In spinal cats, locomotor recovery without rehabilitation is limited, but weight-bearing stepping returns with treadmill training. We studied whether neurotrophins administered to the injury site also restores locomotion in untrained spinal cats and whether combining both neurotrophins and training further improves recovery. Ordinary rat fibroblasts or a mixture of fibroblasts secreting brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT-3) (Fb-NTF) were grafted into T12 spinal transection sites. Cats with each type of transplant were divided into two groups: one receiving daily training and the other receiving no training. As expected, trained cats with/without neurotrophin-producing transplants could step on the treadmill. Untrained cats without neurotrophin-producing transplants could not locomote. However, untrained cats with neurotrophin-secreting transplants performed plantar weight-bearing stepping at speeds up to 0.8 m/s as early as 2 wk after transection. Locomotor capability and stance lengths in these animals were similar to those in animals receiving training alone, suggesting that administration of BDNF/NT-3 was equivalent to treadmill training in restoring locomotion in chronically spinalized cats. Cats receiving both interventions showed the greatest improvement in step length. Anatomical evaluation indicated that all transections were complete and that axons did not enter the cord caudal to the graft. Thus BDNF/NT-3 secreting fibroblasts were equivalent to training in their ability to engage the locomotor circuitry in chronic spinal cats. Furthermore, the rapid time-course of recovery and the absence of axonal growth through the transplants indicate that the restorative mechanisms were not related to supraspinal axonal growth. Finally, the results show that transplants beneficial in rodents are applicable to larger mammals.
Treadmill training improves locomotor function in spinal injured cats (Barbeau and Rossignol 1987; Edgerton et al. 2001; Eidelberg et al. 1980; Lovely et al. 1986), rats (Cha et al. 2007; de Leon and Acosta 2006; Hutchinson et al. 2004; Kunkel-Bagden et al. 1992), and mice (Leblond et al. 2003). In spinal cats, there is a relationship between the type of training provided and the functional recovery observed. A comparison of step- versus stand-trained spinal cats showed improved weight support and standing duration in stand-trained cats and recovery of locomotion in step-trained cats. There was no cross-over effect as step-trained cats performed poorly in the standing task and stand-trained cats did not step effectively (de Leon et al. 1998a). These results suggest that training induces task-specific changes in the locomotor spinal circuitry, as shown by changes in the group I afferent synaptic efficiency that occur with training (Cote et al. 2003).
Another approach to improving locomotor function after transection is to repair the damage by providing a permissive environment for axonal growth with cellular transplantation (Howland et al. 1995; Tessler 1991). The rubrospinal and corticospinal (Grill et al. 1997) tracts are important for locomotion in the cat (Drew et al. 2002; Hongo et al. 1969). Receptors for brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT-3), tyrosine kinase receptor B and C (trkB and trkC), respectively, are present on these axons, and administration of BDNF and NT-3 promotes regrowth and/or sprouting and survival of axotomized neurons at the lesion site or at supraspinal nuclei (Boyce et al. 2002; Grill et al. 1997; Himes et al. 2001; Jakeman et al. 1998; Jin et al. 2000; Liu et al. 1999). Behavioral improvements in a range of tasks have been elicited with transplants of fibroblasts genetically modified to secrete BDNF and NT-3 (Fb-BDNF/NT-3) into incomplete lesions (Mitsui et al. 2005). Furthermore, modest improvements in open field locomotion were obtained after Fb-BDNF/NT-3 transplants into spinalized rats (Nothias et al. 2005). Similar studies have not been carried out in adult cats where more detailed kinematic analyses are standard and the benefits of weight-bearing treadmill training better established. We conducted these experiments in the cat to obtain essential scaling information on whether interventions shown to be beneficial in rodents are also effective in a larger animal.
We show that transplantation of cells engineered to produce NTFs promoted locomotor recovery in untrained spinal cats and that the combination of treadmill training and NTFs provided the greatest improvement in step length. Our results show the potential for intraspinal delivery of NTFs to support recovery of hindlimb locomotion in the cat in the absence of treadmill locomotor training. We also show that this improvement was not associated with regeneration of host axons through the graft and into caudal spinal cord. Thus the improvement because of NTF, like that due to training, is likely to involve local plasticity. Some of these results have been reported previously in abstract form (Boyce et al. 2005).
Sixteen adult female cats (domestic short hair; weight, 2.4–3.6 kg) were used. All animal care and procedures were performed according to National Institutes of Health guidelines and were reviewed and approved by the Institutional Animal Care and Use Committee of Drexel University College of Medicine.
Cats were divided into four groups to evaluate the effects of neurotrophin treatment and/or weight-bearing treadmill training on locomotor recovery after spinal cord transection. One group received no treatment (Op-control, n = 3). Another received treadmill training alone (Trained, n = 4). The third group of cats received transplant of fibroblasts genetically modified to secrete neurotrophic factors but no training (NTF, n = 5), and the last group received both treatments (Combination, n = 4).
All cats were acclimated pre-operatively to a motorized treadmill using 15-min quadrupedal treadmill-walking sessions (5 d/wk) at speeds ranging from 0.1 to 0.8 m/s. To minimize lateral movement of the animal during training, the treadmill was enclosed in a clear Plexiglas frame, and food rewards were used to keep the animal oriented in a forward direction. Locomotion was recorded using two Sony Digital Video cameras at 30 frames/s. The cameras were positioned orthogonal to the treadmill and captured a right lateral view and a rear view of the cat. The animal was shaved, and paint markers were placed on appropriate joints of the right hindlimb (Barbeau and Rossignol 1987; Lovely et al. 1986; Smith et al. 1982) and forelimb. Specifically, markers were placed on the ischium, femoral head, center of rotation of the knee, metatarsophalangeal (MTP) joint, ankle, and the tip of the second digit of the hindlimb (Fig. 1). In addition, the humeral head, center of rotation of the radius (elbow), lateral malleolus, metacarpophalangeal (MCP) joint, and the tip of the second digit of the forelimb were marked (Fig. 1). In two cats, a Vicon Motion Analysis system (Vicon Peak, Lake Forrest, CA) was used in addition to the video cameras, allowing capture of three-dimensional (3D) kinematics at 300 frames/s.
After 4–6 wk of acclimation and collection of pretransection locomotor data, the spinal cord was transected at T11–T12, and the cats received a transplant of either fibroblasts genetically modified to express neurotrophins or unmodified fibroblasts. After 1–2 wk of recovery, cats in the training groups received quadrupedal treadmill training (15 min, 1/d, 5 d/wk). The animals' hindquarters were supported at the tail to provide balance, and perineal and/or tail stimulation was used to initiate stepping (Barbeau and Rossignol 1987; Lovely et al. 1986). As in Belanger et al. (1996), perineal/tail stimulation was used as needed to “induce or maintain locomotion” for cats in all four groups. Although differences in the amount of perineal/tail stimulation needed to initiate/maintain locomotion did exist between animals, it did not correlate with treatment group (comparison between trained, NTF, and combination groups). During the early postoperative period, the swing phase was initiated and completed by the trainer, but over a period of 3–4 wk the animals recovered the ability to generate plantar weight-bearing steps autonomously. Because of the increase in hindlimb adduction that accompanies spinal transection, a 4-cm-high Plexiglas divider was placed between the hindlimbs to prevent interference between the limbs during locomotion. An animal was defined as capable of stepping at a given speed if it could successfully execute 10 consecutive steps without stumbling or dorsi-stepping at that speed (and at all lower speeds). Cats were trained at increasing treadmill speeds from 0.1 up to 0.8 m/s. Locomotor performance of all cats was recorded and evaluated every 2 wk for a period of 12 wk after injury (Fig. 2). For the groups of cats that were not trained, (Op-control and NTF), the evaluation sessions were limited to 5–8 min to minimize training effects that could occur during these sessions.
Spinalization and cell transplantation
All spinalizations were performed under aseptic conditions. Animals were fasted 12 h before spinalization. Anesthesia was induced with an intramuscular injection of ketamine (25 mg/kg) followed by atropine sulfate (0.05 mg/kg) and maintained with isoflurane (1.5–3.0% in oxygen) given through an endotracheal tube. The incision area and lower forelimb were shaved, and stubble removed with a depilatory lotion. The surgical area was scrubbed with Betadine and rinsed with ethanol and then with povidone iodine (10%) solution. A sterile incise drape (Steri-Drape 1040, 3M Health Care, St. Paul, MN) was placed over the skin in the region of the incision. Blood pressure, heart rate, end-tidal CO2, arterial oxy-hemoglobin saturation, respiratory rate, and core temperature were monitored and recorded every 15 min after entubation. Intravenous fluids were administered at a rate of 20 ml/h, and body temperature was monitored and maintained between 37 and 39°C. A laminectomy was performed at the T11–T12 vertebrae. Rongeurs were used to make a ≈5 mm wide by ≈10 mm long opening at the junction of T11–T12 vertebra. The dura was carefully opened and xylocaine (1%) was applied dropwise to the surface of the spinal cord and injected into the spinal cord at the T11–T12 level using a 25G needle to anesthetize the cord and eliminate injury discharge that could occur on spinal transection. The spinal cord was completely severed with microscissors between the T11 and T12 roots. This resulted in separation of the cut ends of the cord by 2–3 mm. Lesion completeness was ensured through visual inspection with a surgical microscope and confirmed histologically.
The procedures for isolation of fibroblasts genetically modified to produce BDNF and NT-3 and the properties of these cells have been described previously (Liu et al. 1999). The fibroblasts were expanded and maintained in T125 flasks in DMEM F12 medium supplemented with fetal bovine serum at 30°C. The fibroblasts were passaged twice weekly until the desired number of cells was obtained. To prevent rejection of the graft, all cats were placed on an immunosuppressive regimen of cyclosporine A (CsA, Sandoz Pharmaceuticals, East Hanover, NJ). Oral doses of 2 mg/kg were administered every 12 h starting 3 days before transection for the duration of the study. On the day of surgery, the fibroblasts were harvested using aseptic techniques (Liu et al. 1997, 1998). The fibroblasts were suspended in Vitrogen (Cohesion Technologies, Encinitas, CA), adjusted to a pH of 7.2–7.4. Vitrogen is a collagen matrix that persists in the lesion cavity and is moderately permissive to axonal growth (Hayashi et al. 2005; Nothias et al. 2005). At the time of transection, all animals received a transplant of unmodified or genetically modified fibroblasts suspended in the Vitrogen matrix.
Fifty microliters of cells in a Vitrogen suspension was injected into the lesion cavity using a 1-ml syringe and a blunted 18G needle. The dura was sutured with 6.0 Prolene, muscles closed in anatomical layers using 4.0 Vicryl (polyglactin 910), and skin apposed with 5.0 Prolene suture. Analgesia was administered for 72 h through a transdermal fentanyl patch (Duragesic, 25 μg/h) (Franks et al. 2000) or by intramuscular injections of buprenorphine (0.01 mg/kg) every 8 h for 3 days. Ampicillin (15 mg/kg), a broad-spectrum antibiotic, was administered to the cat intramuscularly the night before surgery, the morning of the surgery, and at the completion of the surgical procedure. Postsurgical administration of antibiotics (ampicillin, 15 mg/kg, subcutaneously every 12 h) continued for 7 days postsurgery. Lactated Ringer solution (20–30 ml) was administered subcutaneously twice per day to maintain normal hydration in the immediate postoperative period (2 days).
After surgery, the animals were monitored until they achieved sternal recumbancy. They were transferred to individual cages, which were lined with a foam mattress and absorbent pads. Bladders were expressed manually or reflex micturition was initiated by light tactile stimulation of the perineum at least twice per day. The skin of the hindlimbs was massaged to prevent pressure sores, and the hindlimbs were moved throughout their range of motion to prevent orthopedic problems 7 d/wk. If a bladder infection was diagnosed, antibiotics chosen in collaboration with the attending veterinarian, (typically ampicillin 15 mg/kg twice per day, or enrofloxacin 1.3 mg/kg once daily) were administered for 10 days. To maintain hygiene, the animals' hindquarters were cleaned, and hair was clipped at regular intervals.
Video segments containing at least 20 steps with the cat walking forward at the speed of the treadmill were digitized using Adobe Premiere. To generate stick figures of hindlimb kinematics, a body model of the right hind- and forelimbs was superimposed over the joint markers in the digitized video sequence and tracked throughout the stepping sequence using WinAnalyze v1.4 for Windows (Mikromak). The x- and y-positions of each joint marker and the angular excursions about the hip, knee, and ankle joints were exported to Matlab 7.0 (The Mathworks, Natick, MA). Slippage of the knee and elbow markers over the skin was corrected by interpolating their positions, based on the lengths of the femur and tibia (knee) or the humerus and ulna (elbow) (Goslow et al. 1973). Finally, quantitative data on joint angle measures and stepping parameters were extracted in IgorPro (Wavemetrics, Lake Oswego, OR).
Kinematics of pre- and posttransection locomotion was compared for 20 hindlimb step cycles. The metrics of locomotor performance that were examined included stance length, swing height, hindlimb joint excursion, interlimb coordination, and step cycle duration. Stance length was defined as the difference in the horizontal position of the toe marker from the end of swing (touch down of the paw) to the end of stance (toe off). Swing height was determined using the maximal vertical displacement of the MTP joint during the swing phase of each step cycle. Posttransection stance length and swing height were reported as a percentage of their respective pretransection values. They were defined as the stance length index and swing height index, respectively. Angular excursions at the hip, knee, and ankle joints were averaged across steps. The maximum and minimum joint angle values and the ranges of joint excursions were compared.
Interlimb coordination and step cycle duration were measured using analysis of footfall patterns (Bem et al. 1995) for 20 step cycles pretransection and for the last posttransection time-point analyzed. Two types of interlimb coordination were considered. Coordination between the hindlimbs was measured as the variability in the onset of left hindlimb stance with respect to right hindlimb stance. The variability in stance onset between the left hindlimb and forelimb was used to determine coordination between the forelimbs and hindlimbs. In an uninjured cat, there is little variability in stance onset between the hindlimbs and also between the hindlimbs and forelimbs (English 1979). The time between right hindlimb stance onsets gave the duration of each step cycle. Finally, the speed of recovery was evaluated as the time-point at which the animals could locomote at an average treadmill speed (0.4 m/s) and at the maximal treadmill speed tested (0.8 m/s).
Normality of stance length index, swing height index, interlimb coordination (stance onset of one limb relative to another), and hindlimb joint excursion was confirmed using the Kolmogorov-Smirnov test. The effects of treatment on the time-point for locomotor recovery and on posttransection stance length and swing height measures were compared. ANOVA was used for the time-point of recovery and repeated-measures ANOVA for step length and swing height. Tukey HSD posthoc tests were used to determine significant effects between the groups with significance set at P < 0.05. Pre- and posttransection comparisons for stance length, swing height, interlimb coordination, and step cycle duration were performed using paired t-test. In addition, one sample t-tests were used to compare post-transection stance and swing indices to normal (100%). Paired t-test was used to compare pre- and posttransection indices. Circular statistical measures such as angular excursion (Zar 1999a,b) were analyzed using circular statistical methods written as custom routines using IgorPro (Wavemetrics). All other statistical analyses were performed using SPSS 11 (SPSS, Chicago, IL) for Macintosh (Apple Computing, Cupertino, CA).
At the conclusion of the experiments, the animals were anesthetized, and 1 ml of heparin was administered intravenously to the cats and allowed to circulate for 20 min (Howland et al. 1995). The animals were then euthanized with an overdose of pentobarbital sodium (100 mg/kg) and intracardially perfused with 0.9% saline flush followed by ice-cold 4% paraformaldehyde in phosphate buffer. After postfixation of the tissue overnight at 4°C, the thoracic and lumbosacral cord were carefully dissected free of the vertebral column, removed, and briefly washed in 0.1 M phosphate buffer (4°C). The tissue was then cryoprotected in 30% sucrose solution containing 0.5 mM Thimerosal, buffered with 0.1 M phosphate solution for 5–7 days; the sucrose solution was changed every third day. Where possible, the dura was removed from the spinal cord to facilitate infiltration of the sucrose in the spinal tissue. The spinal cord was divided into short segments, usually <1 cm long and embedded in optimal cutting temperature (OCT) medium. The spinal cord was cut into 20-μm-thick transverse sections at the lesion site, and 20- or 40-μm-thick sections in the lumbar (L5–L7) cord on a freezing microtome and mounted onto gelatin/poly-l-lysine–coated glass slides. The mounted slides were air dried at room temperature for 2 h and stored at –20°C until staining was performed.
Histological staining and lesion verification
Mounted tissue sections were allowed to air dry at room temperature for 1 h. The lesion block was cut in 20-μm serial transverse sections. To verify lesion completeness, every seventh slide in the series was Nissl-myelin stained. This involved hydration of the tissue through decreasing alcohol series, followed by a 20-min incubation in myelin stain (Eriochrome Cyanine R). The tissue was differentiated in 2% ammonium hydroxide solution and rinsed in distilled water before staining with cresyl violet (Sigma). The tissue was dehydrated and coverslipped with Permount. The stained sections were examined using light microscopy for the presence of myelinated fibers passing through the graft, which would indicate an incomplete lesion. The rostrocaudal extent of the thoracic lesion was determined based on the length of spinal cord that contained no myelinated fibers.
Mounted tissue sections were ringed with rubber cement to form a well to contain the antibody solutions. After three washes in 0.1 M phosphate buffer (PB), pH 7.4, the tissue was briefly permeabilized in 0.1 M phosphate buffer with 0.3% solution of Triton-X, and rinsed. After blocking for 1 h in 10% goat serum PB solution at room temperature, tissue sections were incubated overnight in a solution with the appropriate primary antibody diluted in a 2% goat serum/PBS with 0.2% Triton. The primary antibodies used to determine axonal sprouting and growth within the lesion included those against serotonin (5HT: Incstar, 1:80,000), calcitonin gene related peptide (CGRP: Peninsula, 1:500), RT-97 (1:100), Tuj-1 (1:200), and growth-associated protein (GAP-43: Chemicon, 1:5,000). Glial fibrillary acidic protein (GFAP, 1:200), ED-1 (1:200), or Vimentin (Dako, 1:100) were used to determine the presence of astrocytes, macrophages, or fibroblasts within and adjacent to the graft, respectively. On the second day, sections were again washed in PBS and incubated with the fluorescent (Jackson Immunobiologicals) secondary antibody (1:200) in 2% goat serum for 2 h at room temperature in the dark. The slides were rinsed in PBS and coverslipped using an aqueous mounting medium (Vectashield).
Untreated spinal (Op-control) cats were incapable of executing plantar weight bearing steps (Fig. 3A) at any speed. Although these cats could perform discernable stance and swing phases of hindlimb movement at slow treadmill speeds (0.1–0.2 m/s), they stepped on the dorsal aspect of their paws and showed little excursion of the knee and ankle during the stance phase. At higher speeds (>0.3 m/s), the hindpaws dragged on the moving treadmill, with little to no motion at the hip, knee, and ankle. Therefore kinematic results for these animals posttransection (Figs. 3–8) are reported for a speed of 0.2 m/s. In contrast, all the other groups, trained cats with unmodified fibroblasts (trained), untrained cats with modified fibroblasts (NTF), and trained cats with NTF-producing fibroblasts (combination), executed plantar weight-bearing steps (Fig. 3, B–D) at all treadmill speeds tested, 0.1–0.8 m/s. Thus both experimental treatments improved treadmill locomotion.
Stance length index
Stance length data are shown for all the cats in Fig. 4A and group averages in Fig. 4B. Within each of the four groups, cats attained similar levels of recovery of stance (Fig. 4A). Average pretransection stance length was 23.51 ± 2.29 cm. After transection, Op-control cats executed short steps on the dorsal aspect of the foot with the hindpaws dragging on the treadmill, and their mean stance length was 6.19 ± 1.52 cm, which was equivalent to 25% of their preoperative stance length (Fig. 4B). Stepping by Trained, NTF, and Combination cats was qualitatively similar and significantly better than Op-controls. The ratio of post- versus pretransection stance lengths (stance length index) was used to quantitatively assess step length recovery after transection. After transection, the stance length index of trained cats decreased to 61% of pretransection level (13.76 ± 2.06 cm). This reduction was comparable to that seen in NTF cats, which was 70% of normal (16.51 ± 2.06 cm). For cats receiving the combination treatment, posttransection stance length index was 87% of pretransection (20.6 ± 2.7 cm), significantly greater than cats with individual treatments. In summary, stance length was reduced with injury but recovered with treatment. The greatest improvement was seen in cats with combination therapy.
Swing height index
Swing height data for all the cats is shown in Fig. 5A and group averages in Fig. 5B. Within each of the four groups, the swing height attained by individual cats varied considerably. The average pretransection swing height was 3.01 ± 0.12 cm. Op-control cats predominantly dragged their hindpaws on the treadmill belt. However, in a few instances, they lifted the hindpaw sufficiently to clear the treadmill belt during swing (0.85 ± 0.14 cm, equivalent to ∼28% of pretransection swing height). Swing height index in Trained cats was 61% of normal (1.83 ± 0.36 cm) and in NTF cats 62% of normal (1.86 ± 0.24 cm). Cats receiving the combination treatment performed at 75% of normal (2.25 ± 0.25 cm).
The ratio of post- versus pretransection swing height (swing height index) was used as a measure of recovery of swing height after transection. Swing height index for Op-control and NTF cats was less than pretransection (P < 0.05). The swing height indices of Trained and Combination cats were not statistically different (P > 0.05) from pretransection values due to the higher variability in swing height index between animals in the Trained and Combination groups compared with the animals in the Op-control and NTF groups. The swing height index was not significantly different among the experimental groups (Fig. 5B). Thus differences in swing height were observed; however, these differences were not found to be significant.
Time-course of locomotor recovery
Op-control cats did not regain the ability to execute plantar weight-bearing stepping at any treadmill speed (Fig. 6). In contrast, cats in the NTF group could plantar step at treadmill speeds of 0.4 m/s by the third week posttransection. Trained cats and those receiving combination therapy were able to locomote at speeds ≤0.4 m/s, after 5 wk of training. A similar trend was observed for the fastest treadmill speed tested (Fig. 6). NTF cats recovered the ability to step at 0.8 m/s by the fourth posttransection week, significantly faster than Trained cats (ANOVA, P < 0.05) and before cats in the Combination group. Thus cats in the NTF group recovered locomotor performance earlier than cats in the other treatment groups.
Interlimb coordination: hindlimb to hindlimb and ipsilateral forelimb to ipsilateral hindlimb
Pretransection, there was little variability (6 ± 2% of step cycle) in ipsilateral hindlimb stance onset with respect to contralateral hindlimb stance onset for each of the groups (Fig. 7). There was a trend toward greater variability in hindlimb to hindlimb stance onset for Op-control cats after transection, but this increase was not significant. In fact, there was no significant difference between pre- and posttransection hindlimb to hindlimb stance onset variability for any of the groups (P > 0.05) and practically no difference for the Trained, NTF, or Combination cats. This suggests that coordination between the hindlimbs was maintained in Trained, NTF, and Combination groups of cats after injury. The increased difference in the smaller group of Op-control cats (n = 3) precludes a similar conclusion for this group.
In contrast, there was an increase in the variability of forelimb–hindlimb stance onset from 5.5% pretransection to 21% posttransection (Fig. 8). This increase in variability was statistically significant (t-test, P < 0.05) for cats in the Op-control, Trained, and Combination groups. There was no difference in pre- and posttransection stance onset variability between the forelimb and hindlimb for NTF cats, although there was a high variance in the performance of NTF cats. Furthermore, there was no difference in pre- or posttransection forelimb–hindlimb coordination between the groups (P > 0.05). Thus forelimb–hindlimb coordination was reduced in all groups.
Hindlimb joint excursion
In Op-control cats, the average angular excursion was reduced about the hip (18 ± 2° posttransection vs. 40 ± 7° pretransection; P < 0.05, repeated-measures ANOVA) and knee (18 ± 3° posttransection vs. 31 ± 5° pretransection; P < 0.05, repeated-measures ANOVA). The ankle joint was consistently more extended posttransection, but the range of movement was not significantly different from pretransection values. For cats in the treatment groups, hip, knee, and ankle ranges of motion were generally within pretransection values (no significant difference except for hip in NTF cats and ankle in Combination cats). Therefore in Op-control cats the hip and knee were more flexed, reflecting the inability of these animals to step. Joint excursions were minimally affected for animals in the treatment groups.
The graft, consisting of fibroblasts within a Vitrogen matrix, was readily identified in all animals. Nissl-myelin staining through the lesion/graft site confirmed the absence of myelinated fibers within the graft, indicating that the lesion was complete in all animals (Fig. 9A).
Transverse frozen sections taken within the graft were stained for markers of axonal growth (GAP-43), neurofilaments (RT-97, Tuj-1), descending raphe (5-HT), and dorsal root afferents (CGRP). GAP-43, RT-97, Tuj-1, 5-HT, and CGRP-positive fibers were present within the grafts of all cats, indicating some axonal growth under each of the conditions (Fig. 9, C and D). Notably, labeled axons did not exit the graft into caudal host tissue in any of the animals. ED-1–positive cells were present within the graft of some cats, suggesting the migration of monocytes or macrophages into the graft (Fig. 9D). In some animals, cysts formed within or at the edges of the graft (Fig. 9A). A GFAP-positive glial scar surrounded the graft, and the cysts that had formed within the graft (Fig. 9B). Group differences in axonal density within the grafts were not quantified because of varying degrees of cyst formation within the grafts (Fig. 9, A and B), which artifactually increased axonal densities in some parts of the graft.
Fibroblasts survived within the Vitrogen matrix in all cats. In some animals, fibroblasts were aggregated, rather than evenly distributed throughout the matrix, as in usually the case in rats (Liu et al. 1999; Mitsui et al. 2005; Nothias et al. 2005). Because host fibroblasts migrate after spinal lesions, it is possible that some of the fibroblasts within the graft originated from the host (Schwab and Bartholdi 1996).
Unmodified fibroblasts (Fb) were larger than the genetically modified neurotrophin secreting cells (Fb-NTF), and fewer were needed to fill the lesion cavity (on average, 2.8 ± 1.0 × 106 Fb vs. 19.5 ± 13.6 × 106 Fb-NTF). This difference did not influence the rostrocaudal extent of the lesion at 12 wk after transection between Fb and Fb-NTF cats (data not shown) or their locomotor recovery. The number of NTF-secreting cells grafted into the NTF and Combination groups also varied (range: 8.5–49 × 106), but again, there was no correlation between numbers of cells grafted and locomotor performance or rostrocaudal extent of the lesion.
We conclude that the thoracic transection injuries were complete in all animals and that no axons grew across the lesion site.
The major findings of this study are 1) fibroblasts genetically modified to secrete neurotrophins restored locomotor ability in untrained spinal cats, 2) combination therapy was superior to individual treatments in restoring stance length, 3) neurotrophin secreting transplants used in the rat can be successfully translated to a larger animal, and 4) this recovery did not correlate with any axonal regeneration through the graft into caudal spinal cord.
Contribution of spontaneous recovery
Op-control cats that received grafts of unmodified fibroblasts were incapable of plantar stepping, although one cat transiently recovered some locomotor function. Limited bipedal locomotor recovery has been reported for untrained cats. Lovely et al. (1986) found that, at 5–7 mo after transection, untrained spinal cats executed plantar, weight-bearing stepping at treadmill speeds of 0.24 ± 0.089 m/s. Another study found significant variability in locomotor performance of untrained spinal cats. Of their nine cats, two were capable of plantar, weight-bearing stepping, two transiently recovered locomotor function, and five never regained the ability to plantar step (de Leon et al. 1998b). We assessed locomotor performance every 2 wk, while it was evaluated weekly in the aforementioned studies (de Leon et al. 1998b; Lovely et al. 1986). It is possible that reduced treadmill exposure could account for the absence of persistent plantar stepping in any of our Op-control cats.
Contribution of interventions
NTF cats could step at 0.1–0.8 m/s, often at the first time-point tested (2–2.5 wk after transection) and stepping persisted for the duration of the experiment (12 wk after transection). Trained cats or cats with combination treatment acquired this ability by the fourth to fifth week after transection, comparable to data previously reported (Barbeau and Rossignol 1987; Belanger et al. 1996; de Leon et al. 1998b). Thus, NTF grafts were as effective as treadmill training in restoring locomotor function, and NTF grafts alone seemed to promote earlier locomotor recovery when used without training.
Recent experiments in adult spinal mice suggest that the type of treadmill training administered greatly influences the rate of locomotor recovery. The hindlimbs of spinal mice were moved in three different trajectories using robotic training (Cai et al. 2006). In the first trajectory, the hindlimbs were actively moved for the animal (fixed). In the other two (band and window), the robot's intervention modified rather than imposed the hindlimb's trajectory. Locomotor recovery of spinal mice trained using the fixed trajectory was slower than for mice trained using band or window trajectories.
During the initial weeks of training, we imposed a trajectory on the hindlimbs of the cat by manually moving the hindlimbs through the swing phase of the step cycle and placing the hindlimbs on the moving treadmill to initiate stance. In contrast, the hindlimbs of the NTF cats were never manipulated during evaluation sessions, allowing recovery of a natural pattern of hindlimb movement rather than the learning of hindlimb trajectories imposed by the experimenter, possibly contributing to the faster locomotor recovery.
Our training paradigm differs from protocols described elsewhere (Barbeau and Rossignol 1987; Lovely et al. 1986). Cats in our study received quadrupedal treadmill training, with the trunk unsupported, and for a shorter duration (15 vs. 30 min/d). Despite these differences, posttransection stance lengths in our Trained cats (13.76 ± 2.06 cm) were comparable to other groups (13.8 ± 0.8 cm) (Barbeau and Rossignol 1987). Swing was also restored with all treatments and the hindlimb kinematics of trained cats was similar to published reports. There was no difference in the performance of Trained and NTF cats, suggesting that training and NTF grafts are equivalent in restoring locomotion to spinal cats.
Finally, cats receiving training might be expected to have increased muscle mass compared with untrained animals as has been shown in rats (Nothias et al. 2005). de Leon et al. (1998b) addressed the issue of muscle atrophy in spinal cats and found that the degree of atrophy occurring in untrained cats was biomechanically insufficient to prevent locomotor recovery. Although training did increase muscle mass (visual observation), we did not observe differences in locomotor performance between Trained and NTF groups in our experiments. Therefore training and NTFs are more likely to promote locomotor recovery through neural mechanisms.
Effects of neurotrophins on spinal circuitry
Because locomotor recovery occurred as early as 2 wk after transection and we did not observe axonal growth into the caudal spinal cord, regeneration of descending axons is an unlikely mechanism for recovery in NTF cats. We suggest instead that locomotor recovery could be caused by plasticity within the lumbar cord. Published reports have shown changes in inhibitory protein expression (GAD mRNA) in cat lumbar spinal cord with step training (Tillakaratne et al. 2002), as well as increases in BDNF and NT-3 in spinal rats with exercise (Gomez-Pinilla et al. 2001, 2002; Ying et al. 2005).
Using this argument, the locomotor improvement of our NTF cats could be caused by BDNF- and NT-3–induced plasticity in the lumbar cord. Retrograde transport of neurotrophic factors within the CNS is well known (DiStefano et al. 1992). There is also evidence to support the anterograde transport of BDNF and NT-3 within the CNS (Conner et al. 1997; von Bartheld et al. 1996). Therefore it is possible that the NTFs secreted by the T12 fibroblast transplants could have been transported to the lumbar spinal circuitry to affect the changes necessary to promote the locomotor improvements observed in our NTF animals.
Neurotrophic factors have local effects on axonal growth and synaptic efficiency that may account for the recovery of function. Rat proprioceptive afferents damaged by pyridoxine (B6) toxicity are rescued by NT-3 administration (Helgren et al. 1997). In addition, BDNF increases axonal sprouting (Lu et al. 2005) and promotes locomotor recovery (Mitsui et al. 2005), whereas NT-3 rescues Clarke's nucleus from retrograde degeneration (Himes et al. 2001) in rats. There are also synaptic effects of NTF (Lu 2003). In the neonatal rat spinal cord, NT-3 application enhanced monosynaptic and polysynaptic excitatory postsynaptic potential (EPSP) amplitude in L5 motoneurons in vitro (Arvanov et al. 2000). In contrast, BDNF significantly reduced monosynaptic EPSP (Arvanian and Mendell 2001). Because we used a combination of two neurotropic factors, we could not ascertain the individual contribution of each neurotrophin to recovery.
The mechanism by which training promotes locomotor recovery may be similar to the mechanism by which neurotrophins promote recovery, because as stated above, exercise up-regulates BDNF and NT-3 mRNA in lumbar spinal neurons and skeletal muscle of spinal cord–injured rats (Gomez-Pinilla et al. 2001, 2002). Thus the effects of exercise or training could be mediated in part by the increase in BDNF and NT-3 occurring with locomotor activity. These trophic factors could lead to the formation of new synapses and circuits or enhance the efficiency of synaptic transmission in locomotor circuits caudal to the injury. Perhaps the increased performance of the stance phase of stepping in Combination cats was caused by the additive effect of NTFs secreted by the grafted cells and NTFs produced during training. Exercise may prolong the time that NTFs are available to the spinal cord and thus the combination may be more effective.
This research was supported by National Institute of Neurological Disorders and Stroke Grants NS-41975 and NS-24707 and The Christopher Reeve Paralysis Foundation.
We thank Dr. Timothy B. Himes, S. Jacob-Vadakot, and M. Obrocka for technical assistance and discussion.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by the American Physiological Society