Materials and methods 90 and 70 alcohol 5 min each and incubated for 1 h

166 G using tests like the BBB score, footprint analysis, fixative 4 paraformaldehyde in 0.1 M phosphate buffer kinematic gait analysis, grid walk, narrow beam, open field with 5 sucrose. The spinal cords were removed and exploration and placing response we covered a broad postfixed overnight. After immersion for 2–3 days in a spectrum of behavioral and physiological parameters. The 30 sucrose solution, the spinal cords were embedded in a outcome of these paradigms was correlated in order to protein matrix [28]. The spinal cords were frozen in define their relationship and the validity of the individual isopentane at 2408C and 50-mm sections were cut in the tests. Based on these results we developed a testing sagittal plane. The sequential series of sections were strategy for functional outcome, optimized for efficiency, mounted on slides. reliability and comparability between different lesion types To visualize axons, a monoclonal antibody against the and severities. neurofilament 200 kDa subunit Boehringer Mannheim, Germany was used. Slides were treated with ethanol 95–acetic acid 5 for 25 min at 48C, rehydrated in

2. Materials and methods 90 and 70 alcohol 5 min each and incubated for 1 h

at room temperature with the neurofilament antibody 2.1. Animals 1:200 in 0.1 M PBS. To localize the primary antibody, slides were incubated for 45 min at room temperature with Two hundred and one adult rats of either sex Lewis, a biotinylated secondary antibody anti-mouse, rat ab- 200–250 g at the beginning of experiment were used. One sorbed, diluted 1:100 followed by the avidin–biotin– hundred and thirty-five rats received a mid-thoracic T8 peroxidase complex Vector, Burlingham, UK. The per- bilateral dorsal hemisection and 66 a mid-thoracic contu- oxidase activity was detected with diaminobenzidine and sion injury. They were housed in groups under a 12:12-h H O Sigma, St. Louis, MO. For background control, the 2 2 light dark cycle with water and food ad libitum. All secondary antibody was applied in the absence of primary experiments were approved by the veterinary department antibodies. of the canton Zurich. 2.4. Behavioral procedures 2.2. Surgery A varying number of animals per test was analyzed For both injury types, the animals were anesthetized since data were collected from control animals in several with a combination of Fentanyl and Fluanisone 0.3 mg kg experiments. All tests were performed before spinal cord i.p., Hypnorm, Janssen, Belgium and Midazolam 0.6 injury and at a medium 5 weeks or a long range 3 mg kg i.p., Dormicum, Roche, Switzerland. Body tem- months after injury. After this period stable neuro- perature was maintained at 378C during surgery using a pathological conditions can be expected [27]. heating pad. After shaving the skin an incision in the Except for the placing response, the results of single mid-thoracic region was made and a T8 laminectomy motor tests were also analyzed in a scoring system. For the performed. paradigms of kinematic analysis, grid walk, narrow beam, For dorsal hemisection, the dura mater was opened and a footprint analysis and exploration we established scores to bilateral dorsal lesion was made with iridectomy scissors, allow comparison of the single parameters. thus completely transecting the rubrospinal tract and the main dorsal corticospinal tract component. 2.4.1. BBB open field locomotion score The contusion injury was performed by using a stan- This open field walking score measures recovery of dardized weight-drop injury device NYU impactor hindlimb movements in rats during free open field locomo- [15,26]. The height of the weight drop impact was varied tion as described by Basso et al. [4]. A score of 0 was 2 2 from 6.25 g cm to 25 g cm . After the lesion, the given if there was no spontaneous movement, a score of 21 muscles were sutured and the skin was closed with surgical indicated normal locomotion. Plantar stepping with full clamps. Postoperative care included analgesia application weight support and complete forelimb–hindlimb coordina- Buprenorphin, 0.03 mg kg, s.c., Temgesic, Reckitt tion is reached when an animal shows a score of 14 points. Colman, UK, regular bladder expression and antibiotic We used a modified version of the BBB score if the treatment Trimethoprim, 0.83 ml kg, i.p., Bactrim, sequence of recovering motor features was not the same as Roche, Switzerland if infections occurred. described in the original score. If this was observed, points for the single features were added independently. For 2.3. Histology example, a rat showing incomplete toe clearance, enhanced foot rotation and already a ‘tail-up’ position, one additional After the behavioral tests were finished, the animals point was added to the score for the tail position. were killed by a pentobarbital overdose 50 mg 100 g and The rats were tested preoperatively in an open field, perfused transcardially with Ringer solution containing which was an 803130-cm transparent plexiglas box, with 100 000 IU l heparin and 0.25 NaNO , followed by a walls of 30 cm and a pasteboard covered non-slippery 2 G .A.S. Metz et al. Brain Research 883 2000 165 –177 167 floor. In postoperative sessions two persons observed each rotation of the feet of more than double values as com- animal for a period of 4 min. pared to its own baseline values; 3 points were recorded if the animal showed no signs of toe dragging but foot 2.4.2. Kinematic analysis rotation; 4 points were rated if the animal showed no signs Kinematic analysis was performed for one hindlimb in of exo- or endo-rotation less than twice the angle of the animals, which were able to perform plantar stepping on a baseline values. treadmill after dorsal hemisection. After shaving the limb, the iliac crest cresta iliaca, hip greater trochanter, ankle 2.4.4. Grid walk lateral malleolus and the fifth metatarsophalangeal Deficits in descending motor control were examined by MTP joints were marked with ink. Because the position assessing the ability to navigate across a 1 m long runway of the knee joint is obscured by loose skin coverage, it was with irregularly assigned gaps 0.5–5 cm between round calculated by using hip and ankle joint positions and metal bars [55]. The bar distances were randomly changed external individual measurements of femur and tibia from one testing session to the next. lengths. Crossing this runway requires that animals accurately Preoperatively, the animals were trained to walk on a place their limbs on the bars. In baseline training and treadmill speed 175 mm s and recorded using a digital postoperative testing, every animal had to cross the grid for video camera. The limb movements of five step cycles at least three times. The number of footfalls errors was were analyzed frame by frame 50 frames s and averaged. counted in each crossing and a mean error rate was The marker coordinates were used to calculate the knee calculated. If an animal was not able to move the hin- position and to quantify flexion extension of the limbs. dlimbs, a maximum of 20 errors was given. The numbers Parameters used were height of the MTP joint, height of of errors counted were also rated in a non-parametric grid the ankle joint and flexion extension in the ankle during walk score: 0–1 error was rated as 3 points, 2–5 as 2 mid swing. For further analysis, data from the different points, 6–9 as 1 point and 10–20 footfalls as 0 points. phases of the step cycle were obtained: Joint angles were measured at the initiation of the swing phase F-phase, in 2.4.5. Narrow beam the middle swing phase E1-phase, and in the phase in The narrow beam test was performed according to the which the paw initially contacts the ground E2-phase. descriptions of Hicks and D’Amato [29]. Three types of Also, the middle stance phase E3-phase was analyzed, beams were used as narrow pathways: a rectangular 2.3-cm which provides most information about the weight support. wide beam, a rectangular 1.2-cm wide beam and a round In order to reduce inter-individual differences, a percentual dowel with 2.5 cm diameter. All beams were 1 m long and ratio was calculated from preoperative values 100. elevated 30 cm from the ground. After training, normal rats were able to traverse the horizontal beams with less 2.4.3. Footprint analysis than three footfalls. When occasionally their feet slipped Footprint analysis was modified from De Medinaceli et off the beam, they were retrieved and repositioned precise- al. [16]. The animal’s hind paws were inked and footprints ly. were made on paper covering a narrow runway of 1 m A scoring system was used to assess the ability of the length and 7 cm width. This ensured that the direction of animals to traverse the beams: 0 was counted as complete each step was standardized in line. A series of at least eight inability to walk on the beam the animals fell down sequential steps was used to determine the mean values for immediately, 0.5 was scored if the animal was able to each measurement of limb rotation, stride length and base traverse half of the beam, 1 point was given for traversing of support. The base of support was determined by the whole length, 1.5 points when stepping with the measuring the core to core distance of the central pads of hindlimbs was partially possible, and 2 points were noted the hind paws. The limb rotation was defined by the angle for normal weight support and accurate foot placement. If formed by the intersection of the line through the print of the scores of all three beams are added, a maximum of 6 the third digit and the print representing the metatar- points can be reached. sophalangeal joint and the line through the central pad parallel to the walking direction. Stride length was mea- sured between the central pads of two consecutive prints 2.4.6. Exploration on each side. To assess their exploratory behavior, rats were tested in To include animals with incomplete weight support in an open field 100380 cm subdivided into nine fields. early postoperative testing sessions, a 4-point scoring Animals were observed individually for 5 min. The system was also used: 0 points were given for constant number of fields crossed during this observation interval dorsal stepping or hindlimb dragging, i.e. no footprint was was noted [6] and a ratio of this performance in com- visible; 1 point was counted if the animal had visible toe parison to the baseline data 100 was modified to a prints of at least three toes in at least three footprints; 2 5-point score. A value of 0–50 was rated as 1 point, points were given if the animal showed exo- or endo- 51–90 as 2 points, 91–110 as 3 points, 111–200 as 4 168 G points and more than 200 of baseline values were rated lesions showed nearly complete disruption of dorsal, as 5 points. lateral and ventral tracts in addition to loss of the gray matter. The loss of gray matter was always more severe 2.4.7. Contact placing response than the loss of white matter. Contact placing is elicited by lightly touching the skin of the dorsal side of the foot without joint displacement 3.2. Behavioral analysis [32,42]. The animal responds by lifting the hind leg and placing it upon the obstacle. 3.2.1. BBB open field locomotion score The animals were held, supported by the upper body, Over-ground locomotion was rated in 137 rats 71 with the hindlimbs hanging free. The dorsum of each foot animals with dorsal hemisection, 66 with contusion using was touched with the edge of a piece of paper. The total the BBB scoring system [4]. Weekly testing began 7 days number of placing responses of ten trials per limb was postoperatively at the end of spinal shock and was noted and the placing rate for individual animals was continued 5 weeks to 3 months after dorsal hemisection or determined from baseline data taken as 100. contusion lesion, respectively. We modified the BBB score by adding earlier occurring features of the score separately 2.5. Statistical analysis see Section 2. The results varied from 6 no weight support to 21 points normal gait. Seventeen rats with Statistical analysis was performed using the Statview dorsal hemisection and 11 rats with contusion showed only software package Abacus Concepts, CA, 1996. The data partial body weight support, absence of toe clearance and are presented in bivariate plots displaying the relationship rotated foot positions during stepping; these animals were between two variables. Regression lines and Pearson’s assigned to the group of low locomotor ability maximal correlation coefficients were calculated. Spearman’s rank BBB score of 12.5 points in both limbs. One hundred and correlation coefficients were computed for paired com- nine animals reached a high locomotor ability. They parisons using ordinal data. For other data, Fisher’s R to Z showed consistent weight support and consistent forelimb– transformation and a ‘Z-test’ were applied to calculate the hindlimb coordination, but often still incomplete toe significance of the correlation coefficients. A P value of clearance. less than 0.05 was chosen as the significance level for all Twelve percent of the animals with a score of 14 or 15 statistical analyses. In the present study, the variable points showed frequent to consistent toe clearance; this number of animals influences the degree of significance. function was considered to occur only at higher stages in All data are presented as the mean6standard error the original BBB scale [4]. We gave an additional point for S.E.M.. the clearance. Another 24 of the animals reached a score of 17–18 points and showed a ‘tail-up’ position, again a function that was rated only at higher score in the BBB

3. Results scale. An additional point was added to include the