Limited Functional Effects of Subacute Syngeneic Bone Marrow Stromal Cell Transplantation after Rat Spinal Cord Contusion Injury

Abstract

Cell transplantation might be one means to improve motor, sensory, or autonomic recovery after traumatic spinal cord injury (SCI). Among the different cell types evaluated to date, bone marrow stromal cells (BMSCs) have received considerable interest due to their potential neuroprotective properties. However, uncertainty exists whether the efficacy of BMSCs after intraspinal transplantation justifies an invasive procedure. In the present study, we analyzed the effect of syngeneic BMSC transplantation following a moderate to severe rat spinal cord injury. Adult Fischer 344 rats underwent a T9 contusion injury (200 kDy) followed by grafting of GFP-expressing BMSCs 3 days postinjury. Animals receiving a contusion injury without cellular grafts or an injury followed by grafts of syngeneic GFP-expressing fibroblasts served as control. Eight weeks post-transplantation, BMSC-grafted animals showed only a minor effect in one measure of sensorimotor recovery, no significant differences in tissue sparing, and no changes in the recovery of bladder function compared to both control groups in urodynamic measurements. Both cell types survived in the lesion site with fibroblasts displaying a larger graft volume. Thus, contrary to some reports using allogeneic or xenogeneic transplants, subacute intraparenchymal grafting of syngeneic BMSCs has only a minor effect on functional recovery.

Introduction

Traumatic spinal cord injury (SCI) usually leads to the disruption of descending motor and ascending sensory pathways resulting in sensorimotor and autonomous deficits such as bowel and bladder dysfunction. Subsequent secondary injury mechanisms including cystic cavitation and demyelination are the consequence of a progressive loss of neurons, astrocytes, and oligodendrocytes. Pronounced cellular inflammatory responses characterized by activation of microglia, infiltration of neutrophils and macrophages, and upregulation of proinflammatory cytokines contribute to this progressive cellular degeneration (18,40,61).

During this early phase postinjury, pharmacological or cell transplantation-based approaches that limit cellular and axonal degeneration by influencing inflammatory reactions might be one means to achieve neuroprotection. Transplantation of immunomodulatory cells may not only have neuroprotective effects but may also fill a lesion site to provide a substrate for axonal regeneration. Toward this goal, bone marrow stromal cells (BMSCs) have received particular attention because they can be easily isolated, rapidly expanded, and transplanted as autologous cell grafts, avoiding graft rejection (68).

BMSCs can integrate into the host spinal cord and have been reported to enhance axonal regeneration and functional outcomes by increasing tissue sparing and reducing cavitation and glial scar formation (5,24,26,45,73). Some studies suggest that BMSCs provide a regenerative host microenvironment by secreting neurotrophic factors, cytokines, and anti-inflammatory molecules (14,22,43,51), reducing the levels of stress-associated proteins, reactive oxygen species, or proinflammatory cytokines (6,37,77). Potential immunomodulatory mechanisms further include inhibition of neutrophil infiltration (54) and shifting macrophage differentiation toward alternatively activated M2 macrophages, while preventing the development of classically activated proinflammatory M1 macrophages (6,21,42,54). Moreover, BMSCs appear to exert immunoregulatory effects by inhibiting T-cell recognition and proliferation (1,68,69,72).

While some of these results together with positive effects in animal models of SCI (22,24,26,42,45) seem to justify the clinical translation of BMSC transplantation in SCI, the interpretation of these potentially promising data is complicated by differences in the source and number of transplanted cells, the time point of cell transplantation, animal models of SCI, limited data on the survival of transplanted cells, or a lack of adequate outcome measures. Indeed, other studies have shown only small or no improvements in functional recovery comparing animals that received grafts of BMSCs after SCI with control animals that received PBS or media (vehicle) injections (5,13,19,26,50,59,64). These discrepancies indicate that more investigations are required to identify the cellular fate of transplanted BMSCs over time and to further elucidate the mechanism of potential neurological recovery. In addition, virtually all studies conducted to date lack an appropriate cellular control, comparing BMSC-grafted animals only to control animals receiving vehicle injections.

The purpose of the present study was therefore to investigate the morphological and functional effects of BMSC grafts in a clinically relevant rat spinal cord contusion injury. Using a study design with two control groups, SCI without treatment and SCI followed by fibroblast grafting, we aimed to exclude the possibility that just an injection of any cell will have similar effects. Furthermore, based on some previous studies and for clinical feasibility and relevance, we chose to transplant cells with a delay of 3 days postinjury (42,44,56). Potential influences of BMSCs on the recovery of bladder function in addition to sensorimotor recovery and tissue sparing were also examined. Our results indicate only a limited if any beneficial effect of intraparanchymally grafted BMSCs compared to both control groups despite the survival of cellular grafts.

Materials and Methods

Animal Subjects

Adult female Fischer 344 rats (Charles River Deutschland GmbH, Sulzfeld, Germany) weighing 160-180 g were used for all in vivo experiments. BMSCs and fibroblasts were isolated from transgenic Fischer 344-Tg (EGFP) rats ubiquitously expressing green fluorescent protein (GFP) under the ubiquitin C promoter (Rat Resource and Research Center, University of Missouri, Columbia, MO, USA). Experiments were carried out in accordance with European Commission Council Directives and institutional guidelines. Animals had ad libitum access to food and water throughout the study.

Isolation of BMSCs and Fibroblasts

For cell isolation, GFP-transgenic Fischer 344 rats were deeply anesthetized using a cocktail of ketamine (62.5 mg/kg; Medistar, Ascheberg Germany), xylazine (3.175 mg/kg; Bayer, Leverkusen, Germany), and acepromazine (0.625 mg/kg; Sanofi-Ceva, Düsseldorf, Germany) in 0.9% sterile saline solution and killed by decapitation. For isolation of BMSCs, bone marrow was harvested from femurs and tibias of 2- to 4-month-old female rats. Cells were mechanically dissociated in minimum essential medium α (α-MEM; Merck Millipore, Darmstadt, Germany) and recovered by centrifugation. Cell pellets were resuspended in α-MEM containing 10% fetal bovine serum (α-MEM-10% FBS; PAA/ GE Healthcare, Freiburg, Germany) and 100 U/ml penicillin/0.1 mg/L streptomycin (Invitrogen, Germany). BMSCs were seeded at 1 × 10⁶ cells/cm² in a humidified incubator at 37°C with 5% CO₂. After 3 days, the medium was changed, and nonadherent cells were discarded. Adherent cells were incubated in fresh α-MEM-10% FBS until cells were confluent. Cells were trypsinized using 0.25% trypsin (Life Technologies, Darmstadt, Germany), seeded in α-MEM-10% FBS at 8,000 cells/cm². This cell culture preparation is highly enriched in BMSCs with virtually no hematopoietic contamination (57) as demonstrated by flow cytometry and immunocytochemistry (see below).

Primary fibroblast cultures were generated from skin biopsies of GFP-transgenic Fischer 344 rats and cultivated under standard culture conditions as previously described (67). Briefly, a small skin biopsy from the depilated abdominal region was taken. The biopsy was put into 70% ethanol for a few seconds, washed in HBSS (Merck Millipore), and all fat tissue was removed. The biopsy was then cut into small pieces (1 × 1 mm) and transferred in culture wells containing Dulbecco's minimal essential medium (DMEM + L-glucose; Life Technologies) supplemented with 2 mM L-glutamine (Life Technologies), 100 U/ml penicillin/0.1 mg/L streptomycin (Life Technologies), and 10% FBS. The medium was changed twice a week, and at 90% confluence, cells were passaged by trypsinization and reseeded at 5 × 10⁴ cells/ml.

Flow Cytometry

BMSCs were characterized by flow cytometry as previously described (17,63); 1 × 10⁵ cells were labeled with mouse anti-CD90-FITC (1:25; Abcam, Cambridge, UK), mouse anti-CD68 (1:100; Abcam), and mouse anti-CD45 (1: 250; Abcam), respectively. Cells were incubated with donkey anti-mouse Alexa Fluor 594 secondary antibody (1:100; Life Technologies). Cells were quantified with a FACSCalibur (BD, Heidelberg, Germany).

In Vitro Immunocytochemistry

Parallel to the flow cytometry experiments, BMSCs were plated in 24-well plates at a density of 6,000 cells/ ml. After 24 h, cells were fixed for 30 min with phosphate-buffered 4% paraformaldehyde (37°C, pH 7.4) and processed for immunofluorescence labeling. Fixed cells were washed in Tris-buffered saline (TBS, 0.15 M NaCl, 0.1 M Tris-HCl, pH 7.5), blocked with TBS/5% donkey serum (GE Healthcare)/0.1% Triton X-100 (only for CD68) for 30 min. The following primary antibodies were applied in the same solution overnight at 4°C: mouse anti-CD90-FITC for nonhematopoetic cells (1:25; Abcam), mouse anti-CD68 for microglia/macrophages (1:2,000; AbD Serotec, Germany), and mouse anti-CD45 for leucocytes (1:1,000; Merck Millipore). Donkey anti-mouse conjugated to Alexa Fluor 594 (1:300; Life Technologies) antibodies were used for immunodetection. A nuclear counterstain was performed with 4′,6′-diamidino-2-phenylindole dihydrochloride hydrate (DAPI) at 0.25 μg/ ml (Sigma-Aldrich, Schnelldorf, Germany). Images were taken on an IX81 Olympus inverted fluorescent microscope (Olympus, Hamburg, Germany) equipped with a XC camera.

Surgical Procedures and Cell Transplantation

For all surgical procedures, animals were anesthetized using a cocktail of ketamine, xylazine, and acepromazine as described above. A total of 38 rats received a spinal cord contusion injury at midthoracic level T9 using the Infinite Horizon (IH) Impactor SCI device (Precision Systems & Instrumentation, Lexington, KY, USA) with an impact force of 200 kilodynes (kdyn) as previously described (60). Postoperatively, animals were kept warm, and buprenorphine (0.03 mg/kg; Reckitt Benckiser, Mannheim, Germany) and ampicillin (167 mg/kg; Ratio-pharm, Ulm, Germany) were given subcutaneously. Bupre-norphine was given for 2 days postoperatively twice a day, and ampicillin was given twice a day as long as manual bladder evacuation was necessary. Manual bladder evacuation was performed twice a day until reflex bladder function returned. Based on previous studies showing enhanced tissue sparing and cell survival with delayed transplantation at 3 days postinjury (42,44,56), the lesion site was reexposed 3 days postcontusion injury in animals receiving cell grafts. The center of the lesion was identified by a slight difference in color/opacity from the surrounding tissue. Cell suspensions (5 μl) containing either BMSCs (10⁵ cells/μl; n = 12) or fibroblast (10⁵ cells/ μl; n = 13) were injected at a single site into the center of the lesion at a maximum depth of 1.5 mm using a pulled glass pipette (100 μm internal diameter) and a PicoSpritzer II (General Valve, Fairfield, NJ, USA). A similar number of cells (2.5 × 10⁵-10⁶ cells) was used in several previously published reports (5,12,26,42,45,73,78). Pipettes were slowly raised during cell injections and left in place for 1 min before being completely withdrawn. Control animals (n = 13) received spinal cord contusions but no cell injections. Animals were killed after 1 and 8 weeks posttransplantation (Table 1).

Table 1.

Animal Numbers and Groups Used for Morphological, Behavioral, and Urodynamic Analyses

Surgery	Survival	BBB	Grid Walk	Urodynamic	Total
Contusion	1 week (n = 5)		—	—	n = 13
	8 weeks (n = 8)	n = 8	n = 7	n = 6
BMSCs	1 week (n = 4)		—	—	n = 12
	8 weeks (n = 8)	n = 8	n = 8	n = 8
Fibroblast	1 week (n = 5)		—	—	n = 13
	8 weeks (n = 8)	n = 8	n = 8	n = 7
Naive	8 weeks	–	–	n = 7	n = 7

BMSCs, bone marrow stromal cells; BBB, Basso, Beattie, Bresnahan open field locomotor rating scale.

Bladder Catheterization

For cystometry, the bladder was catheterized 3 to 4 days prior to urodynamic measurements as previously described (4). Briefly, animals were anesthetized as described above, and the bladder was exposed by a midline incision of the low abdominal wall. A polyethylene tubing (diameter 2.1 mm; PE50; Schubert Medizinprodukte, Wackersdorf, Germany) with a cuff was inserted into the bladder dome and secured by a purse-string suture. The catheter was tunneled subcutaneously and externalized through the skin in the back of the neck. The external tip of the catheter was closed thermally to prevent leakage and infection and sutured to the skin until urodynamic measurements. After catheterization, ampicillin (167 mg/kg) was given daily until measurements were performed.

Cystometry

A total of 21 animals underwent urodynamic measurements 8.5 weeks postinjury as described previously (4). Naive uninjured rats (n = 7) served as controls. The location and intactness of the catheter was verified before cystometry. Awake, unrestrained rats were placed in a metabolic cage (Tecniplast, Buguggiate, Italy) and allowed to accommodate for about 10-15 min. To measure repeated micturition cycles, the bladder was infused via the implanted catheter with saline for 30—45 min at a rate of 6 ml/h and bladder pressure and urine volume were synchronously and continuously recorded.

For measuring micturition volume, released fluid was collected through a funnel into a cup connected to a force transducer (MLT1030/D Wide Range Force Transducer; ADInstruments, Oxford, UK). For pressure measurements, the bladder catheter was connected via a three-way stopcock to a pressure transducer (MLT0699 Disposable BP Transducer; ADInstruments) and to a syringe pump (KR Analytical, Cheshire, UK). Both transducers were connected to an amplifier and a data acquisition system (PowerLab 8/35; ADInstruments). The tracings were followed during the measurements and further analyzed using LabChart software (ADInstruments).

The following parameters were investigated: micturition volume (MV), micturition frequency (MF per hour), basal pressure (P_base: minimum pressure between two micturition events), maximum voiding pressure (P_max: maximum bladder pressure during a micturition cycle), and pressure difference (P_diff: P_max - P_base).

Three rats had to be excluded due to damage to the catheter (n=2) or severe spasticity during the measurement (n = 1) preventing the generation of interpretable data.

Behavioral Assessment

Hindlimb motor function was evaluated according to the Basso, Beattie, Bresnahan (BBB) open field locomotor rating scale (7). BBB scores range from 0 to 21 (0 = no observable hindlimb movement/21 = normal locomotion). All animals had a score of 0 at 1 day postinjury. Animals were scored once a week starting 10 days postlesion by two independent observers blinded to group identity.

To examine differences in sensorimotor integration, the horizontal ladder walking test was performed. Animals were first trained to cross a 100-cm-long horizontal ladder with unevenly spaced rungs toward the home cage. The test was carried out 9 days before the animals were killed. The animals were recorded using a digital video camera, and performance was subsequently analyzed in slow motion. Total steps and the number of footfalls below the plane of the ladder (failure to grasp a rung with their hindlimbs) were quantified in five uninterrupted crossings. The number of missteps per trial was counted, given as a percentile of the total number of steps, and averaged for five trials, by a blinded observer. One rat had to be excluded because it showed severe spasticity and was therefore unable to cross the ladder without stopping. Naive animals show less than 10% missteps in this task.

Tissue Processing and Immunohistochemistry

Animals were transcardially perfused with 0.9% saline solution followed by 4% paraformaldehyde in 0.1 M phosphate buffer at the times indicated above. The spinal cords and bladders were dissected, postfixed overnight, and cyroprotected in 30% sucrose. An 8-mm-long part of the thoracic spinal cord including the lesion site of animals perfused after 1 and 8 weeks post-cell grafting was cut into 30-μm-thick coronal sections and directly mounted onto superfrost slides and processed for immunohistochemistry. Sections were collected in 14 series resulting in equally spaced sections with a distance of 420 μm between adjacent sections on each slide. For analysis of bladders, catheters were removed, bladders were freed from any connective tissue, and the weight was determined following brief drying with a paper towel.

Eriochrome Cyanine Stain for Spared Myelin

Eriochrome cyanine (EC) staining was used to identify myelinated areas in the thoracic spinal cord according to Kigerl et al. (32). In brief, spinal cord sections previously stored at −80°C were air-dried for 1-2 h at 37°C in a dry incubator. After incubation with acetone for 5 min, the slides were air-dried for 20 min and then stained in EC solution (0.2% Eriochrome cyanine R; Merck Millipore), 0.5% H₂SO₄, 10% iron alum in distilled water for 30 min, followed by a 5-min rinse in tap water. Stained sections were then differentiated for 7 min in 5% iron alum, which was followed by a 10-min rinse in tap water. Afterward, sections were placed in borax-ferricyanide solution (1% borax, 1.25% potassium ferricyanide in distilled water) for 10 min. After another 5-min rinse in tap water, slides were dehydrated through a graded ethanol series and coverslipped using Neomount (Merck Millipore).

The total cross-sectional area of the spinal cord and the inner border of spared white matter were measured by an observer blinded for group identity on digitized images of the cord using a computer-assisted image analysis program (Cell F, Olympus). The digital images of EC-stained sections were obtained using an Olympus BX53 fluorescent microscope equipped with a XC camera. White matter was defined as the area that was stained for EC and considered spared if the myelin stain was dense, contiguous, and grossly normal in appearance with absence of cavities and minimal gliosis. The lesion area was identified by the loss of EC staining as well as severe tissue disruption. The injury epicenter was defined visually as the spinal cord section with the smallest visible rim of spared myelin. To calculate the percentage of white matter sparing (WMS) at the lesion epicenter, the total area of spared white matter (WMA) was divided by the WMA of comparable regions of intact animals (n = 3).

Immunohistochemistry

Double/triple immunofluorescence labeling techniques were performed with direct-mounted sections to assess graft survival and lesion extent in vivo. Sections were washed in TBS, blocked in TBS/5% donkey serum (GE Healthcare)/0.25% Triton X-100 (Neolab) for 1 h and incubated with primary antibodies in TBS/1% donkey serum/0.25% Triton X-100 overnight at 4°C. The following day, sections were rinsed in TBS/1% donkey serum and incubated with Alexa Fluor 594 or Alexa Fluor 488 donkey secondary antibodies for 2 h. Sections were coverslipped with Fluoromount G (Southern Biotech, Birmingham, AL, USA).

The following primary antibodies were used: mouse anti-GFAP for astroglia (1:1,000; Merck Millipore), rabbit anti-GFP (green fluorescent protein, 1:750; Invitrogen) to detect the grafted cells, rabbit anti-Ki-67 (1:500; Novacastra/Leica Biosystems Wetzlar, Germany) for proliferating cells, and DAPI as nuclear counterstain. Immunolabeling was visualized using Alexa Fluor 488 (1:300; Life Technologies) or Alexa Fluor 596 (1:300; Life Technologies)-linked donkey secondary antibodies.

Survival of Grafted Cells

The area occupied by GFP+ cells was measured in coronal sections of animals, which survived for 1 (n = 4 BMSCs; n=5 fibroblasts) or 8 weeks (n=8 BMSCs; n=8 fibroblasts) post-cell grafting. Every 14th section spaced 420 μm apart was analyzed using a computer-assisted image analysis program (Cell F; Olympus) by an observer blinded to group identity. The summed up area measurements were multiplied with 30 μm (section thickness) and 14 (number of series) to obtain the total volume occupied by grafted cells.

Proliferation of Grafted Cells

DAPI/GFP/Ki-67 triple-positive cells were quantified in coronal sections at the lesion epicenter of animals that survived for 1 week (n = 4 BMSCs; n = 5 fibroblasts). Three to four fields of view were chosen based on the area with highest density of grafted cells. Image stacks were taken with a confocal microscope (FluoView FV1000; Olympus). A minimum of 100 GFP-labeled cells per animal was analyzed to determine the percentage of Ki-67+ cells.

Statistics

All data are presented as mean ± standard error of the mean (SEM). Differences between groups were investigated using one-way analysis of variance (ANOVA) followed by post hoc Fishers least significant difference (PLSD) test. A two-way ANOVA was performed for analyzing data from BBB testing. For each time point, one-way ANOVA followed by Fisher's PLSD was used. An unpaired t-test was performed to determine differences in cell survival/proliferation. Statistical analyses were performed using PRISM6 software (GraphPad, San Diego, CA, USA).

Results

Adult Fischer 344 rats underwent a 200 kdyn T9 contusion injury. Three days later, syngeneic BMSCs or fibroblasts obtained from Fischer 344 GFP transgenic rats were injected into the lesion epicenter. Animals survived for an additional 1 or 8 weeks. Behavior was assessed weekly using the BBB score and toward the end of the survival using the grid walk. Cystometric measurements were performed 8 weeks post-cell grafting.

BMSC Characterization

Antigenic characteristics of BMSCs were analyzed by flow cytometry and immunocytochemistry. Flow cytometry revealed that BMSCs were strongly positive for the surface marker CD90 (97.9%), while virtually no cells positive for CD45 (0.1%) and CD68 (0.16%) could be detected (Fig. 1A-C). Immunocytochemical labeling fully confirmed these results (Fig. 1D-F).

Figure 1.

Characterization of BMSCs. (A-C) BMSCs from GFP-transgenic rats were labeled with antibodies for CD90, CD45, or CD68 for cytometry. Histograms are shown with labeled population in black and controls (omission of primary antibody) as gray lines. BMSCs are positive for (A) CD90 and negative for (B) CD45 and (C) CD68. Cultured BMSCs are all labeled for (D) CD90 and are negative for (E) CD45 and (F) CD68. Immunocytochemical markers are shown in red, and DAPI nuclear staining is shown in blue. Scale bar: 50 μm.

Tissue Sparing

Several articles have indicated that grafting of BMSCs has neuroprotective effects reflected in increased tissue sparing and reduced cavity formation (5,24,26,44,47,64,73). Therefore, we analyzed white matter sparing in Eriochrome cyanine-stained sections at the lesion epicenter (region of maximal damage) 10 days and 8.5 weeks postinjury (1 and 8 weeks postgrafting). Following the contusion injury, all animals showed characteristic cavitation and tissue loss as early as 10 days postinjury (7 days postgrafting) leaving a peripheral rim of spared white matter. Remaining white matter further decreased 8 weeks postgrafting, and a severe shrinkage of cross-sectional area could be observed from 1 week to 8 weeks as expected (Fig. 2A-F). The Eriochrome cyanine staining of peripherally located white matter appeared closer to normal compared to white matter directly adjacent to the central lesion. The mean percentage of spared white matter at 1 week postgrafting was higher in BMSC-grafted animals (19.7 ± 5.1%) compared to injured control animals (5.4 ± 2.3%). However, grafting of fibroblasts had a similar effect (19.3 ± 5.8%), and overall these differences missed statistical significance (ANOVA p = 0.077) (Fig. 2G). The apparent tissue sparing in grafted animals was transient as white matter continued to degenerate over time in all groups with the largest decline observed in BMSC- and fibroblast-grafted animals (injury only: 2.0 ± 0.5%; BMSCs: 4.1 ± 0.9%; fibroblasts: 3.2± 1.2%) (Fig. 2H). Statistical analysis indicated that neither BMSCs nor fibroblasts had a positive effect on tissue preservation (ANOVA p = 0.292). Similarly, 1 week post-cell grafting, the injured control group exhibited less white matter area (WMA) at the epicenter compared to BMSC- and fibroblast-grafted animals (ANOVA p = 0.079) (Fig. 2I). Compared to the 1-week time point, animals with cellular grafts showed the largest absolute decline in WMA in the subsequent 7 weeks resulting in much smaller insignificant differences between all groups (ANOVA p = 0.28) (Fig. 2J).

Figure 2.

Tissue sparing and representative images of spinal cord cross-sections at the lesion center. Tissue sparing was measured using Eriochrome cyanine-stained coronal sections at the lesion center (A-C) 1 week and (D-F) 8 weeks post-cell grafting. At 1 week postgrafting, animals with (B, C) cellular grafts seem to have a slightly larger rim of spared tissue. No apparent differences are visible at 8 weeks comparing (D) nongrafted control animals with (E) BMSC- and (F) fibroblast-grafted animals. Dorsal is to the top. Scale bar: 250 μm. (G, H) Bar graph showing the percentage of white matter sparing (WMS) compared to control animals and (I, J) white matter area (WMA) at the epicenter at 1 and 8 weeks. WMS represents the spared WMA at the epicenter divided by the WMA of control sections of intact animals. There is no significant difference between treatment groups.

Cell Survival and Proliferation

Since we did not detect significant differences in white matter sparing between animals transplanted with BMSCs or fibroblasts, we determined whether grafted cells equally survived in the injured spinal cord. The area occupied by GFP+ cells was quantified in every 14th section to calculate the volume/area of GFP-positive cells. Already 1 week after cell injection into the lesion site, the area filled by GFP+ BMSCs was smaller (0.31 ± 0.10 mm³) compared to GFP+ fibroblasts (2.01 ± 0.49 mm³; p < 0.05) (Fig. 3A, D, G, I). The volume occupied by GFP-labeled cells declined from 1 to 8 weeks postgrafting, particularly the volume of GFP+ fibroblasts was reduced by more than 50% (to 0.89 ± 0.26 mm³; p < 0.01 comparing 1 and 8 weeks) (Fig. 3D, E, G-J), whereas the volume of GFP+ BMSCs was only slightly reduced to 0.22 ± 0.09 mm³) (Fig. 3A, B, G-J). Most GFP-labeled cells were detected at and around the lesion epicenter in BMSC- and fibroblast-grafted animals (Fig. 3I, J). To determine if the larger number of GFP+ fibroblasts 1 week after cell injection was due to cell proliferation, we performed a triple labeling of either GFP+ BMSCs or GFP+ fibroblasts with Ki-67 and DAPI (Fig. 4). Most Ki-67-labeled nuclei were not within GFP-labeled cells representing proliferating host-derived cells such as macrophages. Only 1.14 ± 0.72% of GFP+ BMSCs were Ki-67+ and 3.7 ± 0.87% of GFP-labeled fibroblasts, respectively; however, there was no significant difference between these two groups (p = 0.11).

Figure 3.

BMSC and fibroblast survival 1 and 8 weeks posttransplantation. Cell survival was analyzed 1 and 8 weeks postgrafting in coronal sections. Representative images of the lesion site showing (A-C) GFP+ BMSCs and (D-F) GFP+ fibroblasts (green). Host astrocytes are labeled with GFAP (red). (C, F) Higher magnification of the areas outlined in (B) and (E), respectively. Dorsal is to the top. Scale bar in (A, B, D, E): 250 μm; (C, F): 25 μm. (G, H) Bar graph showing the volume of GFP+ BMSCs and fibroblasts in mm³ at (G) 1 week and (H) 8 weeks postgrafting. Note the significantly better survival of fibroblasts in the lesion site (unpaired t-test, *p < 0.05). (I, J) Rostrocaudal distribution of GFP-labeled cells was quantified in serial sections spaced 420 μm apart.

Figure 4.

Analysis of cell proliferation 1 week posttransplantation. Proliferating cells were identified by Ki-67 labeling (red) 1 week postgrafting in sections containing the lesion epicenter. (A) Representative image of the lesion site demonstrates that most Ki-67-labeled nuclei are not graft derived (GFP - cells; arrowheads). A single GFP/Ki-67 double-labeled cell is visible (arrow). (B-E) Higher magnification showing a GFP/Ki-67+ fibroblast (arrow). (F-I) Most GFP+ cells have Ki-67- nuclei (open arrows), and Ki67+ nuclei are GFP - (arrowheads). Nuclei are stained with DAPI (blue). Scale bar in (A): 50 μm; (B-I): 10 μm.

Functional Recovery

BBB Locomotor Score.

The amount of tissue sparing at the lesion center strongly correlates to functional recovery (7). Animals were therefore evaluated once a week using the BBB score. All rats showed complete hindlimb paralysis with a BBB score of 0 at 1 day postinjury and regained some hindlimb motor function over time. All groups showed a characteristic plateau in recovery (BBB score between 8 and 9) around 3 weeks postinjury. Two-way ANOVA showed an overall significant change over time (p < 0.0001). However, no significant differences between the three groups were evident (p = 0.196). Performing an exploratory one-way ANOVA at each time point, no significant differences between the groups were detected at each time point including 45 days postinjury. Only at the last time point evaluated (52 days postinjury), BBB scores were significantly higher in BMSC-grafted animals compared to injured control animals without cellular graft (Fisher's PLSD p < 0.05) but not significantly different from fibroblast-grafted control animals. At the end of the experiment, the mean BBB scores of BMSC- and fibro-blast-injected animals and injured control animals were 9.8 ± 0.3, 8.8 ± 0.3, and 8.3 ± 0.5, respectively (Fig. 5A).

Figure 5.

Assessment of functional recovery. Functional outcomes were assessed in BBB and ladder walking after T9 contusion injury and BMSC or fibroblast transplantation. (A) Hindlimb motor function was evaluated according to the BBB open field locomotor rating scale once a week starting 10 days postinjury until day 52. BBB testing shows an improvement in all groups over time (two-way ANOVA p < 0.0001) but no overall group differences (p = 0.196). Only using a one-way ANOVA at each time point, BMSC-grafted animals differ from ungrafted control animals at 7 weeks postgrafting (52 days postinjury; *p < 0.05). Dashed line indicates time of cell grafting (3 days postinjury). (B) Functional outcome on the grid walk task were evaluated 7 weeks postinjury. The number of missteps given as percentile of the total number of steps is not affected by cell grafting (ANOVA, p = 0.59).

Grid Walk

Deficits in sensorimotor integration were also examined using the horizontal ladder walking test. Animals were scored in five trials 9 days before perfusion (50 days postinjury). Total steps and the number of missteps were quantified and are shown as percentile of the total number of steps. Animals receiving BMSC or fibroblast grafts performed only slightly better (BMSCs: 71.3 ± 2.9% and fibroblasts: 72.1 ± 4.5%) than control animals without cell grafts (contusion: 76.3 ± 3.2%), and this difference was not significant (ANOVA p = 0.59) (Fig. 5B).

Urodynamic Measurements

To determine whether BMSCs can improve bladder dysfunction 8 weeks post-transplantation, catheters were implanted in the bladder dome after completion of sensorimotor testing followed by measurements of urodynamic parameters 3-4 days postimplantation. Continuous cystometrograms (CMGs) from micturition cycles over 30-45 min were collected for each animal to evaluate baseline bladder pressure (P_base), maximal voiding pressure (P_max), pressure difference (P_diff = P_max -P_base), micturition volume, and micturition frequency. Compared to naive uninjured animals (Fig. 6A and dashed line in Fig. 6C-H), injured animals of all groups (Fig. 6B) demonstrated a similar baseline bladder pressure (Fig. 6C), a slightly increased maximal voiding pressure (Fig. 6D), and a significantly increased pressure difference (ANOVA p < 0.05) (Fig. 6E). In addition, spinal cord-injured animals frequently showed nonvoiding contractions (NVC), which was not observed in naive animals. These changes were observed irrespective of cell grafting.

Figure 6.

Urodynamic measurements of bladder function. Spinal cord-injured rats underwent cystometric measurements 8 weeks postcontusion injury. Representative 15-min cystometry graph of (A) uninjured control and (B) a spinal cord injured rat. Note the increased micturition frequency, the decreased micturition volume, and the presence of nonvoiding contractions (NVC) in injured animals compared to naive animals. Analysis of (C) P_base (baseline bladder pressure), (D) P_max (maximum voiding pressure), (E) P_diff (pressure difference), (F) micturition frequency, (G) micturition volume, and (H) bladder weight. Dashed lines in (C-H) represent control values of uninjured naive animals (one-way ANOVA, *p < 0.05, ***p < 0.001, ***p ≤ 0.0001 compared to naive animals). No significant differences are detected in any measure between injured control animals without cell injection and BMSC- or fibroblast-grafted animals.

BMSC- and more pronounced fibroblast-grafted animals showed a trend toward reduced micturition frequency in comparison to injured animals without injection; however this difference did not reach significance (ANOVA p = 0.0535 comparing all injured groups) (Fig. 6F). The voided volume per micturition of all injured animals was decreased compared to intact animals, but this did not reach significance (ANOVA p = 0.06) (see dashed line in Fig. 6G). The voided volume per micturition was slightly higher in animals with BMSC grafts compared to injury-only control animals, but this was also evident in fibroblast-grafted animals, and statistical analysis did not indicate significant differences when injured groups were compared (ANOVA p = 0.28) (Fig. 6G).

Bladder-sphincter dyssynergia (DSD) often developing after SCI is usually reflected in an increased bladder weight. Weight measurements of dissected bladders showed a significantly increased bladder weight in all injured groups compared to intact animals (Fig. 6H), but there was no difference between animals with BMSC grafts, fibroblast grafts, and injured control animals (ANOVA p = 0.19).

Discussion

Bone marrow stromal cells (BMSCs) have been the subject of several studies in animal models of SCI. Some of these studies have suggested that BMSCs have neuroprotective properties and thereby might influence tissue preservation and functional outcome. Indeed, at least nine ongoing phase I/II clinical trials in Chile, India, China, Korea, Spain, and the US are examining whether BMSC transplantation is safe and might improve functional recovery in patients with SCI (http://www.clinicaltrials.gov). Yet, until today, neither a clear mechanism underlying a potential beneficial effect nor the major parameters influencing treatment effects have been sufficiently defined.

In the present study, we systematically analyzed the effect of BMSC transplantation in a rat spinal cord contusion injury model extending previous analyses to include appropriate cellular controls and outcome measures beyond sensorimotor recovery. Similar to clinical studies using autologous cell grafts, syngeneic grafts were used avoiding the need for immunosuppression. To the best of our knowledge, only two previous studies used fibroblasts as a cellular control (25,50) for grafting BMSCs into rats, but both studies used only a moderate contusion injury (BBB score of 11-14), animals were immunosuppressed (25,50), or tissue sparing was not analyzed (50).

Our studies indicate only a minor functional improvement in an open field test (BBB) in BMSC-grafted animals. The observed effect size is much smaller (1-1.5 points on the BBB scale) than some previous studies have suggested; the improvement is only observed at 8 weeks post-cell grafting, and significant differences (p < 0.05) are only observed in comparison to lesioned animals without cell injection. Importantly, sensorimotor recovery of BMSC-grafted animals was not different from fibroblast-grafted control animals at any time point. These findings were confirmed using the horizontal ladder test, which is more sensitive to small motor improvements. Similar findings, that is, a lack of significant differences in tissue sparing and functional recovery between animals receiving BMSC or fibroblast grafts, were also found in one previous study using human cells with immunosuppression (25). Thus, the “effect” relative to untreated animals cannot be exclusively attributed to BMSCs, but may at least partially be the result of cellular grafting or an injection/reopening of the lesion (36).

Many studies that reported tissue sparing after BMSC transplantation only measured cyst size or analyzed sagittal or longitudinal sections (25,46,52,56,64). Because cyst size can be influenced by the presence of grafted cells and because white matter and not gray matter sparing is relevant for hindlimb motor recovery after midthoracic contusions, quantification of white matter should provide the most objective outcome measure. Consistent with the limited functional recovery, transplantation of BMSCs had no significant effect on white matter sparing compared to control animals receiving fibroblasts or animals that received a contusion injury but no cell graft. Interestingly, 1 week postlesion, fibroblast- and BMSC-grafted animals had a larger area of white matter sparing compared to non-grafted animals. However, this short-term delay in white matter degeneration did not result in long-term tissue sparing. At 8 weeks postgrafting, only minor differences in the amount of remaining white matter were observed between all three groups.

As an additional clinically relevant outcome measure, animals underwent urodynamic assessments. To date, only a few studies have investigated the influence of cell transplantation approaches on recovery of bladder function after SCI (15). The extent of recovery depends on the severity of the injury and the degree of white matter sparing at the injury site (53). The function of the lower urinary tract depends on complex neural circuits that are located in the brain and spinal cord (53). Lesions above the lumbosacral level can lead to the loss of supraspinal control of the lower urinary tract altering the coordination between bladder and sphincter and impairing micturition (34,38). SCI patients perceive bowel and bladder dysfunction as one of the most debilitating functional deficits, significantly reducing the quality of life (3,23). In all of the urodynamic outcome measures examined, a difference between BMSC-, fibroblast-, and nongrafted groups could not be detected. These data are similar to one study that examined urodynamic parameters after BMSC transplantation but did not provide any measures of tissue sparing (50). To examine voiding functions, we implanted a catheter into the bladder dome, which avoids bladder outlet obstruction and the impairment of voiding efficiency, which can occur with a transurethral bladder catheter (65,75). Another advantage of this method for bladder catheterization is the ability to perform urodynamic measurements in awake rats, whereas many other studies examined animals under urethane anesthesia (34,35,71). This anesthetic can induce changes in voiding mechanisms since it may alter synaptic reflexes (4,75,76). However, a drawback of bladder dome catheterization is the direct manipulation and potential irritation of the bladder wall (75), which might influence the outcome and could account for the relatively small differences between intact and spinal cord-injured animals.

As mentioned above, several previous studies have suggested that BMSCs improve locomotor recovery after SCI (24,26,46,55,78). A meta-analysis of studies using BMSC grafting versus controls in rat models of SCI suggested a mean increase of 3.9 points in the BBB score [95% confidence interval (CI): 3.1 to 4.7; p < 0.001] (47). However, for interpreting this impressive average improvement, several factors need to be considered. 1) The BBB score is a nonlinear rating scale, and a critical point is the onset of weight-supported stepping (7,8,41,47,62). An increase in the BBB score could be clinically meaningless, depending on where it is observed on the scale (47). 2) Several studies included in this meta-analysis suggested an increase of up to 17 points on the BBB scale, an outcome that is at best questionable. 3) In the meta-analysis by Oliveri et al., locomotor recovery was associated with the recipient rat strain (p < 0.001) and the SCI trauma model (p=0.017) performed (47). Studies in Fischer 344 rats have not been done to date. 4) The smallest BBB improvement was observed in studies using syngeneic and autologous transplants, whereas xenogeneic and allogeneic transplants sometimes in conjunction with immunosuppression generally resulted in better outcomes. Indeed, locomotor recovery seems to correlate with the immunosuppressive status in allogeneic or xenogeneic grafting (p = 0.047) (47). Thus, factors other than BMSCs, for example, the administration of immunosuppressive agents (mostly cyclosporin) or graft rejection seem to influence the outcome (2,24,45,46). This is of particular relevance as autologous cell transplants are most commonly used in ongoing clinical studies (see below).

Interestingly, although syngeneic BMSCs were used in the current study, cell survival of BMSCs was already significantly reduced 1 week after cell injection in comparison to fibroblast grafts. This could be explained, at least partially, by a slightly increased proliferation of the grafted fibroblasts compared to BMSCs (3.7% vs. 1.1%, respectively) 1 week postgrafting, even though this difference did not reach statistical difference. In both groups, cell survival was further reduced at the last assessment 8 weeks postgrafting. A decrease in the survival of grafted BMSCs over time has previously been reported with BMSCs becoming undetectable in some studies after 3-4 weeks (27,44,46,55,73). However, limited cell survival does not seem to influence reported improvements in functional outcome (46,55,73). This might indicate that the transient presence of BMSCs is sufficient, and graft rejection is possibly even necessary to improve functional recovery. Grafting BMSCs at a later time point postinjury might have improved cell survival and would bring this study closer to a clinically feasible time frame. However, if BMSCs have neuroprotective effects, early administration might be necessary to observe beneficial effects as indicated in previous studies (42,44,56).

The mechanisms underlying BMSC-mediated functional improvement have not been fully defined. Many studies report functional recovery within 2 to 3 weeks post-BMSC grafting into the injured spinal cord (12,46,55,70,78), suggesting that neuroprotection rather than regeneration is the underlying mechanism. The release of trophic factors by BMSCs might be one influence on host tissue survival (14,22,39,43,51). However, the amount of growth factor production is very low; a positive influence on the host environment might therefore be limited. Alternatively, influences on inflammatory responses and in particular macrophage differentiation by secretion of cytokines or direct cell-cell interaction could underlie the observed effects (6,21,31,42,54).

Despite the limited mechanistic insights into BMSC-mediated effects and significant influences of rat strain, immunosuppression, injury model, and autologous/syngeneic grafting versus allogeneic/xenogeneic cell sources, a number of case series/clinical trials with bone marrow-derived cell transplantation in spinal cord-injured subjects have been conducted (9-11,16,20,28-30,33,48, 49,58,66,74). Most of the studies used autologous bone marrow-derived cells for transplantation, injecting them either directly into the spinal cord or administering them into the cerebrospinal fluid (CSF) by lumbar puncture. While most of these clinical studies included only a small number of patients, making a therapeutic evaluation difficult, the administration of BMSCs was generally found to be safe. Even though studies reported some form of sensory or motor improvement, the effects were small, or patients did not show significant beneficial effects after treatment with BMSCs (9,33,48,74).

The heterogeneity of SCI patients, the limited beneficial effect of syngeneic/autologous BMSC grafting in animal models and the difficulty to translate the effect of BMSCs seen in one animal model or strain to another make it difficult to predict the success of future larger clinical trials with BMSCs. The small if any beneficial effect observed in this and several other studies (5,19,47,50,59) does not meet the recently proposed minimal criteria for clinical translation of invasive cell therapies in SCI (36).