Neonatal Rat Hypoxia‐Ischemia: Long‐Term Rescue of Striatal Neurons and Motor Skills by Combined Antioxidant‐Hypothermia Treatment

Abstract

Perinatal hypoxia‐ischemia can cause long‐term neurological and behavioral disability. Recent multicenter clinical trials suggest that moderate hypothermia, within 6 h of birth, offers significant yet incomplete protection. We investigated the effect of combined treatment with the antioxidant N‐tert‐butyl‐(2‐sulfophenyl)‐nitrone (S‐PBN) and moderate hypothermia on long‐term neuronal injury and behavioral disability. S‐PBN or its diluent was administered 12‐hourly to rats from postnatal day (PN) 7 to 10. On PN8, hypoxia‐ischemia was induced. Immediately post‐hypoxia, additional S‐PBN and 6 h of moderate hypothermia or additional diluent and 6 h of normothermia were administered. At 1 week, and at 11 weeks, after hypoxia‐ischemia, the absolute number of surviving medium‐spiny neurons was measured in the coded right striatum. In a separate experiment, skilled forepaw ability was investigated in coded 9‐ to 11‐week‐old rats. Normal, uninjured animals were also tested for motor skills at 9‐ to 11‐weeks‐of‐age. The combination of S‐PBN and moderate hypothermia provided statistically significant short‐ and long‐term protection of the striatal medium‐spiny neurons to normal control levels. This combinatorial treatment also preserved fine motor skills to normal control levels. The impressive histological and functional preservation suggests that S‐PBN and moderate hypothermia is a safe and attractive combination therapy for perinatal hypoxia‐ischemia.

INTRODUCTION

Perinatal cerebral hypoxia‐ischemia is a leading cause of acute brain injury which can lead ultimately to cerebral palsy (52). Recent clinical trials in hypoxic‐ischemic (H/I) neonates indicate that moderate hypothermia within 6 h of birth results in modest improvements in the combined outcome of death or severe disability 20, 61. For further improvements in outcome, the combination of hypothermia with other pharmacological interventions may be beneficial.

The process by which neurons succumb to hypoxia‐ischemia is complex and appears to include energy failure, increased release of glutamate, the formation of free radicals and the activation of apoptotic pathways (52). Because of the complex nature of the neurotoxic cascade, the targeting of multiple sites in the cascade may yield an effective treatment. In nature, hibernating animals exhibit an impressive tolerance to trauma and fluctuations in blood flow through numerous mechanisms operating on multiple targets (12). These adaptive mechanisms appear to include an up‐regulation of antioxidant defenses and hypothermia 5, 65. Therefore, the first aim of this study was to investigate whether combined treatment with an antioxidant and moderate hypothermia rescues neurons after perinatal H/I brain injury.

More specifically, this study investigated whether combined treatment with the antioxidant N‐tert‐butyl‐(2‐sulfophenyl)‐nitrone (S‐PBN) (29) and with moderate hypothermia provides short‐ and long‐term rescue of striatal medium‐spiny neurons after acute perinatal hypoxia‐ischemia. S‐PBN was chosen because it had the best therapeutic ratio, out of 11 potentially neuroprotective compounds, for experimental stroke in adult animals (10). The striatum was chosen because it is one of the main sites of injury after acute perinatal hypoxia‐ischemia in both the rat and human 17, 55, 68. In humans, striatal injury is associated with the choreoathetotic movement disorders of cerebral palsy (68). The predominant medium‐spiny neurons of the striatum were investigated because they comprise more than 97% of all striatal neurons 51, 59. The absolute number of surviving striatal medium‐spiny neurons was measured in each animal using modern stereological methods. This allowed reliable comparisons to be made 8, 17, 18, 49.

The second aim of this study was to investigate whether combined treatment with S‐PBN and moderate hypothermia provides long‐term protection of motor skills after acute perinatal hypoxia‐ischemia. The staircase test 46, 47 was used to investigate skilled forepaw ability. Deficits in the staircase test are correlated with the degree of striatal injury 21, 67.

When a new treatment is examined, the probability of finding an effect is usually greater if it is administered before the insult. Thus, pre‐ and post‐treatment with S‐PBN, in combination with immediate post‐treatment with S‐PBN and moderate hypothermia, was investigated. Pre‐ and post‐treatment with an antioxidant could permit the scavenging of oxygen‐free radicals produced during and after hypoxia (3). The inclusion of an immediate post‐hypoxia booster of antioxidant, in combination with moderate hypothermia, may permit the scavenging of oxygen‐free radicals produced during reperfusion (3). Moderate hypothermia slows free radical production (39) which may allow S‐PBN to effectively scavenge the oxygen‐free radicals produced.

MATERIALS AND METHODS

Litters of 10 postnatal day (PN) 6 Sprague‐Dawley rat pups and their dam were obtained from the University of Otago Animal Breeding Station. The day of birth of the pups was designated as PN0. The rats were housed in controlled environmental conditions. Food and water were available ad libitum, except during behavioral testing. All procedures were approved by the Animal Ethics Committee at the University of Otago.

Fifty‐six male pups from 14 litters underwent hypoxia‐ischemia and treatment with either S‐PBN/hypothermia or diluent/normothermia. In the long‐term behavioral study, another seven male rats (ie, three pups from one litter and four pups from a second litter) were handled regularly (see below), but were otherwise free from experimental manipulation. These animals formed the uninjured, normal control group.

S‐PBN

S‐PBN‐treated pups were injected subcutaneously with 100 mg/kg S‐PBN (Aldrich, St Louis, MI, USA) 12‐hourly 32, 34. The injections were administered from 8:45 pm on PN7 until 8:45 am on PN10 (see Figure 1). The S‐PBN‐treated pups were also injected with S‐PBN immediately after exposure to hypoxia on PN8 (see below). Thus, a total of seven injections of 100 mg/kg S‐PBN were administered per S‐PBN‐treated pup. The diluent for S‐PBN was phosphate buffered saline (pH 7.2). Diluent‐treated, control pups were injected with the diluent at the same time‐points as the S‐PBN‐treated pups. Each litter contained two S‐PBN‐treated and two diluent‐treated pups. Each S‐PBN‐treated pup was matched‐for‐weight with a diluent‐treated pup.

Figure 1.

Timeline of experimental procedures for the subcutaneous injections of N‐tert‐butyl‐(2‐sulfophenyl)‐nitrone (S‐PBN) or diluent, the exposure to a hypoxic‐ischemic (H/I) insult, the post‐treatment with hypothermia or normothermia, the stereological analyses and the behavioral testing. PN, postnatal day.

Hypoxia‐ischemia

On PN8, four experimental pups per litter were treated according to a variation of the Rice et al (55) model of acute H/I brain injury during the human third trimester‐equivalent. These methods have been previously described 8, 17, 18. A hypoxic exposure of 1.5 h was chosen as it induces brain pathology with minimal mortality 8, 17, 18.

Hypothermia

Immediately after hypoxia, the rectal core temperature of each pup was measured. A correlation between rectal core temperature and brain temperature in the PN7 rat has previously been established 39, 64. After this measurement, each pup was weighed and then immediately placed in an individual click‐clack box in a waterbath to commence the 6 h of hypothermia or normothermia. The waterbath was maintained at 26°C to achieve moderate hypothermia or 37°C to achieve normothermia. The body weight was used to prepare an appropriate injection volume and then each pup received an injection of S‐PBN or diluent. Each rat was resting on a 26°C or 37°C gel pad during the measurement of rectal temperature, the injection, and the brief transfers to and from a waterbath. S‐PBN/hypothermia rats were processed before diluent/normothermia rats. Rectal temperatures were also measured at 0.5, 3 and 6 h. The pups were then returned to their littermates and dam. From PN7 to PN14 each pup was weighed daily to monitor their general well‐being.

Short‐term study

On PN14, 16 rats from four litters were anesthetized with Fentazin 5 (fentanyl citrate 0.2 mg/kg, azaperone 1.6 mg/kg and xylazine hydrochloride 29 mg/kg; Parnell Laboratories, Auckland, New Zealand) and Hypnovel (midazolam 4.1 mg/kg; Janssen Pharmaceutica, Belgium). Each rat was then perfused intracardially, the brain was dissected and weighed, and each right cerebral hemisphere was embedded in toto in Technovit as previously described (50). Serial 40‐µm coronal sections were then cut through the entire striatum and stained with cresyl violet as previously described (50).

Prior to stereological analyses, all sections were coded. Stereological quantitation was carried out on every eighth section of the brain containing the striatum. The Cavalieri and optical disector methods, as previously described (50), were used to measure the total number of striatal medium‐spiny neurons. Cavalieri's method was used to measure the total striatal reference volume (Vref). The number of medium‐spiny neurons per unit volume of the striatum (ie, the neuronal density or N_V) was measured using the optical disector method. The boundaries of the striatum, and the identification of a medium‐spiny neuron (Figure 2), were as previously defined (50). An estimate of the absolute number (N) of medium‐spiny neurons was then obtained by N = Vref × N_V.

Figure 2.

Light micrograph illustrating cresyl violet‐stained medium‐spiny neurons and a large cholinergic interneuron in the right striatum of a postnatal day 14 hypoxic‐ischemic rat. In addition to the difference in somal size, the medium‐spiny neurons were characterized by a spherical, or slightly ovoid pale, homogeneous nucleus which dominated the somal volume. The nucleus usually contained one prominent nucleolus. The medium‐spiny neurons were also the most prevalent neurons in the field of view because they constitute more than 97% of striatal neurons 51, 59. Other, very rare, medium‐size interneurons had a darker homogeneous nucleus and a nucleolus. Glia were smaller than the medium‐sized neurons and had overt chromatin clumping on the edge, or within, their nucleus, yielding darkly stained cells. Note that the spines of the medium‐spiny neurons cannot be identified after cresyl‐violet staining. Note also that only some neurons are in focus because this photograph was taken at one focal plane in a 40‐µm section.

Estimates of precision, termed the coefficient of error (CE), of the stereological estimates of absolute volume and neuronal number were calculated according to Gundersen et al (24). The CE of the N_V estimates were calculated as 1/(ΣQ^‐)^1/2 (40). A repeated measures analysis of variance (anova) was used to compare the effect of treatment with S‐PBN/hypothermia on body weight. All other quantitative data were statistically compared using a paired Student t‐test.

Long‐term study

In the long‐term study on neuroprotection, 24 rats from six litters were weaned on PN28. Body weight was recorded daily until PN35 and then twice‐weekly until the end of the study. On PN83, each animal was anesthetized with ketamine (165 mg/kg, Parnell Laboratories) and xylazine (16.5 mg/kg, Phoenix Pharm Distributors, Auckland, New Zealand) and intracardially perfused as described above. Each brain was also dissected, sectioned and stained as described above. All sections were coded prior to stereological analyses. Every tenth 40‐µm section that contained the striatum was analyzed using Cavalieri's method. The optical disector method was performed on the same subset of every tenth striatal section. Twelve animals per group were used to increase the statistical power. Statistical analyses were completed as described above.

In the long‐term behavioral study, the H/I pups were coded from PN7 onwards. All H/I and normal pups were weighed regularly and weaned as described above. From 9‐weeks‐of‐age, eight H/I‐S‐PBN/hypothermia‐treated and eight H/I‐diluent/normothermia‐treated animals from four litters, and seven uninjured, normal animals from two litters, underwent behavioral testing of their skilled forepaw abilities. A modified version of the staircase test (46) was used. Prior to testing, all animals were food‐deprived for 48 h. This reduced their body weight to 90% of their free‐feeding weight. Rats were kept at 90% of their free‐feeding weight throughout the 3‐week period of behavioral testing (47). Hence, each rat did not put on any weight during this period. It is necessary to food‐deprive the animals to ensure that they are adequately motivated to enter the staircase apparatus and to attempt to complete the test (47). Food‐restriction to 90% of a rat's free‐feeding weight is standard ethical practice 36, 47 and is considered mild food‐restriction (36).

During training and testing over PN63–81, one staircase only was baited per trial. This allowed the coordinated grasping ability of each forepaw to be assessed independently. During the 5 days of training (PN63–67), each rat was allowed 15 minutes per staircase (ie, side). Over this period, a small piece of chow was placed on each step along with sucrose pellets. The chow served as an olfactory cue because of the novelty and lack of odor of the sucrose pellets. The top two steps were baited with a single sucrose pellet each and the five lower steps were baited with three sucrose pellets each. During the testing period (PN70–81), the interval inside the apparatus was reduced to 5 minutes per staircase and the chow was omitted. Testing was repeated on five consecutive days per week, for 2 weeks. In each trial, the number of pellets successfully consumed, and the maximum stair level from which the pellets were retrieved, was recorded. The number of pellets retrieved and the maximum step reached provide different measures of performance. The maximum step reached is a measure of how far a rat can reach per se, while the number of pellets retrieved is a measure of how far a rat can make a coordinated reach, grasp and retrieval of a pellet (33). At the completion of behavioral testing, the rats were returned to free‐feeding and were sacrificed by intracardial perfusion on PN83.

Motor skills and body weight were statistically compared using a repeated measures anova. Post hoc comparisons were made using Bonferroni's formula and were corrected for the number of comparisons.

RESULTS

Hypothermia

Body temperature over 0–6 h was significantly different between each S‐PBN‐hypothermia‐treated group and its respective control ( Figure 3). At 6 h, the S‐PBN/hypothermia‐treated pups in the short‐term histological study were 5.62°C cooler, on average, than their diluent/normothermia‐treated littermates (ie, 27.92°C vs. 33.54°C, Figure 3A). In the long‐term histological study, the S‐PBN/hypothermic pups were 5.74°C cooler, on average, than the diluent/normothermia‐treated animals at 6 h (ie, 27.92°C vs. 33.66°C, Figure 3B). For the behavioral study, the S‐PBN/hypothermic pups were 5.75°C cooler, on average, than the diluent/normothermia‐treated animals at 6 h (ie, 29.29°C vs. 35.04°C, Figure 3C). Thus, moderate hypothermia [ie, a decrease in temperature of 4–7°C, (69)] was achieved.

Figure 3.

Mean rectal temperatures [±standard error of the mean (SEM)] over the 6‐h exposure to normothermia or hypothermia on postnatal day 8. A. Short‐term study on striatal neuroprotection, B. Long‐term study on striatal neuroprotection, C. Long‐term behavioral study. The temperature differential between the groups over 0 h, 30 minutes and 6 h was statistically significant using repeated measures anova[A. F _1,6 = 14.26, P < 0.009; B. F _1,10 = 46.93, P < 0.001; C. F _1,6 = 43.16, P < 0.001]. S‐PBN, N‐tert‐butyl‐(2‐sulfophenyl)‐nitrone.

Body and brain weights

There was no significant difference in the average body weight from PN7 until PN14, or from PN7–PN83, when the H/I‐diluent/normothermia‐treated and the H/I‐S‐PBN/hypothermia‐treated groups were compared (ie, PN7–PN14, F _1,14 = 0.015, P < 0.904; PN7–PN34, F _1,14 = 0.017, P < 0.899; PN35–83, F _1,14 = 0.003, P < 0.957; anova repeated measures). Both the H/I‐normothermia‐ and H/I‐hypothermia‐treated pups also showed a significant overall increase in body weight from PN7 to PN14 (F _7,98 = 480, P < 0.001) or from PN7–PN83 (ie, PN7–PN34, F _27,378 = 1520, P < 0.001; PN35–83, F _14,196 = 1092, P < 0.001). Thus, combinatorial treatment with S‐PBN/hypothermia did not have any obvious adverse effects on the general health of the animals.

Comparison of the average weight of the right cerebral hemisphere revealed that there was no significant difference between the H/I‐S‐PBN/moderate hypothermic pups and the H/I‐diluent/normothermic pups on PN14 (Table 1). A significant difference in the weight of the right cerebral hemisphere was evident on PN83 (Table 1).

Table 1.

Stereological data for the right striatum, and brain weight, for all experimental conditions. Abbreviations: Vref = reference volume; N_V = neuronal density; N = absolute number; PN = postnatal day; SD = standard deviation; SEM = standard error of the mean; CV = coefficient of variation = SD/mean; S‐PBN = N‐tert‐butyl‐(2‐sulfophenyl)‐nitrone; H/I = hypoxic‐ischemic.

Treatment group	Insult	Number of animals	Weight (g) of the right cerebral hemisphere*****	Vref of striatum (mm3)	NV of medium‐spiny neurons in the striatum (104/mm3)	N (106) of medium‐spiny neurons in the striatum
6 day survival, PN14 Diluent/Normothermia	HI on PN8	8
Mean			0.402	10.032	21.286	2.164
SD			0.083	2.369	2.056	0.640
SEM			0.031	0.896	0.777	0.242
CV			0.205	0.236	0.097	0.296
S‐PBN/Hypothermia	H/I on PN8	8
Mean			0.434	11.449	24.635*	2.812**
SD			0.048	1.047	1.622	0.219
SEM			0.018	0.396	0.613	0.083
CV			0.111	0.091	0.066	0.078
11 week survival, PN83 Diluent/Normothermia	H/I on PN8	12
Mean			0.711	19.535	9.787	1.893
SD			0.114	6.664	1.063	0.635
SEM			0.034	2.009	0.320	0.192
CV			0.160	0.341	0.109	0.335
S‐PBN/Hypothermia	HI on PN8	12
Mean			0.768	23.999***	10.731	2.578****
SD			0.048	4.009	1.207	0.514
SEM			0.014	1.209	0.364	0.155
CV			0.063	0.167	0.112	0.199

^* P = 0.020 compared with HI‐exposed, diluent/normothermia treated control N_V on PN14;

^** P = 0.015 compared with HI‐exposed, diluent/normothermia‐treated control N on PN14;

^*** P = 0.035 compared with H/I‐exposed, diluent/normothermia‐treated control Vref on PN83;

^**** P = 0.004 compared with H/I‐exposed, diluent/normothermia‐treated control N on PN83; **,****Using an unpaired two‐tailed t‐test, P = 0.853 and P = 0.346, respectively, compared with normal, uninjured total number of striatal medium‐spiny neurons of 2.791 × 10⁶ ± 0.084 × 10⁶ (mean ± SEM) in the PN28 Sprague‐Dawley rat (50);

^***** PN14, P = 0.295; PN83 histology, P = 0.022; PN83 behavioral (data not shown), P = 0.967.

Stereology

In the short‐term histological study, the absolute number of medium‐spiny neurons in the right striatum on PN14 was significantly increased after treatment with S‐PBN/hypothermia compared with diluent/normothermia (Table 1). The N_V estimate after treatment with S‐PBN/hypothermia were also significantly higher than after treatment with diluent/normothermia (Table 1). There was no significant difference in striatal volume after treatment with S‐PBN/hypothermia compared with diluent/normothermia (Table 1).

In the long‐term histological study, there was a significant increase in the absolute number of medium‐spiny neurons in the right striatum after treatment with S‐PBN/hypothermia compared with diluent/normothermia (Table 1). The striatal volume in the S‐PBN/hypothermia‐treated group was significantly larger than the diluent/normothermia‐treated group (Table 1). The striatal N_V of the two groups was not statistically significant (Table 1).

In the short‐ and long‐term histological studies, the absolute number of medium‐spiny neurons in the right striatum after treatment with S‐PBN/hypothermia was not statistically different from the absolute number of medium‐spiny neurons in the normal, uninjured right striatum [Table 1, (50)].

The mean CEs for the Vref, N_V and N measurements ranged from 2%–11% (Table 2). The observed mean variance for Vref and N (ie, CE², Table 2) was generally less than half of the total variance for each parameter for the respective group (ie, CV², where CV is the coefficient of variation, Table 1). Thus, reliable estimates were obtained for these parameters 24, 50, 71. When the CE²/CV² ratio was not less than 0.5 (ie, 0.62–1.19 for N_V or N, 1, 2), there was a low biological variance (ie, CV, 50, 62) for the specific parameter (ie, 6.6%–11.2%, Table 1). A higher CE²/CV² ratio is acceptable when statistical outcomes are not jeopardized (62), as was evident in this study (Table 1).

Table 2.

Final mean CE values of three useful parameters from the right striatum of 8 H/I‐S‐PBN/moderate hypothermia and 8 H/I‐diluent/normothermia rats in the short‐term study, and 12 H/I‐S‐PBN/moderate hypothermia and 12 H/I‐diluent/normothermia rats in the long‐term study, based on the formula CE = (1/n × Σi CE_i ²)^1/2. Abbreviations: CE = coefficient of error; Vref = reference volume; N_V = neuronal density; N = absolute number; PN = postnatal day; H/I = hypoxic‐ischemic; S‐PBN = N‐tert‐butyl‐(2‐sulfophenyl)‐nitrone.

CE	Vref*	NV †	N ‡
6 day survival, PN14
H/I‐diluent/normothermia	0.020	0.076	0.077
H/I‐S‐PBN/moderate hypothermia	0.019	0.072	0.073
11 week survival, PN83
H/I‐diluent/normothermia	0.025	0.106	0.111
H/I‐S‐PBN/moderate hypothermia	0.019	0.090	0.091

^* The mean number of points counted to estimate the Vref ranged from 208 to 259 for the four groups. This exceeded the guideline of counting 100 points per structure (24).

^† The mean number of disector neurons sampled per disector volume ranged from 0.760 to 2.04 for the four groups. This generally complied with the guideline of sampling 1–2 neurons per disector volume (71).

^‡ The mean total number of disector neurons sampled per striatum ranged from 82 to 222 for the four groups. This generally complied with the guideline of counting 100–200 neurons per striatum (23).

Behavior

There was a significant difference in the number of sucrose pellets retrieved with the left forepaw (ie, the forepaw contralateral to the ligated artery in the two hypoxia‐ischemia exposed groups) between the H/I‐S‐PBN/hypothermia‐treated, the H/I‐diluent/normothermia‐treated and the uninjured, normal control animals [anova, F _2,20 = 5.130, P < 0.016, Figure 4A]. Post hoc analysis of the mean number of pellets retrieved with the left forepaw on the fifth (final) day of the trials on each of the 3 weeks revealed that both the H/I‐S‐PBN/hypothermia‐treated animals and the uninjured, normal control animals retrieved and ate more pellets than the H/I‐diluent/normothermia‐treated animals (see Figure 4A for P‐values). There was no significant difference between the H/I‐S‐PBN/hypothermia‐treated group and the uninjured, normal control group (see Figure 4A for P‐values). The number of pellets retrieved with the right forepaw between the H/I‐S‐PBN/hypothermia‐treated, the diluent/normothermia‐treated, and the uninjured, normal control animals was not significantly different over the 3 week period [anova, F _2,20 = 1.125, P < 0.344, Figure 4B].

Figure 4.

Mean number of sucrose pellets retrieved [±standard error of the mean (SEM)] with the left forepaw (A.), or the right forepaw (B.), and then eaten from 9‐weeks‐of‐age. Left forepaw, N‐tert‐butyl‐(2‐sulfophenyl)‐nitrone (S‐PBN)/hypothermia‐treated group vs. the diluent/normothermia‐treated group, *P < 0.0015; **P < 0.0015; ***P < 0.0015. Left forepaw, normal, uninjured group vs. the diluent/normothermia‐treated group, *0.003 < P < 0.0075; **P < 0.0015; ***P < 0.0015. Left forepaw, S‐PBN/hypothermia‐treated group vs. the normal, uninjured group; *,**,***P > 0.25.

The deepest step level reached with the left forepaw differed significantly between the H/I‐S‐PBN/hypothermia‐treated, the H/I‐diluent/normothermia‐treated, and the uninjured, normal control animals [anova, F _2,20 = 6.417, P < 0.007, Figure 5A]. Post hoc analysis of the deepest step level reached with the left forepaw on the fifth (final) day of the trials on each of the 3 weeks revealed that both the H/I‐S‐PBN/hypothermia‐treated animals and the uninjured, normal control animals reached deeper than the H/I‐diluent/normothermia‐treated animals (see Figure 5A for P‐values). There was no significant difference between the H/I‐S‐PBN/hypothermia‐treated group and the uninjured, normal control group (see Figure 5A for P‐values). The deepest step level reached with the right forepaw did not differ significantly between the S‐PBN/hypothermia‐treated, the diluent/normothermia‐treated, and the uninjured, normal control animals [anova, F _2,20 = 0.326, P < 0.726, Figure 5B].

Figure 5.

Mean maximum step level from which a sucrose pellet was retrieved [±standard error of the mean (SEM)] with the left forepaw (A.), or the right forepaw (B.), from 9‐weeks‐of‐age. Left forepaw, N‐tert‐butyl‐(2‐sulfophenyl)‐nitrone (S‐PBN)/hypothermia‐treated group vs. the diluent/normothermia‐treated group, *P < 0.0015; **P < 0.0015; ***P < 0.0015. Left forepaw, normal, uninjured group vs. the diluent/normothermia‐treated group, *0.015 < P < 0.003; **P < 0.0015; ***P < 0.0015. Left forepaw, S‐PBN/hypothermia‐treated group vs. the normal, uninjured group, *P > 0.15; **,***P > 0.20.

DISCUSSION

To our knowledge, this is the first study to report the long‐term rescue of striatal neurons and motor skills by combined antioxidant/hypothermia treatment for neonatal hypoxia‐ischemia. The extent of neuroprotection and behavioral rescue was to normal adult levels.

Methodological considerations

A strength of this study was the use of modern stereological methods to directly measure the absolute number of surviving neurons. The use of indirect measures of neuroprotection, such as the N_V, striatal volume or brain weight, can lead to misinterpretations of biological processes 8, 17, 18, 49. For example, if the N_V, striatal volume or brain weight had been used as the sole measure of neuroprotection in the current study, the efficacy of S‐PBN, in combination with moderate hypothermia, would have been undetected in either the short‐ or long‐term experiment. The short‐ and long‐term effectiveness of combined antioxidant‐hypothermia treatment was only evident when the absolute number of surviving neurons was compared.

The usefulness of measuring the absolute number of surviving neurons was also evident when comparisons were made with qualitative assessment of striatal injury. In the diluent/normothermia‐treated animals, there was qualitative evidence of neuronal loss in the right striatum in two of eight animals on PN14 and two of 12 animals on PN83. There was a quantitative decrease in medium‐spiny neurons in the right striatum of six of these eight animals on PN14, and in all 12 animals on PN83. Qualitative evidence of gliosis in the right striatum was seen in three of eight animals on PN14 and three of 12 animals on PN83. A qualitative decrease in medium‐spiny neurons or gliosis in the right striatum was not evident in S‐PBN/hypothermia‐treated animals on PN14 nor PN83. There was a quantitative decrease in medium‐spiny neurons in the right striatum in one of eight, and four of 12, S‐PBN/hypothermia‐treated animals on PN14, and PN83, respectively. Quantitation was needed, therefore, to detect moderate striatal injury.

Male rat pups were investigated to enable comparisons with the relevant literature [eg, 8, 9, 10, 13, 17, 18, 21, 31, 38, 42, 43, 45, 53, 60, 63, 74]. Adult male rats were used in the behavioral studies as size differences between the genders may affect aged‐matched performance on the staircase test. Future studies should investigate the neuroprotective potential of combined antioxidant/hypothermia treatment for neonatal hypoxia‐ischemia in female rats.

Hypoxia‐ischemia was induced on PN8 to enable comparisons with previous antioxidant studies 8, 31. The previous experiments required the surgical implantation of micro‐osmotic pumps on PN7, to allow continuous infusion of an antioxidant into the striatum 8, 31, and then further surgery for hypoxia‐ischemia on PN8. As we designated the day of birth as PN0, rather than PN1 28, 37, our PN8 rats could be considered as PN9 rats. PN9 rats may be more term‐equivalent (25) than the PN7 rats that have been extensively used in the Rice et al model of neonatal H/I brain injury (57). The use of PN9 rats may increase the relevance of this study to H/I brain injury at term.

Because of the novelty of this study in the immature rat, a 12‐hourly dose of S‐PBN in general (Figure 1), and the specific dose of 100 mg/kg subcutaneously, was based on previous studies in adult rats 32, 34, 35. In these adult studies the neuroprotective efficacy of S‐PBN was evident 32, 34, 35. A more recent study in the adult rat indicated that, after systemic administration of S‐PBN, the half‐life in the plasma was 9 minutes (45). This compared with a half‐life in the plasma for PBN of 3 h (7). In spite of reportedly unfavorable pharmacokinetics, S‐PBN reduces free radical production several hours after post‐injection and induces significant neuroprotection in a wide variety of experimental models of brain injury [for a review of experiments, see (53)]. For example, a dose of 100 mg/kg of S‐PBN significantly decreased free radical production in the adult brain at 2‐h postinjection (60). S‐PBN was also more effective than PBN in reducing infarct volume after focal ischemia in adult rats (38). A degradation product of PBN was also more potent than PBN in delaying senescence in lung fibroblast cultures (2). The degradation product of PBN was effective at concentrations at least 20 times lower than PBN (2). Taken together, these findings may indicate that a degradation or metabolic product of S‐PBN, with a much longer half‐life than S‐PBN, contributes to the documented effective neuroprotection and effective free radical scavenging at hours post‐injection.

During the 6‐h normothermic exposure, the average rectal temperature of approximately 34°C was 3°C cooler than the average “normothermic” rectal temperature reported in earlier studies 4, 6. The present study could therefore represent mild hypothermia in the normothermic group. However, the normal rectal temperature of a rat pup can be as low as 33°C (56). Most researchers have chosen the same temperature range that is applicable to humans (ie, 36–37°C) for the ease of transferring the results. The cooler normothermic temperature of 34^oC in the current study is likely to have contributed to the mild‐to‐moderate striatal injury observed. It is noteworthy, however, that injury is evident at 34°C because the total number of striatal medium‐neurons in the H/I right striatum was 1.8 million (9) compared with the normal total number in the right striatum of 2.8 million (50).

Each rat's performance on the staircase test was investigated at 9–11‐weeks‐of‐age to investigate the longer‐term effect of the combinatorial treatment. The rats were approximately 400 g at this age, which was within the recommended weight range of 250–500 g for the standard staircase apparatus [http://www.psych.ualberta.ca/~fcolbour/staircase.pdf; 42, 43]. As the minimum weight of 250 g was attained when the rats were 7‐weeks‐old, testing could begin at an earlier age in future studies.

The training and the testing on the staircase test covered a recommended 3‐week period 14, 66. Three weeks allowed time for the response of each individual to stabilize and for an increased probability of detecting statistical significance if present. For example, the repeated measures anova for the number of pellets eaten with the left forelimb was not statistically significant after 1.5 weeks (P < 0.066) but was significant after 2 weeks (P < 0.037) and after 3 weeks (P < 0.016). Statistical significance by repeated measures anova across all three groups at 2 weeks for the left forelimb (ie, for the number of pellets eaten, and the step level reached) may indicate that a shortened training and testing period of 2 weeks may suffice in future studies.

Biological considerations

The mild‐to‐moderate brain injury (ie, a loss of 22% to 32% of the right striatal medium‐spiny neurons after unilateral neonatal hypoxia‐ischemia) resulted in deficits in fine motor control of the left forepaw. To our knowledge, this is the first study to show that a small, covert injury to the striatum (ie, an injury that is without conspicuous histological damage) contributes to significant long‐term functional injury of the forelimb in the extensively used Rice et al rat model of neonatal H/I brain injury.

The demonstrated motor deficits on the staircase test are likely to involve the striatal medium‐spiny neurons through their connections. As the principal neurons of the first and largest basal ganglion, the striatum, these medium‐spiny projection neurons receive a massive input from the cerebral cortex and project via their axons to other basal ganglia. Their axonal projections are thought to regulate the activity of the substantia nigra reticulata and the entopeduncular nucleus, which then convey the output of the basal ganglia to the thalamus and onto the supplementary motor area and premotor area of the frontal cortex (22). Neurons in these cortical areas contribute to the corticospinal and corticobulbar tracts (30), which in turn generate movement in the limbs, and head and neck, respectively. As both the input and output of the basal ganglia are intimately related to motor (and other) areas of the cerebral cortex, the basal ganglia are thought to control the direction and velocity of movement that is initiated by the cerebral cortex 1, 73. The death of striatal medium‐spiny projection neurons after H/I brain injury is likely to impair these motor skills.

The deficits in fine motor control of the left forepaw of the diluent/normothermia‐treated animals, after a right‐sided H/I brain injury, reflects the known anatomy of the rat striatum and corticospinal tract. The right striatum predominantly influences the ipsilateral cerebral cortex, including corticospinal neurons (19). The corticospinal tract is largely crossed at the level of the medulla, with a minor uncrossed component (54). Thus, the right injured side of the striatum exerts dominant control over movement on the left side of the body.

It could be argued that the lack of improvement in the ability of the diluent/normothermia‐treated animals to reach and grasp with the left forepaw, when compared with the S‐PBN/hypothermia‐treated animals, primarily reflects a deficit in learning. For instance, the rats may have failed to progress past their early achievement because they struggled to learn the task. However, all of the rats had tasted sucrose pellets by the end of training on the second day. Once the sucrose pellets were tasted, the rats invariably spent the majority of the training and testing period in the testing box and remained determined to attempt the task, regardless of their success. This suggests that the deficit was primarily due to motor ability rather than learning. This perseverance was detected because the animals were watched during each trial.

A right‐sided H/I striatal injury usually produces nominal damage to the left cerebral hemisphere (55). For example, on PN18, the total number of striatal medium‐spiny neurons in the left hemisphere (ie, contralateral to the H/I injury) is not significantly different from normal (17). Consistent with this, there was no significant difference in the reaching and grasping abilities of the right forepaw when the H/I‐diluent/normothermia, the H/I‐S‐PBN/hypothermia, and the control, uninjured animals were compared. A trend for a difference in the behavioral data with the right forepaw was possibly evident, however, in the diluent/normothermia‐treated H/I animals vs. normal control animals (Figure 4B). A similar trend, without statistical significance, has been reported for the ipsilateral forepaw after unilateral striatal injury in adult rats [see figure 3B in reference (14)]. Two reasons may account for this apparent trend. First, the output pathways of the basal ganglia are not completely lateralized (19). Striatal injury on the right side may therefore impair right ipsilateral motor skills to a small extent. Second, an impairment in motor skills may be a dysfunction of postural adjustment on the contralateral side, leading to a dysfunction of balance, which consequently impairs the control of reaching and pellet retrieval with the ipsilateral or “good” limb (14).

In the diluent/normothermia‐treated animals, there was qualitative evidence of neuronal loss in the right cerebral cortex and the right thalamus in three of eight animals on PN14 and two of 12 animals on PN83. Injury to the thalamus is thought to contribute only moderately to the observed motor deficits in the staircase test (13). It is possible that injury to the cerebral cortex, when it included the forelimb region as in the above animals, may contribute to the observed motor deficits in the staircase test 13, 47. Yet, the vast majority of diluent/normothermia‐treated animals did not exhibit lesions in the forelimb cortical region. In addition, long‐lasting deficits on the staircase test require full thickness lesions of the forelimb region of the rat cerebral cortex (63). These data suggest that focal striatal injury, rather than thalamic or cortical injury, contributed predominantly to the observed motor deficits.

Individual performance in the staircase test was not correlated with striatal injury in the current study. This is because the behavioral animals were not included in the histological assessment as it was unclear what impact behavioral testing may have on the total number of surviving striatal medium‐spiny neurons. The extra handling involved in the behavioral testing, the novel testing environment and the staircase task itself may represent a form of enrichment (ie, synaptic, dendritic or neurotrophic) that may in turn affect striatal neuronal survival after moderate brain injury. To avoid this potential confound, the effect of treatment with S‐PBN/hypothermia was investigated in behaviorally naive animals. This strategy was also adopted because of the time‐consuming nature of stereological studies.

As treatment with S‐PBN/hypothermia rescued the total number of striatal medium‐spiny neurons to normal levels in the current study, it is now known that behavioral testing is unlikely to be a confounder. This is because it is biologically unlikely that rat striatal neuronal numbers can be increased above normal levels. Further stereological analysis of the total number of striatal medium‐spiny neurons in the behavioral animals would now permit an analysis of the correlation, if any, between the loss or preservation of striatal medium‐spiny neurons and fine motor skills in each individual.

S‐PBN is a free radical trapping nitrone that is likely to react with oxygen‐free radicals that are produced extracellularly (38) as a result of perinatal hypoxia‐ischemia. Injury to the blood–brain barrier during perinatal hypoxia‐ischemia allows the entry of S‐PBN, which has a reduced lipid solubility, into the brain 11, 74. Oxygen‐free radicals may be released into the extracellular space of the injured brain by the membrane‐associated β‐nicotinamide adenine dinucleotide phosphate (NADPH) oxidase, which is abundant in macrophages, neutrophils and resident microglia(15). S‐PBN may also scavenge extracellular oxygen‐free radicals produced by endothelial cells at the blood–brain barrier 26, 38. The statistically significant, long‐term neuroprotection reported in this study suggests that a high concentration of extracellular oxidants may contribute significantly to the death of striatal medium‐spiny neurons in perinatal hypoxia‐ischemia. By contrast, mitochondrially derived oxidants may not contribute significantly to the death of these neurons 8, 31. Nitrones like S‐PBN, and moderate hypothermia, may also act on neuroinflammatory and signal transduction mechanisms to achieve neuroprotection 16, 44, 48, 72.

Antioxidants, including vitamins E and C and allopurinol, have been used extensively in clinical studies involving babies and pregnant women without major adverse effects 3, 27, 58. NXY‐059, a nitrone that is structurally related to S‐PBN, is also well tolerated in healthy young and elderly humans and in stroke patients 41, 70. The safety of S‐PBN in animal studies (10), and its potential clinical relevance as an antioxidant, led to its use in the current study.

The significant neuroprotection on PN14 in this combinatorial study was not evident in previous singleton studies that replicated treatment with S‐PBN alone (31) or moderate hypothermia alone (9). The absolute number of medium‐spiny neurons in the right H/I striatum on PN14 after treatment with S‐PBN alone was 1923000 ± 261000 (mean ± standard error of the mean, n = 8 per group) vs. 2246000 ± 228000 neurons in the diluent‐treated condition (31). The absolute number of medium‐spiny neurons in the right moderately injured H/I striatum on PN14 after treatment with moderate hypothermia alone (ie, 29^oC) was 1891000 ± 210000 vs. 1845000 ± 92000 neurons in the normothermia‐treated condition [ie, 34^oC, n = 5 per group, adapted from reference (9)]. These data highlight the effectiveness of combinatorial treatment with S‐PBN and moderate hypothermia.

Whether combinatorial treatment with S‐PBN and moderate hypothermia is protective against more severe H/I striatal injury (3) warrants investigation. Further research is also needed on the effects of post‐treatment alone with S‐PBN and moderate hypothermia. This is because exposure to perinatal hypoxia‐ischemia can be unpredictable. This situation requires the investigation of a treatment that is effective when administered after hypoxia‐ischemia.

In summary, a small, covert injury to the striatum contributes to significant long‐term functional injury of the forelimb in the extensively used Rice et al rat model of neonatal H/I brain injury. The combination of pre‐ and post‐treatment with S‐PBN and post‐treatment with moderate hypothermia protected striatal medium‐spiny neurons, and preserved motor dexterity, to normal levels in an established rodent model of perinatal hypoxia‐ischemia. The impressive histological and functional preservation, well into adulthood, suggests that further experimental observations in larger animals, and possibly in the clinic, are warranted.