|Home | About | Journals | Submit | Contact Us | Français|
Transplantation of mesenchymal stem cells (MSCs) derived from bone marrow has been shown to improve functional outcome in spinal cord injury (SCI). We transplanted MSCs derived from human bone marrow (hMSCs) to study their potential therapeutic effect in SCI in the rat. In addition to hMSCs, we used gene-modified hMSCs to secrete brain-derived neurotrophic factor (BDNF-hMSCs). After a dorsal transection lesion was induced at T9, cells were microinjected on each side of the transection site. Fluorogold (FG) was injected into the epicenter of the lesion cavity to identify transected corticospinal tract (CST) neurons. At five weeks after transplantation, the animals were perfused. Locomotor recovery improvement was observed for the BDNF-hMSC group, but not in the hMSC group. Structurally there was increased sprouting of injured corticospinal tract and serotonergic projections after hMSC and BDNF-hMSC transplantation. Moreover, an increased number of serotonergic fibers was observed in spinal gray matter including the ventral horn at and below the level of the lesion, indicating increased innervation in the terminal regions of a descending projection important for locomotion. Stereological quantification was performed on the brains to determine neuronal density in primary motor (M1) cortex. The number of FG backfilled cells demonstrated an increased cell survival of CST neurons in M1 cortex in both of the hMSC and BDNF-hMSC groups at 5 weeks, but the BDNF-hMSC was greater. These results indicate that transplantation of hMSCs hypersecreting BDNF results in structural changes in brain and spinal cord, which are associated with improved functional outcome in acute SCI.
Transplantation of bone marrow-derived cells into spinal cord injury (SCI) models has been reported to promote axonal regeneration, reduce lesion size and improve functional outcome (Chopp et al., 2000; Hofstetter et al., 2002; Ankeny et al., 2004; Kamada et al., 2005; Cízková et al., 2006; Himes et al., 2006: Shichinohe et al., 2008; Someya et al., 2008; Zurita et al., 2008). The precise cell type within bone marrow responsible for these beneficial effects is not fully established, but is thought to reside within the marrow stromal or mesenchymal stem cell (MSC) population (Li et al., 2002; Iihoshi et al., 2004). MSCs can be isolated and expanded as plastic adherent cells having a flattened fibroblast-like morphology (Friedenstein, 1976; Woodbury et al., 2000) that are CD34−, CD45−, SH2+, and SH3+ (Majumdar et al., 1998; Kobune et al., 2003). MSCs have been suggested to differentiate in culture into osteoblasts, chondrocytes, adipocytes and hepatocytes (Prockop et al., 1997; Pittenger et al., 1999; Sanchez-Ramos et al., 2000; Krause et al., 2001; Kobune et al., 2003) and neuronal and glial lineage (Azizi et al., 1998; Kopen et al., 1999; Brazelton et al., 2000; Sanchez-Ramos et al., 2000; Woodbury et al., 2000; Iihoshi et al., 2004). However, current thinking is that the potential beneficial effect of MSCs in various models of central nervous system (CNS) injury is not from neuronal or glial differentiation, but from release of trophic factors which can provide for neuroprotection (Chen et al., 2002; Parr et al., 2007), induction of axonal sprouting (Shen et al., 2006), neovascularization (Onda et al., 2008) and immunomodulation (Ohtaki et al., 2008; Bai et al., 2009). The release of trophic factors from transplanted hMSCs or stimulation of their endogenous production within the host spinal cord may contribute to the recovery of function following SCI (Chen et al., 2002; Lu et al., 2005; Koda et al., 2007).
Isolated cultured bone marrow-derived MSCs can secrete trophic factors (Hamano et al., 2000; Chen et al., 2002; Iihoshi et al., 2004; Kurozumi et al., 2004) including brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF), vascular endothelial growth factors (VEGF), and hepatocyte growth factor (HGF). BDNF is of particular interest because it has been shown to promote survival and sprouting of corticospinal tract (CST) neurons after axotomy (Namiki et al., 2000; Hiebert et al., 2002; Zhou and Shine, 2003). BDNF applied to cortex after SCI led to preservation of CST neurons and to collateral sprouting with increased number of contacts with propriospinal interneurons (Vavrek et al., 2006), which are important in functional recovery after SCI by their ability to increase axonal contacts on motor neurons below the spinal cord lesion (Bareyre et al., 2004).
In this study, we transplanted both hMSCs and genetically modified hMSCs that express BDNF (BDNF-hMSCs) into a rat SCI model in order to investigate the impact of cellular delivery of BDNF by hMSCs on functional outcome, axonal regeneration and neuroprotection in acute SCI.
Human bone marrow from healthy adult volunteers was obtained by aspiration from the posterior iliac crest after informed consent was obtained; this study was approved by the institutional review board at Sapporo Medical University, Japan (Kobune et al., 2003). Bone marrow mononuclear cells were isolated, plated in 150-cm2 plastic tissue culture flasks and incubated overnight. After washing away the free cells, the adherent cells were cultured in mesenchymal stem cell basal medium (MSCBM, Cambrex, Walkersville, MD, USA) containing mesenchymal cell growth supplement (MCGS, Cambrex, Walkersville, MD, USA) with 4mM l-glutamine, and kept in a humidified atmosphere of 5% CO2 at 37°C. After reaching confluency, they were harvested and cryopreserved as primary MSCs or used for gene transduction.
Adenoviral vectors carrying a human BDNF cDNA were constructed as described previously (Kurozumi et al., 2004; Nomura et al., 2005). Briefly, human BDNF cDNA was cloned using the reverse-transcription polymerase chain reaction (RT-PCR) method from the total RNA extracted from primary hMSC as the template. The identity of BDNF cDNA obtained in this manner was confirmed by sequencing and comparing it to the GeneBank sequence XM_006027.
The human BDNF primer sequence was forward, 5′-CGGAATTCCACCATGACCATCCTTTTCCTTACTATGGTTA-3′, and reverse, 5′-CCAGATCTATCTTCCCCTTTTAATGGTCAATGTA-3′.
The BDNF cDNA was inserted between the EcoR I site and the BglII site in the pCAcc vector and the resulting plasmid was designated pCAhBDNF. The plasmid pCAhBDNF was digested with ClaI, and the fragment containing the BDNF cDNA expression unit was isolated by agarose gel electrophoresis. The adenoviral BDNF expression vector, pWEAxCAhBDNF-F/RGD, was prepared using LipofectAMINE 2000 (Invitrogen, Tokyo, Japan). Before being used, the above viral vectors were evaluated for their viral concentration and titer, and viral stocks were examined for potential contamination with replication-competent viruses. To determine viral concentration (particle unit (pu)/ml), the viral solution was incubated in 0.1% sodium dodecyl sulfate and A260 was measured. The viral titers of AxCAhBDNF-F/RGD were 1.0×1012 pu/ml.
Adenovirus-mediated gene transfection was performed as previously described (Nomura et al., 2005). Briefly, the cells were seeded at a density of 2×106 cells per 15cm plate. MSCs were exposed to the infectious viral particles in 7.5ml DMEM at 37°C for 60 min; cells were infected with AxCAhBDNF-F/RGD at a multiplicity of infection (MOI) of 3.0×103 pu/cell. The medium was then removed, the cells washed once with DMEM and then recultured with normal medium for 12 hours; after which transplantation was performed. Commercial BDNF ELISA kits (Promega, Madison, WI, USA) were used to quantify the concentration of BDNF in the supernatant of cultured BDNF-hMSCs with an MOI of 3.0×103 pu/cell were at a rate of 4.73±0.59 ng/105 cells/48 hours (Nomura et al., 2005).
Flow cytometric analysis of primary hMSCs and BDNF-hMSCs were performed as described previously (Honma et al., 2006; Liu et al., 2006; Onda et al., 2008). Briefly, cell suspensions were washed twice with phosphate-buffered saline containing 0.1% bovine serum albumin. For direct assays, aliquots of cells at a concentration of 1 ×106 cells/mL were immunolabeled at 4°C for 30 min with the following anti-human antibodies: fluorescein isothiocyanate (FITC) conjugated CD45, CD34, CD105 (Immunotech, Marseilles, France), and anti-human CD73 (SH-3; BD Biosciences Pharmingen, San Diego, CA, USA). As an isotype-matched control, mouse immunoglobulin G1-FITC (IgG1-FITC; Immunotech) was used. Labeled cells were analyzed by a FACSCalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ) with the use of CellQuest software. Dead cells were gated out with forward- versus side-scatter window and propidium iodide staining.
Cells were pre-labeled with the membrane dye PKH26 according to the manufacturer's instructions (Sigma-Aldrich, St. Louis, MO). The efficiency of cellular labeling (routinely 100%) was examined with a fluorescence microscope.
Experiments were carried out in accordance with National Institutes of Health (NIH) Guidelines for the Care and Use of Laboratory Animals, and the VA Connecticut Healthcare System Institutional Animal Care and Use Committee (IACUC) approved all animal protocols. Adult female Sprague Dawley rats (n = 66) (150-179 gm) were anesthetized with ketamine/xylazine (90/4 mg/kg, i.p.). Under sterile technique, a T9 laminectomy was performed, and dura was opened to expose the spinal cord. The dorsal vein of the spinal cord was coagulated, and the dorsal funiculus was transected using an ophthalmic microscalpel (P-715; Feather Safety Co., Osaka, Japan) (Sasaki et al., 2004). The microscalpel blade was marked for a 1.0 mm depth of cut across the entire dorsal region of the spinal cord. This lesion shows consistent damage to the spinal cord from animal to animal as observed with plastic embedded semi-thin toluidine blue sections (Sasaki et al. 2004). After the cell injection, Gelfoam (Pharmacia and Upjohn, Kalamazoo, MI) impregnated with 5 μl of Fluorogold (FG; 4% w/v in saline, pH 7.4; Molecular Probes, Eugene, OR), was placed into the epicenter of the lesion cavity (Hains et al., 2003; Sasaki et al., 2006). The overlying muscles and skin were closed in layers with 4-0 nylon sutures, and the animal was allowed to recover on a 37°C heating pad. This surgery resulted in paresis of the hindlimbs in all animals subjected to the procedure but did not impair eating, drinking, or elimination.
Immediately after transection of the dorsal funiculus, the cells were injected into the dorsal funiculus using a drawn glass micropipette (Sasaki et al., 2006). Two injections were made at approximately 0.5 mm rostral and two at 0.5 mm caudal to the lesion at depths of 0.7 and 0.4 mm (1.0μl per site; 3.0 × 104 cells/μl for a total of 1.2 × 105 cells transplanted per rat). Transplant-receiving rats and media (DMEM) injected rats were immunosuppressed with Cyclosporin A (10 mg/kg/day) beginning one day before surgery and continued daily for the survival duration. Experimental condition was as follows: PKH26-labelled hMSCs (n = 10), PKH26-labelled BDNF-hMSCs (n = 10), media (DMEM) (n=10), FG+hMSCs (n = 6), FG+BDNF-hMSCs (n = 6), FG+DMEM (n=6), hMSC (n=6), BDNF-hMSC (n=6), DMEM (n=6).
Behavioral analysis was performed (n = 16 animals/group). Intact animals (n=6) were tested for control. Preoperative testing began 2 days before injury and was performed at 3 days and weekly for 5 weeks after surgery. Locomotor function was recorded by a blinded observer using the Basso, Beattie, and Bresnahan (BBB) locomotor rating scale (Basso et al., 1995) to ensure reliability of hind limb somatosensory testing and to assess treatment outcome. Values displayed are means ± SEM.
Another subset of locomotor function was assessed for their sensory-motor ability using a motorized Rotorod system and Rotamex software package, version 1.32A (Columbus Instruments, Columbus, OH). Performance was then tested at 5 weeks post-surgery. The rod (diameter, 7.0 cm) was accelerated at 0.1 rpm/s and increased in speed from 5 to 30 rotations per min (RPM) at a constant rate of acceleration over 3 min. Scores (maximum RPM achieved, total run time) were averaged over three trials per session (Tan et al., 2008).
Animals were perfused transcardially with 0.9% saline, followed by 4% paraformaldehyde in 0.14 M Sorensen's buffer; brain and spinal cord tissue were cryoprotected overnight at 4°C in 30% sucrose. Sixteen micrometer longitudinal cryosections of the spinal cord were cut and mounted on Sigma Silane-Prep glass slides, and the sections were processed for immunolabeling for monoclonal TUJ-1 (1/100; Millipore), monoclonal GFAP (1/1000; Sigma) and polyclonal NG2 (1/200; Millipore). Secondary antibodies used were as follows: goat anti-mouse IgG-Alexa Fluor 488 (1:1000; Invitrogen, Eugene, OR), goat anti-rabbit IgG-Alexa Fluor 488 (1:1000; Invitrogen, Eugene, OR) and mounted with Vectashield mounting medium for fluorescence with DAPI (4′,6-diamidino-2-phenylindole; Vector Labs). The sections were examined under a fluorescence scope equipped with confocal lasers (Zeiss LSM510, Thornwood, NY). Images were captured using LSM 510 software (Zeiss, Thornwood, NY) and arranged using CorelDRAW 12 (Corel Corporation, Mountain View, CA) and Photoshop (Adobe Systems, San Jose, CA).
The spinal cord (n=5/group) 5-10mm rostral to and 5-10mm caudal to the lesion center was embedded in OCT. Five millimeters length blocks of spinal cord from T7 and T8 were frozen in OCT (Triangle Biomedical Sciences) and 160 serial sections (8μm thick) in the coronal plane were cut using a 2800 Frigocut (GMI, inc.; Ramsey MT) cryostat and thaw mounted onto gelatinized glass slides. The sections were washed in blocking solution (potassium phosphate buffered saline (KPBS), 0.1% Triton X-100, 4% normal donkey serum) and incubated overnight at 4°C in rabbit anti-protein kinase C-gamma (PKC-γ) (1:500; Santa Cruz Biotechnology, Santa Cruz, CA) diluted in the same blocking solution. After washing in blocking solution, sections were incubated for 1 hour in donkey anti-rabbit Alexafluor488 (1:2000; Invitrogen, Carlsbad, CA) at room temperature. Slides were washed in clean KPBS and cover slipped with Aqua Polymount mounting solution (Polysciences Inc.; Warrington, PA). Sections were visualized with a Nikon Eclipse E800 microscope with a HQ Coolsnap camera (Roper Scientific; Tucson, Arizona) (Barritt et al., 2006).
Image analysis was performed similar to Barritt et al. (2006) (Barritt et al., 2006). Four coronal sections were selected for analysis. MetaVue software (Scanalytics, Fairfax, VA) was used to calculate relative intensity of PKC-γ immune-positive signal intensity within three bins (300 μm × 200 μm). These bins were superimposed over the dorsal columns avoiding the superficial gray matter of the dorsal horn. Using unstained tissue to normalize for background signals, digital images were processed into binary images through a threshold algorithm across groups. The mean pixel density within each bin was calculated within each group and expressed as a percentage of fields (pixel area/total bin area). Each bin was statistically compared against the control treated group.
The spinal cord (n=5/group) 5-10mm rostral to and 5-10mm caudal to the lesion center was embedded in a glutaraldehyde-polymerized albumin matrix and cut parasagittally in the thickness of 40μm on a vibratome. Transverse sections (40μm) were collected from the spinal cord 11 to 16mm rostral to and 11 to16mm caudal to the lesion center. For analysis of serotonin innervation, the sections were incubated with avidin-biotin-peroxidase complex plus anti-5-hydroxytryptamine (anti-5-HT) antibody (1:10,000; Immunostar, Hudson, WI) to detect raphespinal fibers (Wang et al., 2006). Immunoreactive serotonin fibers in the ventral horn of transverse sections rostral or caudal to the lesion center were photographed using conventional fluorescence microscopy (Nikon Eclipse 800; Spot RT Color CCD camera; Diagnostic Instruments). Immunoreactivity was quantified using NIH Scion Image software to count the number of immunopositive pixels set above a threshold to selectively detect serotonin immunoreactivity. The total number of immunoreactive pixels was averaged to represent the 5-HT immunoreactivity (5-HT-IR) for each animal (Lee et al., 2007).
Stereologic counting methods were used to obtain an accurate estimate of the number of FG-labeled backfilled neurons after DC lesion (n = 6 animals/group) (Smolen et al., 1983; Mori et al., 1997; Kwon et al., 2002; Hains et al., 2003; Sasaki et al., 2006). Serial coronal sections (20μm) were made through the M1 cortex, and every tenth section was saved and digitized spanning bregma -2.0 to 2.0. Only cells that displayed prominent nuclear profiles were scored. The total number of FG-positive cell profiles (Ntotal) was estimated by the following formula:
where section thickness t = 20μm, and f = 10. The calculated distance from one disector to the next was 200μm. Qi = counted cell profiles in the uniformly sampled disectors (crude number); n = 10 (number of equidistant sections used in the analysis). The number of cells in each of the 1 to i th-sampled sections was used to calculate the coefficient of error (Gundersen et al., 1988; Abusaad et al., 1999; Hains et al., 2003; Sasaki et al., 2006).
Comparison against control group was performed using a one-way ANOVA with Tukey's test post hoc. Data management and statistical analyses were performed using SigmaStat (version 3.0.1a; Jandel Scientific, Corte Madera, CA) and Microsoft Office Excel (2003) and graphed as mean ± SEM using SigmaPlot (version 8.02a).
Primary hMSCs were cultured as plastic adherent cells to subconfluency, which took approximately 1 week (see Materials and Methods). Both hMSCs (Fig.1A) and BDNF-hMSCs (Fig.1B) were pre-labeled with PKH26 to identify the transplanted cells in vivo. Superimposed Nomarski light microscopic and fluorescent images of PKH26 labeled cells nuclear stained with DAPI are shown in Fig. 1A (hMSCs) and B (BDNF-hMSCs), respectively. PKH26 is a fluorescent lipophilic dye whose detection does not require immunostaining. PKH26 was localized in the cytoplasm of the characteristic flattened hMSCs and BDNF-hMSCs in vitro (Fig. 1A, B). Flow cytometric analysis of the BDNF-hMSCs (Fig. 1C) was essentially identical to primary hMSCs (Fig. 1D). Both had CD34-, CD45-, SH2+ (CD105), and SH3+ (CD73) cell surface phenotype. These data are in agreement with our previous work (Kobune et al., 2003; Kurozumi et al., 2004; Nomura et al., 2005).
The transplanted PKH26 labeled BDNF-hMSCs and hMSCs were easily identified within the lesion zone by their red fluorescence at both 1 week and 5 weeks after transplantation. BDNF-hMSC distribution in the injured SCI at 1 and 5 weeks are shown in Supplemental Figure 1. At 1 week, the transplanted BDNF-hMSCs were confined largely to the lesion zone (Supple. Fig. 1A). At 5 weeks, the BDNF-hMSCs survived in the lesion zone, but a number of cells distributed outside of the lesion throughout the spinal cord and in the meningeal region (Supple. Fig. 1B). The distribution pattern was similar for hMSCs (not shown). Cytoplasmic PKH26 localization in the perinuclear region of the hMSCs was also observed in vivo at 5 weeks after transplantation indicating cell survival (Supple. Fig. 1C, D).
Immunostaining of spinal cord sections five weeks after transplantation of hMSCs and BDNF-hMSCs was carried out for GFAP, TUJ1 and NG2. The transplanted cells were identified by the distribution of PKH26 in the cytoplasmic regions outside of the nucleus (Fig. 2A, E, I). There was no obvious co-localization of GFAP (Fig. 2A-D), TUJ1 (Fig. 2E-H) or NG2 (Fig. 2I-L) with PKH26 identified hMSCs. The hMSCs were embedded in areas rich in glial and neuronal processes, but did not differentiate into glial or neuronal elements.
Open-field locomotor scores for BDNF-hMSC, hMSC and DMEM injection groups (n = 16/group) were tested 1 week before and 3 days, 1, 2, 3, 4, and 5 weeks after transplantation (Fig. 3A). Intact animals (n=6) scored 21 on the BBB test. All experimental animals exhibited a gradual improvement in hindlimb locomotor function during the 5-week recovery period. The BDNF-hMSC transplant group recovered to near 20 on the BBB score and displayed consistent weight-bearing plantar stepping behavior. Statistical analysis indicated that the BBB scores of BDNF-hMSC transplantation were higher than the hMSC and DMEM groups at 4 and 5 weeks after transplantation, respectively (Fig. 3A). Although a trend was observed, a difference between the hMSC and the DMEM groups at 5 weeks did not reach statistical significance (p=0.775 with Tukey post hoc test).
Rotation rates (maximal RPM scores; rpm) and duration (total run time; sec) were evaluated for the BDNF-hMSC, hMSC and DMEM control groups with Rotorod testing at 5 weeks after transplantation (Fig. 3B, C). Intact animals (n=6) were able to stay on the Rotorod at a rotation of 21.6 ± 2.5 rpm, and duration of 242.7 ± 25.9 sec. In the DMEM group, rotation rates of 6.79 ± 0.89 rpm and duration of 67.11 ± 13.65 sec were observed. Rotation rates and duration were greater in the BDNF-hMSC group (12.0 ± 1.26 rpm; 116.16 ± 13.69 sec), but not in the hMSC group (9.7 ± 1.37 rpm; 95.2 ± 12.5 sec) as compared to the DMEM group (p<0.05). As with the BBB test, a trend in the hMSC group was observed. Error bars represent mean ± SEM.
Image analysis on tissue with PKC-γ immunostaining was performed to identify CST axons using similar methods as described previously (Barritt et al., 2006). Three bins were superimposed over the dorsal funiculus from coronal tissue sections sampled from rostral and caudal to the SCI site. Levels of PKC-γ immuno-positive fibers were compared across the three experimental groups (Fig. 4M, N). In the DMEM (sham control) group, PKC-γ immuno-positive reactivity rostral to the lesion revealed dorsal CST fibers in the ventral portion of the dorsal funiculus (Fig4A, B) and small diameter interneurons in the superficial dorsal horn (Fig. 4A). This pattern of PKC-γ immunostaining was also consistent for tissue sections examined from animals within the BDNF-hMSC (Fig. 4I, J), or hMSC (Fig. 4E, F) groups. Five-weeks after BDNF-hMSC transplantation, more PKC-γ positive CST fibers appeared to sprout superficially into the intermediate region of the rostral dorsal funiculus (Fig. 4I, J, M) compared to the DMEM group. Compared to the DMEM group, the hMSC group also had increased PKC-γ immunoreactivity in the ventral region. Image analysis of PKC-γ immunostaining in caudal spinal cord tissue sections located below the injury site demonstrated more dramatic effects of the transplant groups on CST sprouting (Fig. 4D, H, L). Comparison across experimental groups in the ventral region showed significantly more PKC-γ fibers in the BDNF-hMSC transplant group (Fig. 4K, L, N) compared to the DMEM group (Fig. 4C, D). In addition, more CST fibers were found in this ventral region with BDNF-hMSC transplants than with the hMSC group (Fig. 4G, H, N). More PKC-γ immuno-positive CST fibers sprouted into the dorsal region bin in the BDNF-hMSC group compared to both the hMSC and the DMEM groups. Error bars represent mean ± SEM.
5-HT fiber growth was increased after BDNF-hMSC transplantation after SCI. Transverse sections 11 to 15mm rostral to the lesion center of spinal cord stained with anti-serotonin antibodies from the DMEM (Fig.5A), hMSC (Fig. 5B) and BDNF-hMSC (Fig. 5C) groups showed similar numbers of serotonergic fibers in the ventral horn (Fig. 5G). Transverse sections of spinal cord 11 to 15mm caudal to the lesion center showed a significantly greater 5-HT fiber immunoreactivity of serotonergic fibers in the ventral horn of BDNF-hMSC (Fig. 5F) group compared with the DMEM group (Fig. 5D). Quantification of the optical density of the 5-HT immunoreactivity of serotonergic fiber in the ventral horn rostral and caudal to the SCI lesion indicate that there is a significant increase of serotonergic fibers in the ventral horn caudal (Fig. 5H) to the lesion in the BDNF-hMSC group, compared to the DMEM group, but no statistical difference rostral to the lesion (Fig. 5I).
Bilateral transection of the dorsal CST and FG injection at T9 results in robust retrograde labeling of the cell bodies of these axons within layer V of the primary motor cortex (M1) (Fig. 6); which spans from approximately bregma 1.6 through -1.6 (Hains et al., 2003; Sasaki et al., 2006). In the coronal plane, the anatomical location of labeled cells was within a region lateral to the midline longitudinal fissure and medial to the forelimb region of the primary sensory cortex (S1) and within cortical layer V. In the sagittal plane, the anterior distribution of FG-labeled cells extended to the rostral margin of the anterior commissure and extended caudally to the posterior margin of the optic chiasm. Retrogradely labeled cells displayed pyramidal-shaped somata, which give rise to a singular apical dendrite and smaller basal dendrites. In the rostrocaudal dimension, the density of backfilled CST neurons was highest at approximately bregma -0.3. At 5 weeks after SCI, quantification of the number of FG-positive neurons was significantly higher in both the BDNF-hMSC and hMSC groups (p<0.01) as compared to the DMEM group, but the BDNF-hMSC group was greater. Figure 6 (A-C) shows corresponding sections through bregma -0.3 of the three experimental groups. Stereological analysis indicates that the estimated total neuronal counts among DMEM, hMSC, and BDNF-hMSC groups are 675 ± 44.2, 798± 29.3, and 909± 34.8 at 5 weeks, respectively (Fig. 6D).
Bone marrow-derived cells (mononuclear or marrow stromal cells) from experimental animal sources transplanted into SCI models have demonstrated variable degrees of functional improvement (Chopp et al., 2000; Hofstetter et al., 2002; Ankeny et al., 2004; Kamada et al., 2005; Cízková et al., 2006; Himes et al., 2006: Shichinohe et al., 2008; Someya et al., 2008; Zurita et al., 2008). In the present study, we prepared highly purified and identified MSCs (Kobune et al., 2003) from human bone marrow to study their impact on SCI. The hMSCs increased axonal sprouting and increased the number of surviving M1 cortical neurons, but did not lead to improved functional outcome with our testing paradigms although a trend was observed. Several studies have demonstrated that BDNF delivery is protective to injured CST neurons and can improve functional outcome after SCI (Jakeman et al., 1998; Namiki et al., 2000; Kim et al., 2004; Kwon et al., 2007). We previously observed an increase in BDNF levels in the SCI transplantation zone after olfactory ensheathing cell (OEC) transplantation and an increased survival of M1 CST neurons and improved functional outcome suggesting a contribution of BDNF (Sasaki et al., 2006). Moreover, genetically modified hMSCs that express BDNF have additive neuroprotective effects and reduce lesion volume in a cerebral infarction model as compared to hMSCs alone (Kurozumi et al., 2004; Nomura et al., 2005). Therefore, we transplanted BDNF-hMSCs into a SCI model to determine if these cells would enhance functional recovery over hMSCs alone because of the additive effect of BDNF delivery in this study. Transplantation of BDNF-hMSCs increased axonal sprouting and M1 cortical neuronal survival and led to a statistically significant improvement in functional outcome. Thus, the cellular delivery of BDNF can positively impact functional outcome in SCI. The comparison of hMSCs alone and BDNF-hMSCs provides more direct evidence for the role of BDNF in enhancing this cell therapy approach for experimental SCI.
Both transplanted hMSCs and BDNF-hMSCs distributed in the lesion site at 1 week and survived in the spinal cord for the 5-week study period, suggesting that the difference in functional outcome between the two groups was not the result of a difference in cell survival. Although only a trend in improvement in functional outcome was observed with hMSCs alone, the effects of hMSCs alone on increasing cortical neuronal survival and increasing sprouting should not be discounted. MSCs secrete a variety of bioactive substances such as neurotrophins (including BDNF), interleukins, macrophage colony-stimulating factor, Flt-3 ligand, and stem-cell factors that could contribute to these changes (Eaves et al., 1991; Majumdar et al., 1998).
PKC-γ co-localizes with CST fibers and small diameter neurons located in the substantia gelatinosa of the adult rat spinal cord (Akinori et al., 1998). Because of its expression pattern in the spinal cord motor system, specifically in rodent species, PKC-γ has been shown to be useful for investigating CST sprouting and plasticity in SCI models (Bradbury et al., 2002; Barritt et al., 2006; Cafferty and Strittmatter, 2006). Transplantation of hMSCs and BDNF-hMSCs promoted sprouting of PKC-γ immuno-positive CST fibers both above and below the SCI site. hMSC and BDNF-hMSC transplantation resulted in CST fiber sprouting into more superficial regions (ventral) of the dorsal funiculus in rostral tissue above the injury site. In addition, the BDNF-hMSC group sprouted into the intermediate dorsal funicular region rostral to the lesion, but the hMSC group did not. Therefore, the increased CST fiber sprouting in the BDNF-hMSC group in the rostral dorsal funiculus tissue regions suggests that BDNF induces increased plasticity within the spinal cord. This growth pattern may be explained by either of two ways: first, by the extrinsic chemoattractive properties of BDNF (Nakahara et al., 1996; Ebadi et al., 1997; Bamber et al., 2001), or second, by the intrinsic growth-promoting stimulus that BDNF can provide to subclasses of injured axons (Filbin et al., 2003; Zhou and Shine, 2003). As the impact of BDNF-hMSC has a longitudinal effect, providing impetus to enhance growth bidirectional from the injury site, it may be interesting for future investigation to determine whether BDNF-hMSCs can affect other axonal types (i.e., ascending sensory afferents) after SCI. Interestingly, only the BDNF-hMSC group promoted CST sprouting into the dorsal funiculus caudal to the injury site compared to both the hMSC and DMEM groups. Thus, these results together support the importance of BDNF in reported functional improvement and that BDNF can be delivered by a cell therapy approach.
It is possible that the rostral CST axonal sprouting induced by BDNF-hMSC may contribute to recovery through an enhancement of polysynaptic connections to the caudal spinal cord. Bareyre et al. (2004) demonstrated that the course of endogenous functional recovery is associated with sprouting of CST axons above the lesion that synapse on long propriospinal neurons. In turn, the propriospinal neurons, which project below the lesion, sprout and provide greater synaptic drive to motor neurons below the lesion.
Raphespinal fibers were not completely severed by the dorsolateral transection SCI model. In the BDNF-hMSC group, increased 5-HT fibers were detected in the ventral horns of spinal cord. Several studies suggest the importance of raphespinal inputs to spinal motor neurons in motor performance (Morrison and Gebber, 1985; Wang et al., 2006). Moreover, a number of treatments including application of Nogo inhibitors results in 5-HT axonal sprouting and improved functional outcome in rodent SCI models (Wang et al., 2006). The increased 5-HT fiber sprouting we observed after BDNF-hMSC transplantation may at least partially account for the improvement in functional outcome.
Several groups have reported that CST neurons of M1 cortex undergo atrophy (McBride et al., 1989; Tang et al., 2004; Carter et al., 2008) after SCI, as well as dynamic changes in the density and morphology of dendritic spines of M1 neurons in rat (Kim et al., 2006). In mice, Carter et al. (2008) observed significant atrophy of CST neuron somata after thoracic SCI using a Yellow Fluorescent Protein (YFP-H) transgenic mouse model. In humans, structural abnormalities with reduced tissue volume after SCI by MRI approaches have been demonstrated (Wrigley et al., 2009). Cell loss in M1 cortex after SCI has also been reported (Hains et al., 2003; Lee et al., 2004; Klapka et al., 2005; Sasaki et al., 2006). Klapka et al. (2005) reported functional recovery and increased cell numbers on pyramidal motor neurons in M1 cortex following scar-suppressing treatment with local application of an iron chelator (BPY-DCA) following SCI in rats. In the present study, hMSC and BDNF-hMSC transplants demonstrated the preservation of the number of CST neurons in M1 as compared to the sham control (DMEM group). Rescue of CST neurons by hMSC and BDNF-hMSCs could provide for a greater upper motor neuronal pool that through sprouting above the lesion might enhance new intraspinal circuits to regions below the lesion, e.g. propriospinal (Bareyre et al., 2004). It is important to note that Lu et al. (2005) did not see improved functional outcome after transplantation of BDNF-hypersecreting rat MSCs in a rodent SCI model. One difference between their study and ours was that they injured the spinal cord at C3 and studied forelimb function that requires more precise targeting of regenerated axons to the motor neurons than for hindlimb function. Our lesion was at T9. It is possible that the improvement we report in hindlimb function with BDNF-hMSC was because spinal cord target sites for hindlimb function may not need as extensive and precise reinnervation as would targets for forelimb function.
In summary, the transplantation of hMSCs expressing BDNF promotes functional recovery, sprouting of CST and raphespinal fibers, and protection of CST neurons after SCI. Thus, cellular delivery of BDNF secreted by hMSCs may have a therapeutic effect following some forms of acute SCI.
Montage images of sagittal frozen sections through the injury and BDNF-hMSC transplantation site 1 (A) and 5 (B) weeks after injury. Transplanted PKH26 induced fluorescent BDNF-hMSCs survive and are distributed within the transplantation site at 5 weeks (B). At 5 weeks, the BDNF-hMSCs survived in the lesion zone, but a number of cells distributed outside of the lesion throughout the spinal cord and in the meningeal region (B). (C) was taken from * in (B). Cytoplasmic PKH26 localization was observed in vivo after transplantation; counterstained with DAPI staining for nuclei (C, D). Note: C was taken by epifluorescence microscope and D is taken by confocal microscope. Scale bars = 1mm (A, B), 200μm (C), 50μm (D).
This work was supported in part by the Medical and Rehabilitation and Development Research Services of Department of Veterans Affairs (VA), the VA Spinal Cord Injury Translational Collaborative Consortium for Regenerative Medicine, the National Institutes of Health (NS43432 to J.D.K.), the National Multiple Sclerosis Society (RG2135; CA1009A10 to J.D.K.), the Paralyzed Veterans of America Research Foundation (2469 to M.S.), the Connecticut Stem Cell Research Grants Program (08-SCA-YSME-011 to M.S.), and the Japanese Ministry of Education, Science, Sports and Culture (19591691, 20390388, 20591717 to O.H.). This material is based upon work supported by the State of Connecticut under the Connecticut Stem Cell Research Grants Program. Its content are solely the responsibility of the authors and do not necessarily represent the official views of the State of Connecticut, the Department of Public Health of the State of Connecticut or Connecticut Innovations, Incorporated. The Center for Neuroscience and Regeneration Research is a collaboration of the Paralyzed Veterans of America and the United Spinal Association with Yale University. We thank Heather Mallozzi, Margaret Borelli and Cynthia Santaniello for excellent animal care and technical assistance and Dr. Bryan Hains for contribution in the initial study and Drs. Edgardo Arroyo and Yu-Wen Chang for assistance with confocal microscope.