Ethics Statement

All procedures were performed in accordance with NIH guidelines and were approved by Boston Children’s Hospital’s Institutional Animal Care and Use Committee (IACUC, protocol 13-04-2373R). All animals used in the study were anesthetized by either a Ketaset/dexdormitor cocktail (dosages below) or isoflurane and were maintained in a deep state of anesthesia throughout all surgical procedures. Animals were euthanized with an overdosed by isoflurane prior to transcardial perfusion (details below). All efforts were made to minimize suffering.

Spinal Cord Surgery

Adult male Sprague Dawley rats (250-300 g; Charles River Laboratories) were maintained on a 12 h light/dark cycle. Experimental and control cases were generated on a random basis, and the subsequent analyses of behavior and anatomy were blinded. Animals included in the study were generated over a course of 2 years, with no more than two surgeries on a single day. Rats were anesthetized with Ketaset (ketamine, 75 mg/kg, s.c.; Fort Dodge) and dexdormitor (dexmedetomidine, 0.25 mg/kg, s.c.; Pfizer), and a skin incision was made at the T8 level. Muscles were cut in layers along the midline and retracted under a surgical microscope. A laminectomy was performed at the T8 level. Lidocaine-HCl (10 mg/ml; Hospira) was applied topically to the spinal cord to minimize pain and sensation during surgery. The dura mater was opened and removed using #3 jewelers’ forceps and a #11 scalpel blade. The dorsal half of the spinal cord was transected using a #11 scalpel blade to a depth of 1.1 mm. Transection of the dorsal CST was confirmed visually and subsequently verified histologically. Gelfoam was packed into the injury site and muscles were closed in layers using 3-0 silk sutures. Sham-operated animals served as normal controls and underwent all procedures up to the point of the dorsal hemisection and pump placement.

Osmotic Minipumps

Immediately after spinal cord surgery, rats were fitted with Alzet osmotic minipumps to deliver saline or inosine either intracerebroventricularly (i.c.v.) or intravenously (i.v.). Minipumps were prepared the day before for priming per the manufacturer’s instructions (Durect). Animals receiving i.c.v. treatment had a borehole drilled over the lateral ventricle of the right hemisphere (coordinates: 0.08 cm A/P, 0.14 cm M/L relative to bregma) and an Alzet brain infusion needle (Brain Infusion Kit 2; Durect) was inserted to a depth of 5.0 mm and secured using a silicon-based glue. The brain infusion kit was attached to an osmotic pump (Alzet 2002; Durect) that was placed subcutaneously behind the shoulder blades. Animals in all experiments were assigned on a random basis to receive either saline (0.9%, Baxter) or inosine (50 mM in saline, Sigma-Aldrich) at a flow rate of 0.5 µl/hour (~ 0.5 mg/kg body weight/day). Pumps were replaced biweekly.

For intravenous (i.v.) delivery, rats were purchased from Charles River Laboratories with pre-fitted jugular vein catheters. Immediately after surgery, rats were fitted with Alzet osmotic pumps (2ML1, Durect) attached to the catheters to deliver saline or inosine (260 mM, ~ 70 mg/day/kg body weight). In a subsequent dose-response study, rats also received i.v. inosine at either 30 or 100 mM (8 - 27 mg/day/kg body weight). Pumps were changed weekly. To ensure that catheters remained functional, they were flushed weekly during the pump change with heparin lock flush (100 USP Units/ml, APP Pharmaceuticals). Inosine solubility increases with elevated pH, and we adjusted the pH of both the inosine and vehicle (saline) solutions to 9.2 to allow solubilization of the highest concentration used. Intravenous delivery of inosine elevates cerebrospinal fluid concentrations within 45 minutes of beginning infusion (Hurtt et al., U.S. Patent 20090221521). Following pump placement, the skin was closed with 3-0 silk sutures and rats were allowed to recover on a warming pad before being returned to the animal facility. All animals were given Buprenex (0.05 mg/kg, s.c.; buprenorphine; Bedford Laboratories) every 8-12 hours for 72 hr and the antibiotic Enrofloxacin (2.5 mg/kg, s.c., Bayer) daily for one week. Bladders were expressed manually twice daily until voluntary control was restored. Rats received subcutaneous boluses of lactated Ringer’s solution twice daily for 2 weeks to maintain hydration.

Behavioral Testing

Rats were tested for general locomotion using the Basso, Beattie, Bresnahan (BBB) scale [22] and for skilled use of the hindlimbs using the horizontal ladder rung walking test [23]. The BBB was evaluated at 24 hours and 7 days after surgery and then weekly over a 4-week period. Two observers were trained to assess hindlimb function according to standard criteria [22] and scored each behavioral exam independently. The number of rats tested in Study I was: saline (combined i.v. and i.c.v.), n = 14; inosine i.c.v., n = 10; inosine i.v., n = 8. In Study II: saline i.v., n = 14; inosine i.v., n = 20. The same rats were evaluated for skilled use of the hindlimbs using a horizontal ladder rung walking test the week after spinal cord surgery and every week after that over the 4-week period (n values same as above). This task assesses sensorimotor coordination and requires descending cortical control [23]. Animals were placed on a horizontal ladder 14 cm wide and 100 cm long suspended 30 cm above the surface. Rungs were spaced irregularly at 2 - 5 cm intervals to prevent rats from learning a fixed pattern of stepping. Performance was video-recorded and scored later (successful steps/total steps) by lab members blind to animals’ treatments. Unsuccessful steps are as defined by Metz and Whishaw [23], where a hindlimb either misses a rung completely or initially makes contact with a rung but slips during weight-bearing.

Anatomical Tracing

Twenty-eight days after SCI, animals received bilateral injections of the anterograde tracer biotinylated dextran amine (BDA, 10,000 MW, 10% w/v solution in sterile saline; Invitrogen) into 9 sites within the hindlimb motor area under stereotaxic guidance. In the first study, BDA placement was guided by the boundaries of the hindlimb area shown in a standard rat brain atlas [24]. We subsequently discovered, however, that these injections infringed upon the forelimb area, causing considerable labeling of CST axons that normally project to the cervical spinal cord. This labeling made it difficult to assess the sprouting of hindlimb CST fibers after SCI, and we therefore modified the injection coordinates based on an electrophysiological mapping study of the motor cortex [25] (new coordinates in cm with respect to bregma: 0.05 A/P, ± 0.10 M/L; 0.00 A/P, ± 0.10 M/L; 0.00 A/P, 0.0 M/L; -0.05 A/P, ± 0.10 M/L; -0.05 A/P, ± 0.15 M/L; -0.10 A/P, ± 0.10 M/L; -0.10 A/P, ± 0.15 M/L). BDA injections (Nanoject II, Drummond Scientific) were placed at 3 different depths (2.0 mm, 1.0 mm, 0.5 mm, 1 min each) for each injection site. Retrograde labeling of long propriospinal interneurons was carried out by injecting 1 µl of Cholera Toxin Subunit B (CTB; 1%; List Biological) into each side of the lumbar spinal cord (level L2 – L3) using a glass micropipette and Nanoject II (200-300 µm lateral to midline; ~ 400 µm depth). Micropipettes remained in place for 1 min after injections. The skin and muscles were then sutured closed and rats recovered on a warming pad before returning to their cages. Two weeks later, rats were euthanized with an overdose of anesthesia and perfused transcardially with 0.9% saline followed by 4% paraformaldehyde (PFA). The brain and spinal cord were post-fixed in 4% PFA overnight (4°C) and cryoprotected in sucrose (10% for 24 h, then 30% for 24 h). A 2 cm segment of the thoracic spinal cord centered around the lesion site was embedded in gelatin-albumin (Type-A; Sigma-Aldrich) and cut at 40 µm in the parasagittal plane using a Vibratome. The brain and portions of the cervical and lumbar spinal cord were embedded in Tissue-Tek (Sakura Finetek USA) and frozen (–20°C) for cryostat sectioning. Although rats from the first study were not analyzed for axon sprouting, they were analyzed for behavior, lesion depth, and serotonergic reinnervation, which are unaffected by the tracer injections.

Visualization of BDA injections

Free-floating sections through the forebrain (coronal plane, 40 µm thick) were reacted with avidin-biotin complex conjugated to horseradish peroxidase (Vectastain ABC Kit, Vector) followed by a DAB-peroxidase substrate kit (Vector). Sections were mounted onto slides (Superfrost Plus; VWR), dried, rehydrated through graded ethanol solutions, stained with Cresyl Fast Violet (Certified; Ted Pella, Inc.), dehydrated through serial ethanol solutions and xylene, and covered using Permount (Fisher). BDA placement was confirmed anatomically in each rat by reference to a standard brain atlas [24]. The observation that very few CST fibers were labeled in the cervical enlargement of sham-operated controls provides further evidence that our BDA injections did not infringe into the forelimb motor area.

Lesion Depth and Area

The cross-sectional area of the lesion and width of the remaining tissue bridge were measured using NIH ImageJ software (Wayne Rasband, NIH, Bethesda, MD), averaging the results from several sections close to the midline. To further verify that we had transected the CST, free-floating coronal sections at the cervical and lumbar levels (40 µm) were stained to visualize BDA-positive axons as above. Rats were excluded from the study if (i) spared tissue at the center of the lesion was either ≤ 15% or ≥ 50% the width of the spinal cord (measured in adjacent uninjured areas); or if (ii) BDA-positive fibers were seen in the dorsal CST in the lumbar region. These criteria excluded rats with lesions that spared fibers in the principal (dorsomedial) CST or that were too large to allow functional recovery (see below). Lesion area is presented as the percentage of tissue that was lost or necrotic in an area spanning 4 mm on either side of the lesion epicenter.

CST Sprouting

Coronal sections through levels C3 – C5 were stained to visualize BDA-positive axons as above. We identified the midline using the central canal and dorsal median fissure as landmarks, then quantified the number of BDA-positive axon profiles ≥ 50 μm in length within the gray matter at the edge of the dorsal CST, at the midline (M), and at two fixed distances from the midline labeled Zones I and II (see below). Zone I was defined by lines parallel to the midline passing through the lateral edges of the dorsal CST on either side, whereas Zone II included virtual lines at twice this distance from the midline. Results were averaged over 12-20 sections and were normalized by the total number of labeled axons in the dorsal CST for each case. This latter number was established by staining 10 coronal sections through the cervical enlargement with fluorescently tagged strepavidin-biotin complex (Strepaviden Alexa Fluor 647, Invitrogen) and capturing digital images through 3 µm optical sections (Ultraview Vox Spinning Disk confocal microscope). Axons, which appear as dots in the coronal plane, were quantified using Stereo Investigator (MBF Bioscience) and averaged across the 10 sections to determine the number of labeled axons in the CST for each case.

CST contacts onto LPSNs

Free-floating sections through cervical levels C3-C5 were blocked with a solution containing 3% bovine serum albumin (BSA) and 10% normal goat serum (NGS) in PBS with 0.5% Triton X-100, then incubated in a rabbit anti-CTB antibody (1:10,000, Genway) in a similar buffer but with 5% NGS overnight at 4°C. Sections were rinsed and incubated with goat anti-rabbit IgG conjugated to Alexa Fluor 594, 1:500, and Strepavidin conjugated to Alexa Fluor 647, 1:1000 (Invitrogen). Digital images of all CTB-positive cells (typically 4-7 per section, avg. 45/case) were captured in 10-12 sections, and the number of BDA-labeled boutons, defined as swellings at least twice the thickness of the axon [26] falling on or within 20 µm of a CTB-labeled soma or dendrite, were quantified using the Stereo Investigator program. All CTB-labeled somata were counted and results are presented as the number of BDA-positive contacts onto CTB-positive LPSNs per section normalized by the number of BDA-positive axons in the CST and averaged over the experimental group.

Synaptophysin-positive contacts

A separate set of free-floating sections through the cervical enlargement was rinsed in PBS, transferred to a preheated solution of 10 mM sodium citrate (pH 8.5, 80° C, 30 min), then to 3% H₂O₂ in methanol (10 min), blocked with 5% NGS in PBS, and incubated overnight with a monoclonal anti-synaptophysin antibody (1:200, Millipore) in PBS containing 3% Triton X-100 at 4°C. Sections were rinsed in PBS and incubated in an Alexa Fluor 488-conjugated antibody to mouse IgG (made in goat: Invitrogen) in 0.3% Triton X-100 in PBS and for 2 h, RT. Sections were rinsed and stained for BDA in presynaptic axons and CTB in long propriospinal interneurons as above. Sections were then mounted and covered. Stacked images were collected in the vicinity of all CTB-positive cell bodies (Ultraview Vox confocal) and transferred to the Stereo Investigator software. This enabled us to quantify (i) total BDA-positive bouton-like swellings, (ii) total BDA+ boutons that were double-stained for synaptophysin, and (iii) Synaptophysin-positive, BDA-positive terminals that fell on or within 20 µm of CTB-positive long propriospinal interneurons. Results were averaged over 10-12 sections and normalized by the total number of BDA-positive axons in the CST as evaluated above.

5-HT quantitation

Free-floating sections through the lumbar enlargement (L1-3) were rinsed in high-salt buffer (HSB), blocked with 5% NGS in HSB containing 0.3% Triton X-100, incubated with a rabbit anti-5HT antibody (1:40,000, Immunostar), rinsed thoroughly, and incubated with an Alexa Fluor 488-conjugated antibody to rabbit IgG made in goat (Invitrogen). A stacked image of the ventral horn was captured (Ultraview Vox Spinning Disk confocal microscope), and the integrated density of 5HT immunoreactivity was quantified using ImageJ software. Results were averaged over 10-12 sections per case and then over the experimental group.

Statistics

Results are presented as group means ± SEM based on ≥ 6 rats per group. Statistical analyses used Mann-Whitney rank-sum tests and, where appropriate, one-way and two-way ANOVA. Scatter plots and Pearson’s product-moment correlation coefficients were used to evaluate the relationship between behavioral scores and lesion depth. Reported P values are all two-tailed.

Improvement of hindlimb function after inosine treatment

Three-to-four weeks after a dorsal hemisection of the thoracic spinal cord, rats treated with inosine performed markedly better than saline-treated controls on tests of general locomotion and fine-motor coordination. Inosine was equally effective whether administered intracranially or intravenously. The behavioral effects of inosine were seen in two independent studies.

Study I: Intravenous vs. intracerebroventricular delivery.

In the first of two studies, we compared the efficacy of inosine delivered by direct infusion into the lateral ventricle (intracerebroventricularly, i.c.v., n = 5) vs. intravenously (i.v., n = 7). i.c.v. delivery of inosine has been shown to enhance CST sprouting and to improve skilled use of an affected paw after stroke or traumatic brain injury [18,20,21,27]. Clinically, however, i.c.v. delivery is undesirable due to the risk of intracranial infection, and we therefore investigated whether i.v. delivery would be equally effective. Vehicle-treated controls in the present study performed similarly regardless of whether saline was delivered i.c.v. or i.v., and we therefore pooled these animals to increase statistical reliability (final BBB score for i.c.v. saline = 13.9 ± 1.3, for i.v. saline = 13.1 ± 1.1; average = 13.5 ± 0.9; ladder rung walking test score for i.c.v. saline at 4 weeks = 19.9 ± 14.1%, for i.v. saline = 21.8 ± 8.6%; avg. = 21%).

All groups were severely impaired in open-field behavior one day after surgery (BBB score ≤ 3: Figure 1A), possibly reflecting spinal shock and secondary injury. Inosine-treated rats (i.c.v., n = 13; i.v., n = 6) began to outperform controls by two weeks and maintained this superiority for the remainder of the study, averaging 4 points higher on the BBB scale at 4 weeks (Figure 1A: F = 21.35, P < 0.01). Rats performed similarly after 4 weeks regardless of whether inosine was delivered i.v. or i.c.v. (Figure 1A). Rats receiving inosine i.c.v. appeared to recover somewhat faster than those receiving i.v. inosine, possibly due to the greater bioavailability in the former case. However, in our second study (see below), the rate of recovery in the i.v. inosine group was comparable to that of the i.c.v. inosine group in Study I. The difference in performance between inosine- and saline-treated rats after 4 weeks translates into a greater consistency of forelimb-hindlimb coordination, plantar stepping, toe clearance, and positioning of the hindlimb parallel to the direction of movement.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Inosine improves open-field behavior and skilled use of hindlimbs.

Behavioral studies were carried out in 2 separate cohorts of rats. General locomotor function was evaluated using the Basso-Beattie-Bresnahan (BBB) scale, whereas fine motor coordination was tested using the ladder rung-walking test, scoring the percentage of steps in which either hindlimb was placed on a rung without error. A, B. Study 1. Inosine was delivered either i.c.v. (black) or i.v. (blue); saline-treated controls are shown in red. Note the superior performance of animals treated with either i.c.v. or i.v. inosine on the BBB (A) and the ladder rung walking test (B). C, D, Study 2. Inosine and saline were delivered i.v. Inosine-treated rats again performed better than saline-treated controls on both the BBB (C) and ladder-rung walking test (D) at weeks 3 and 4. *P < 0.05, **P < 0.01, ***P < 0.001. Data in all figures represent group means ± SEM (Two-way ANOVA, Bonferroni a priori test).

https://doi.org/10.1371/journal.pone.0081948.g001

Unlike open-field behavior, which is largely independent of the CST [28,29], the ladder rung-walking test requires this pathway [23]. Inosine-treated rats outperformed controls on this test from the second week on (P < 0.01, F = 14.56), and by 4 weeks, they were able to place their hindlimbs correctly on the ladder rungs ≥ 50% of the time (Figure 1B). As with open-field behavior, the final scores of inosine-treated rats on the ladder rung test were nearly identical regardless of whether inosine was delivered i.v. or i.c.v.

Inosine treatment does not affect lesion size

Histological analyses were carried out to investigate whether inosine affects secondary injury and to verify that we had completely transected the CST while leaving enough tissue intact to allow for the formation of detour circuits [29].

Because even a small percentage of CST fibers can mediate skilled behavior [30], it was essential to verify that this tract was fully severed. This was accomplished by (a) visualizing the loss of the CST during surgery; (b) verifying the complete interruption of the CST in parasagittal sections through the lesion site; (c) verifying the absence of BDA-positive axons in coronal sections through the lumbar cord; and (d) excluding rats whose lesions were < 50% the depth of the spinal cord (Figure 2C-E). There were no significant between-group differences in the percentage of rats excluded based on these criteria (Figure 3A: 1 out of 28 saline-treated rats vs. 1 out of 39 inosine-treated rats: _X² = 0.07, P ~ 0.8). Rats were also excluded if lesions were ≥ 85% the depth of the spinal cord based on earlier studies showing that transection of ≥ 90% of the spinal cord prevents recovery of open-field behavior [29]. Our data show a marked drop-off in both open-field behavior and the horizontal ladder rung walking test when lesions exceeded 85% the depth of the cord despite inosine treatment (Figure 4). There were no significant between-group differences in the number of rats excluded using this second criterion (3 out of 28 saline-treated rats vs. 9 out of 39 inosine-treated rats: _X² = 1.12, P ~ 0.3; Figure 3A). To investigate possible between-group differences in numbers of rats excluded in the two groups further, we carried out a Mann-Whitney rank sum test (non-parametric) of the lesion depths of all animals and again found no significant difference between the inosine- and saline-treated groups (P = 0.84). In light of the appreciable motor function seen even in rats with the greatest lesions (Figure 4), it is possible that many of the apparently complete transections actually reflected tissue damage during sectioning or histological processing. A Mann-Whitney rank sum test of lesion depth vs. behavioral data that includes all rats, including those that that fell outside of our lesion criteria, continued to show highly significant differences between inosine- and saline-treated rats (BBB: P < 0.0001; Ladder-rung P < 0.0001). Animals remaining in the study after applying our exclusion criteria had lesions that were, on average, ~ 70% the depth of the cord regardless of treatment (Figure 3B). Saline and inosine-treated groups were also similar to each other in the cross-sectional areas of the lesions (Figure 3C). Together, these results indicate that (a) inosine treatment did not affect lesion size, and that (b) inter-group differences in performance are not a consequence of differences in lesion size. After applying our exclusion criteria, although there were clear inter-group differences on both tests, there were no significant within-group correlations between lesion depth and test performance (Figure 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Interruption of the dorsal corticospinal tract (CST) after spinal cord injury (SCI).

A. Hindlimb CST axons were labeled by injecting biotinylated dextran amine (BDA) into 9 sites in the hindlimb motor area of each hemisphere 4 weeks after SCI. Rats survived 2 more weeks to allow for BDA transport. B. Identity of brain structures (from the atlas of Paxinos and Watson, 1997). C. Parasagittal section through the thoracic cord shows labeled CST axons (arrowhead) rostral to the lesion (dashed lines). Note the absence of labeled fibers distal to the injury. D, E. Transverse sections of the spinal cord at the cervical (D) and lumbar (E) enlargements. Note heavy labeling at the cervical level (D, D') but absence of labeling in the lumbar cord (E, E'). Scale bar, 500 µm.

https://doi.org/10.1371/journal.pone.0081948.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Inosine does not affect lesion size.

A. Depth of lesions for all animals in Studies 1 and 2 (bars show group means; dashed lines show exclusion limits). B. lesion depth after excluding rats in which the lesion was either ≤ 50% or > 85% the depth of the spinal cord. C. Area of lesions (percentage of tissue lost within an 8 mm segment of spinal cord centered at the lesion epicenter). Inosine and saline-treated groups did not differ from each other in lesion depth or area (One-way ANOVA, Dunnet a priori test).

https://doi.org/10.1371/journal.pone.0081948.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Behavioral outcome depends upon treatment but not depth of lesion.

Scattergrams show open field behavior (A) and ladder rung scores (B) as a function of lesion depth. Blue dots represent rats treated with either i.c.v. or i.v. inosine, and red dots are rats treated with saline; dotted lines show the upper and lower bounds of our exclusion criteria. Within these bounds, inosine-treated rats out-performed controls, but there is no significant correlation (n.s.) between performance and lesion size. Performance on both tests falls off for lesions outside the lower limit for inclusion in the study.

https://doi.org/10.1371/journal.pone.0081948.g004

Inosine-treated rats show increased CST sprouting and synaptic contacts

CST axons transected at the thoracic level sprout collateral branches that can form synapses onto long propriospinal interneurons (LPSNs) at the cervical level of the spinal cord, forming a “detour circuit” that can partially restore cortical control to the lumbar level in the case of incomplete SCI [9-11]. We investigated the sprouting of hindlimb CST axons at 4 loci within the cervical spinal cord as shown in Figure 5B: (a) immediately adjacent to the principal bundle of the CST, (b) at the midline (M), (c) at a distance from the midline equal to the width of the dorsomedial CST (Zone I), and (d) at twice the latter distance (Zone II). Axon counts were normalized by the total number of BDA-labeled axons in the dorsal CST for each animal. In normal rats (n = 6), BDA labeling of the hindlimb motor cortex resulted in fewer than 1% of all labeled axons in the dorsal CST extending into the gray matter at the cervical level. Inosine- and saline-treated rats were selected for detailed histological analysis on a random basis and were coded so the person doing the analyses was unaware of the animals’ treatment. Several cases were rejected based on inadequate fixation or inadequate labeling of the CST.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Inosine increases CST axon branching in the cervical spinal cord.

A. Transverse section through the cervical enlargement. B. BDA-labeled CST collaterals were quantified at the border of the main (dorsomedial) CST (B), the midline (M), and at distances I and II from the midline. C - E, SCI alone increases CST sprouting in the gray matter compared to a sham-treated controls (D vs. C), and inosine enhances it further (E). F - H, Quantitation of CST sprouting into the gray matter at the indicated loci. #P < 0.05, ^##P < 0.01 compared to sham-operated controls; *P < 0.05, ** P < 0.01 compared to saline-treated rats with SCI. Axon counts were normalized by the total number of axons counted in the main CST at the cervical level (Mann-Whitney test).

https://doi.org/10.1371/journal.pone.0081948.g005

CST transection and inosine treatment both increased axon sprouting. In the absence of additional treatments, dorsal hemisections of the thoracic spinal cord (n=5) increased the number of hindlimb axons that extended into the cervical gray matter two-fold (Figure 5F, P < 0.05; n.s. in G and H). Over and above this change, rats receiving inosine (n = 7) showed twice as many CST collaterals as saline-treated controls (Figure 5F - H: all P ≤ 0.05). This increase was highest adjacent to the dorsomedial CST, where the number of BDA-labeled axon branches was ~ 5-fold higher than in intact animals, and decreased at greater distances from the CST. Few CST fibers crossed the midline in any group, though there was a small increase after SCI (saline-treated compared to normal: P ≤ 0.05) and a further increase with inosine treatment (P ≤ 0.01, not shown).

We next investigated whether hindlimb CST axon collaterals contact LPSNs. Following the methods of Bareyre et. al. [9], we injected cholera toxin B fragment (CTB) into the lumbar enlargement to retrogradely label LPSNs that project from the cervical level of the spinal cord to the lumbar level, i.e., the level of hindlimb control. We then counted bouton-like swellings made by BDA-labeled CST axons onto CTB-labeled somata or within 20 µm of their visible dendrites (Figure 6). As before, results were normalized by the total number of BDA-positive axons in the CST to account for inter-animal variability in the overall extent of labeling. Axosomatic contacts were detected far more frequently than axodendric contacts, most likely because CTB labeling does not extend far into the dendritic arbor of LPSNs (Figure 6) and because, even if dendritic arbors were labeled, they may extend outside the plane of section. Rats with SCI and saline treatment had 3 times as many BDA-positive boutons that contacted CTB-positive somata as normal controls (P < 0.01). The inosine-treated group had twice as many of these contacts as saline-treated controls (P < 0.01, Figure 6D) and a qualitatively similar, though not statistically significant, increase in boutons within 20 µm of CTB+ dendrites (Figure 6E). Compared to normal rats, rats receiving inosine after SCI had ~ 10-fold more contacts of CST collaterals onto LPSN somata (Figure 6D, P ~ 0.001) and 4-fold more putative axodendritic contacts (Figure 6E, P < 0.05).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Inosine increases CST contacts onto LPSNs.

A. Transverse section of an inosine-treated animal though the cervical enlargement shows position of LPSNs (boxes). B, C. Enlargement of areas in (A) show LPSNs (red) that project to the lumbar enlargement. B', C'. Higher magnification of cells in B and C. Arrowheads show BDA-labeled CST collaterals (green) contacting an LPSN soma (B') and proximal dendrites (C'). D, E, Quantitation of CST synapses upon LPSN somata and within 20 µm of labeled dendrites. SCI alone increases axosomatic synapses compared to sham-operated controls, and inosine increases the number of these to a greater extent (D). Inosine also increases putative axodendritic contacts compared to normal animals (E). *P < 0.05 compared to saline treatment, ^#P < 0.05, ^##P < 0.01 compared to sham-operated controls (Mann-Whitney).

https://doi.org/10.1371/journal.pone.0081948.g006

Conceivably, the increase in BDA-labeled axons contacting CTB-labeled somata or dendrites noted above could reflect increases in the number of LPSNs that get labeled in the cervical enlargement following CTB injections in the lumbar area. Inosine-treated rats showed ~ 15% more labeled LPSNs following lumbar CTB injections than saline-treated controls, but this difference did not achieve statistical significance and is too small to account for the observed 3-fold difference in numbers of contacts (Normal, mean ± SEM = 50.7 ± 2.6; Saline, 42.8 ± 1.2; Inosine, 49.1 ± 1.8; differences not significant, one-way ANOVA). These results also indicate that, for rats that met our inclusion criteria, lesions did not greatly disrupt LPSN projections that course through the ventrolateral funiculus [31] and serve as the likely substrate for functional recovery [9].

We next examined whether BDA-labeled boutons that contact LPSNs express synaptophysin and are therefore likely to represent physiological synapses. Normal rats showed almost no synaptophysin-positive CST contacts onto LPSNs (Figure 7C), whereas rats with SCI treated with saline showed significantly more (P < 0.01). Compared to the latter group, rats treated with inosine after SCI had nearly 4 times as many synaptophysin-positive boutons on CTB-positive LPSNs (Figure 7, P < 0.05).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Inosine increases synaptic contacts between CST collaterals and LPSNs.

A. Spinning-disc confocal image shows BDA-labeled boutons (red) in the cervical enlargement that arise from CST axon collaterals and express the presynaptic protein synaptophysin (green). B. Confocal image shows CST boutons co-localized with synaptophysin (green)-positive contacts upon a CTB-labeled LPSN (blue). Side images show co-localization in the x-z and y-z planes. C. Quantitation of synaptophysin- and BDA-positive boutons on LPSN somata. SCI alone increases synaptic contacts on LPSNs, and inosine causes a further 4-fold increase. Data are normalized by the total number of labeled axons in the main bundle of the CST. *P < 0.05 compared to saline treatment, ^##P < 0.01 compared to normal control (Mann-Whitney test).

https://doi.org/10.1371/journal.pone.0081948.g007

Restoration of serotonergic innervation in the lumbar spinal cord

Quantitation of 5HT immunostaining in the ventral horns of the lumbar spinal cord showed a 60% decline in levels of serotonin immunoreactivity following dorsal hemisections (n = 12). Inosine-treated rats (n = 14) exhibited levels of serotonin immunoreactivity comparable to those of normal rats (Figure 8). All cases used in this analysis had lesions depths that fell within the criteria described above.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. Inosine increases serotonergic input to the lumbar spinal cord.

A. Schematic diagram of the lumbar enlargement. Box delineates area of 5-HT analysis in the ventral horn. B. Integrated intensity of 5-HT immunofluorescence. SCI decreases 5-HT immunofluorescence compared to normal controls. i.v. inosine restores 5-HT levels back to normal (F = 10.93; **P < 0.01 compared to saline treatment, ^†P < 0.05 compared to sham-operated controls: One-way ANOVA, Dunnet a priori test). C - E. Photomicrographs show serotonergic fibers in the lumbar enlargement of sham-operated rats (C) and after SCI in rats treated with i.v. saline (D) or i.v. inosine (E).

https://doi.org/10.1371/journal.pone.0081948.g008

Dose-response study

A dose-response study demonstrated high levels of recovery in rats treated with 60% less inosine than in the above studies. In addition to the dosage tested above (260 mM), we tested i.v. inosine at nominal concentrations of 30 mM (n = 5) and 100 mM (n = 6). For statistical robustness, we included in the analysis rats that received 0 or 260 mM i.v. inosine in Studies 1 and 2 (total n = 17 for saline and n = 16 for 260 mM inosine). At 30 mM, inosine treatment did not improve outcome on either the BBB scale or the horizontal ladder rung-walking test. Although rats treated with 100 mM inosine showed a slow initial rate of recovery, their scores on the BBB scale and ladder-walk after 4 weeks were similar to those of rats receiving 260 mM inosine (P < 0.01 compared to saline-treated controls, both tests: Figure 9).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 9. Dose-response relationship for intravenous inosine.

A, B. Behavioral results 4 weeks after SCI. The effects of inosine reach a plateau at 100 mM in both the open-field (A) and ladder rung walking (B) tests. C, D. Behavioral test results over the 4 week test period. *P < 0.01; **P < 0.01; ***P < 0.01 (One-way ANOVA, Dunnet a priori test (a, b); two-way ANOVA, Bonferroni a priori test (c, d)).

https://doi.org/10.1371/journal.pone.0081948.g009