was significantly higher in mice treated with either BMT or GCSF alone compared with vehicle treatment (**p < 0.01; Fig. 4a, e, g, i). However, the number of immunopositive cells in mice treated with BMT + GCSF was significantly higher than that in mice treated with vehicle, BMT, or GCSF alone (**p < 0.01; Fig. 4a, e, g, i). The percentage of GDNF-, VEGF-, and ANG-immunopositive large neuronal cells per motor neuron in mice treated with BMT + GCSF was also significantly higher than that in mice treated with vehicle (**p < 0.01; Fig. 4j).

Double immunofluorescence analysis showed that large neurons in the VH of G93A mice treated with vehicle or GCSF alone strongly expressed GDNF (Fig. 5a). GNDF was partially expressed by endogenous microglial cells and weakly expressed by astroglial cells. In the lumbar cords of G93A mice treated with BMT or BMT and GCSF, GFP-positive cells showed strong expression of GDNF compared with endogenous microglial cells. Large neurons and endogenous microglial and astroglial cells in the VH of G93A mice treated with vehicle or GCSF only weakly expressed BDNF (Fig 5b). In contrast, GFP cells in the lumbar cords of mice treated with BMT or BMT + GCSF strongly expressed BDNF compared with endogenous microglial cells. Large neurons in the VH of G93A mice treated with vehicle or GCSF alone showed weak expression of VEGF (Fig 5c). Astroglial cells strongly expressed VEGF, but it was only weakly expressed by endogenous microglial cells (Fig 5c, arrows). Compared with endogenous microglial cells, GFP cells in the lumbar cords of G93A mice treated with BMT or BMT and GCSF strongly expressed VEGF. Large neurons in the VH of G93A mice treated with vehicle or GCSF alone strongly expressed ANG, which was also partially expressed by endogenous microglial cells and weakly expressed by astroglial cells (Fig. 5d). GFP cells in the lumbar cords of G93A mice treated with BMT or BMT + GCSF showed strong expression of ANG compared with endogenous microglial cells.

Figure 5.

Expression of neurotrophic factors by GFP cells in the VH of G93A SOD1 Tg mice. (a–d) Double immunofluorescence analysis of GFP, neurofilament (NF), Iba1 and GFAP (green), and GDNF, BDNF, VEGF, and ANG (red) expression in the VH of G93A mice (GFP = BMT and GCSF-treated mice; NF, Iba1, and GFAP = GCSF-treated mice). Scale bar: 20 µm. See Results section for details.

These results suggest that treatment with BMT and GCSF significantly increases the expression of neurotrophic factors in the VH, especially within the motor neurons of G93A mice compared with vehicle treatment, and that GFP cells strongly express neurotrophic factors compared with endogenous microglial cells.

Combined BMT + GCSF Treatment Activates Neovascularization in the VH of G93A SOD1 Tg Mice

We also evaluated the vascular endothelium in the lumbar VH by immunostaining with anti-CD31 antibodies (a marker of vascular endothelial cells). The density of the vascular endothelium in the VH of G93A mice treated with vehicle was significantly less than that in WT mice (Fig. 6a, b). However, treatment with BMT + GCSF led to a significant improvement in G93A mice, whereas mice treated with BMT or GCSF alone showed little improvement.

Figure 6.

Vascular changes and neovascularization in the VH of G93A SOD1 Tg mice. (a) Immunohistological analysis of vascular endothelial cells (CD31) in the VHof WT or G93A mice. Scale bar: 50 μm. (b) Increased density of CD31-positive vascular endothelium in the VH of mice treated with BMT and GCSF compared with vehicle. ##p < 0.01 compared with vehicle-treated G93A mice. ††p < 0.01 compared with BMT-treated G93A mice. ‡‡p < 0.01 compared with GCSF-treated G93A mice. **p < 0.01 (between each treatment group; one-way ANOVA followed by Tukey-Kramer post hoc comparison). (c) Double immunofluorescence analysis of GFP (green) and CD31 (red) expression in the VH of G93A mice treated with BMT and GCSF. Scale bar: 10 µm. (d) Triple immunofluorescence analysis of 5-bromodeoxyuridine (BrdU; blue), GFP (green), and CD31 (red) expression in the VH of G93A mice treated with BMT and GCSF. Scale bar: 10 μm. (e) Number of BrdU/CD31 double-positive cells in the VH of G93A mice. Data represent means ± SD (*p < 0.05, one-way ANOVA followed by Tukey-Kramer post hoc comparison). See Results section for details.

To investigate the angiogenic effects of the different treatments, we performed double immunofluorescence studies on the lumbar cords of G93A mice treated with BMT or BMT + GCSF using a mixture of anti-GFP and anti-CD31 antibodies. Some GFP-positive cells were observed within blood vessels and perivascular sites, but no GFP-positive cells had merged with vascular endothelial cells (Fig. 6c), suggesting that BM-derived cells do not differentiate into vascular endothelial cells in G93A mice.

Next, we performed triple immunofluorescence studies on the lumbar cords of G93A mice using a mixture of BrdU, GFP, and CD31 antibodies to investigate neovascularization. All G93A Tg mice were given intraperitoneal injections of BrdU along with GCSF after BMT to label proliferating cells. The total number of BrdU-labeled cells in the lumbar cord, and the number of BrdU-labeled cells in the VH, DH, and WM of G93A mice treated with BMT + GCSF significantly increased compared with other treatments (data not shown). Upon regional analysis, the majority of BrdU-labeled cells were found in the WM, but the number of BrdU-labeled cells in the VH was larger than that in the DH in all treatment groups. All the BrdU/CD31 double-positive cells, which were GFP negative, were present only in the VH (Fig. 6d). The number of BrdU/CD31 double-positive cells in the VH of mice treated with BMT + GCSF was significantly higher than that seen with the other treatments (Fig. 6e), suggesting that treatment with BMT + GCSF activates neovascularization in the VH of G93A mice.

DISCUSSION

Previous reports show that leukocyte numbers and spleen weight in G93A mice decrease during the late stages of disease, which may suggest immunodeficiency in ALS (2,34). Our study also showed that both leukocyte numbers and spleen weight in G93A mice treated with vehicle were reduced at 130 days of age. However, these were significantly ameliorated by treatment with BMT and GCSF (Table 1). In a murine stroke model, combined treatment with BMT and GCSF showed significant neuroprotective and neuroregenerative effects compared with BMT alone (30). Neural regeneration can be divided into three steps: proliferation, migration, and differentiation (27,46). Therefore, GCSF was administered not only to provide a neuroprotective effect but also to stimulate the proliferation of BM cells, which have potential as neural stem cells (29,58). The administration of GCSF was stopped for only 18 days so that we could observe the migration and differentiation of the BM-derived cells after GCSF administration. Immunofluorescence studies using GFP and Iba1, PSA-NCAM, DCX, GFAP, or MAP2 showed that the number of neuronal precursor cells in the lumbar VH (but not in the DH and WM) of G93A mice treated with BMT and GCSF significantly increased compared with that in mice treated with vehicle (Fig. 3a–c), suggesting that combined treatment significantly activates the proliferation of neuronal precursor cells in the VH of ALS mice without any proliferation of microglial and astroglial cells. Since there are few reports about neurogenesis in the spinal cord of ALS mice (46), we performed double immunofluorescence studies using PSA-NCAM and DCX. We confirmed that PSA-NCAM-positive cells were neuronal precursor cells, as most of them also expressed DCX (Fig. 3j). A significant increase in the proliferation of BM-derived GFP cells was seen in the VH of mice treated with BMT and GCSF, but the majority were microglial cells (30.3% of total microglial cells); some were neuronal precursor cells, but no GFP cells formed mature neurons or astroglial cells (Fig. 3d–j). These results suggest that combined treatment with BMT + GCSF induced the proliferation of BM-derived WT cells and efficiently replaced 30.3% of mutant SOD1 microglial cells with BM-derived WT microglial cells. However, it did not activate neurogenesis by BM-derived WT cells in the VH of symptomatic ALS mice.

Combined BMT with GCSF treatment resulted in a marked reduction in motor neuron loss in the lumbar VH rather than in the cervical VH of G93A mice (Fig. 2a, b), and induced the proliferation of endogenous neuronal precursor cells (Fig. 3c, f, i). This suggests that combined BMT and GCSF treatment had a significant neuroprotective effect in the lumbar VH of G93A mice. BM cells produce several neurotrophic factors including GDNF, BDNF, VEGF, and fibroblast growth factor 2 (10,14,25), and GCSF treatment increases the expression of BDNF by microglial cells (6). Therefore, we next investigated the expression of GDNF, BDNF, VEGF, and ANG (which are reported to be neuroprotective in ALS mice) (32,37,59) in the VH of G93A mice in all treatment groups. It is known that GDNF and VEGF expression increases, while that of BDNF decreases, in the spinal cords of symptomatic ALS mice compared with those of WT mice (3,41), and that serum ANG levels are elevated in ALS patients compared with healthy controls (16). In the present study, immunofluorescence experiments showed that the expression of GDNF, VEGF, and ANG increased (Fig. 4a, b, e–h) and that BDNF expression decreased in the VH of G93A mice treated with vehicle compared with WT mice (Fig. 4c, d). Combined treatment with BMT + GCSF significantly increased the expression of these factors in the VH, especially in motor neurons of G93A mice compared with vehicle treatment (Fig. 4a–j). Although endogenous microglial cells in G93A mice strongly expressed GDNF and ANG, and weakly expressed BDNF and VEGF, WT (GFP) cells (which mainly consist of microglial cells) showed strong expression of these neurotrophic factors compared with endogenous microglial cells (Fig. 5a–d). The expression of mutant SOD1 in astroglial and microglial cells accelerates motor neuron degeneration in ALS mice (5,42,57), and WT microglial cells are neuroprotective for motor neurons (3,15), although ablation of proliferating mutant SOD1 microglial cells does not affect motor neuron degeneration (23). Combined treatment with BMT and GCSF significantly increased BM-derived WT (GFP) microglial cells in the VH without causing the proliferation of astroglial or microglial cells. Also, WT microglial cells strongly expressed neurotrophic factors compared with endogenous G93A microglial cells. These results suggest that combined treatment with BMT and GCSF has a significant neuroprotective effect, due to the efficient microglial replacement of mutant SOD1 cells with BM-derived WT cells in the VH of symptomatic ALS spinal cords. Transplantation-based astrocyte replacement also has a significant neuroprotective effect in ALS mice (36). Thus, therapies that induce a nonneuronal environmental change may be an effective clinical tool for the treatment of ALS.

Recently, disruption of the blood–spinal cord barrier and a decrease in capillary length were reported in both asymptomatic and symptomatic ALS mice with SOD1 mutations, which were related to neurotoxicity (21,22,60). BM-derived cells do not differentiate into vascular endothelial cells in normal brain and spinal cord (19), but VEGF or GCSF treatment induces BM-derived cells to differentiate into vascular endothelial cells in normal, or ischemic, mouse brain (26,54). In the present study, combined treatment with BMT and GCSF increased the expression both of VEGF and ANG in the VH of G93A mice (Fig. 4e–h). Because both VEGF and ANG activate angiogenesis (35), we first evaluated the vascular changes within the VH using an antibody to CD31. Compared with WT mice, the density of vascular endothelium was significantly reduced in the VH of G93A mice treated with vehicle (Fig. 6a, b). However, combined treatment with BMT and GCSF led to a significant improvement in G93A mice (Fig. 6a, b). Although BM-derived GFP cells did not differentiate into vascular endothelial cells in the spinal cords of G93A mice (Fig. 6c), BrdU/CD31 double-positive (GFP negative) cells (signifying proliferating endogenous vascular endothelium) were observed only in the VH (Fig. 6d) of G93A mice. Combined treatment with BMT and GCSF further increased the amount of proliferating, endogenous vascular endothelium in the VH (Fig. 6e). The mechanism underlying vascular disruption in ALS is unclear, but combined treatment with BMT and GCSF led to an improvement in these vascular changes and activated neovascularization in ALS mice, which would be induced by increased VEGF and ANG expression due to microglial replacement.

This is the first report showing the effects of combined treatment with BMT + GCSF on blood vessels in ALS. The results suggest that treatments that provide protection for blood vessels and promote angiogenesis may be successful for ALS patients.