Introduction

Moderate-to-severe hypoxia–ischemia (HI) in term infants is associated with a high risk of death or disability,1,2 which is partially improved by therapeutic hypothermia.1 The reasons for incomplete protection are poorly understood. Magnetic resonance imaging (MRI) studies in human term infants with HI encephalopathy (HIE) show HI is associated with considerable damage to white matter tracts as well as gray matter,3,4,5 which is highly associated with poor neurodevelopmental outcome.6,7 Common patterns include damage to the deep nuclei with injury of the posterior limb of the internal capsule (PLIC)5 and a watershed pattern of brain injury involving the white matter and overlying cortex.5

Axonal injury likely mediates at least part of the association between injury of white matter tracts and disability in term and preterm infants after HI brain injury.8,9 Experimentally, axonopathy has been reported in areas of focal necrotic injury 1–3 days after repeated umbilical cord occlusions in near-term fetal sheep and in areas of focal necrotic but not diffuse white matter injury 1–2 weeks after carotid artery occlusion in preterm fetal sheep.10,11 However, the severity of axonal damage after global cerebral ischemia is unclear, and it is unknown whether or to what extent it is attenuated by delayed therapeutic hypothermia.

In near-term fetal sheep, when neural maturation is broadly equivalent to term human infants,12 global cerebral ischemia results in watershed brain injury, affecting the gray and white matter regions.13 We have previously reported that therapeutic hypothermia for 72 h after ischemia partially reduced neuronal loss, oligodendrocyte loss, and microglial induction and improved qualitatively assessed myelin integrity.13,14 Further, compared with 72 h of hypothermia, 48 h of hypothermia was associated with secondary deterioration of electroencephalographic (EEG) power during rewarming and impaired neuronal survival and suppression of microglia.13

In the present study, we examined the hypothesis that the delayed EEG deterioration after 48 h of hypothermia, compared with the standard 72 h clinical protocol, would be associated with impaired axonal protection after reversible global cerebral ischemia in near-term fetal sheep. We used quantitative light microscopy to analyze axonal and myelin integrity in the intragyral white matter tracts.

Materials and methods

Fetal surgery

We have previously reported the physiological data and changes in neuronal, oligodendrocyte, and microglial number in the same cohort of animals as the present report.13 One sham control fetus had to be excluded because of fixation artifacts leading to poor-quality immunochemical labeling. All the procedures were approved by the Animal Ethics Committee of The University of Auckland, under the New Zealand Animal Welfare Act, and the Code of Ethical conduct for animals in research established by the Ministry of Primary Industries, Government of New Zealand. The experiment has been reported in compliance with the ARRIVE guidelines.15

Briefly, 32 time-mated Romney/Suffolk fetal sheep were instrumented at 124 ± 1 days gestation (term = 145 days) under sterile conditions. Food but not water was withdrawn from the ewe 18 h before surgery. At 30 min before the start of surgery, ewes were administered long-acting oxytetracycline intramuscularly (20 mg/kg, Phoenix Pharm, Auckland, New Zealand). Anesthesia was induced with intravenous (i.v.) propofol (5 mg/kg, AstraZeneca Limited, Auckland, New Zealand) and maintained with 2–3% isoflurane in O2 during surgery. The depth of anesthesia, maternal respiration, and heart rate were monitored by trained staff. Ewes were given a constant infusion of isotonic saline infused at approximately 250 mL/h to maintain fluid balance.

After a maternal midline abdominal and uterine incisions, the fetus was partially exteriorized and instrumented for continuous data recording of physiological parameters and monitoring of fetal health. Fetal brachial arteries were catheterized with polyvinyl catheters to monitor mean arterial pressure and for blood sampling. An amniotic catheter was secured onto the fetal shoulder to monitor amniotic pressure. Electrocardiograph electrodes (Cooner Wire Co., Chatsworth, CA) were secured across the fetal chest to record fetal heart rate. Inflatable carotid artery occluders were placed around both carotid arteries, after ligating vertebral–occipital anastomoses. A 3S Transonic ultrasonic flow probe (Transonic Systems, Ithaca, NY) was placed around the carotid artery to measure carotid blood flow. Two pairs of seven stranded stainless steel wire electroencephalograph electrodes were placed on the dura over the parasagittal cortex (10 and 20 mm anterior to bregma and 10 mm lateral) and secured with cyanoacrylate glue. The reference electrode was sewn over the occiput. A pair of electrodes to measure electromyogram activity was sewn over the nuchal muscle. A thermistor was placed on the dura over the parasagittal cortex to measure extradural temperature, and another thermistor was inserted into the esophagus to measure body temperature. A cooling cap made out of silicon tubing (3 × 6 mm, Degania Silicone, Israel) was attached onto the fetal head.

The fetus was returned into the uterus and the incision was closed. Antibiotics (80 mg Gentamicin, Pharmacia and Upjohn, Rydalmere, NSW, Australia) were administered into the amniotic sac. Ten milliliters of 0.5% bupivacaine plus adrenaline (AstraZeneca Ltd., Auckland, New Zealand) was injected into the maternal laparotomy skin incision. All fetal catheters and leads were exteriorized through the maternal flank. The maternal long saphenous vein was catheterized for post-operative maternal care and euthanasia.

Post-operative care

Sheep were housed together in individual metabolic cages with ad libitum access to food and water. The rooms were temperature controlled (16 ± 1 °C, humidity 50 ± 10%) at 12 h light/dark cycle. Antibiotics were administered daily for 4 days i.v. to the ewe (600 mg benzylpencillin sodium, Novartis Ltd, Auckland, New Zealand and 80 mg gentamicin). A continuous infusion of heparinized saline was used to keep fetal catheters patent (20 U/mL at 0.15 mL/h) and the maternal catheter was maintained by daily flushing.

Data recording

Data recordings began 24 h before the start of experiments and continued for the duration of the experiments. Data were recorded and saved continuously for off-line analysis using custom-made acquisition programs LabView for Windows (National Instruments, Austin, TX). Arterial blood samples were taken for pre-ductal pH, blood gas, base excess (ABL800 Flex Analyzer, Radiometer, Auckland, New Zealand), glucose, and lactate measurements (YSI model 2300, Yellow Springs, OH). All fetuses had normal biochemical variables for their gestational ages before the start of the experiment.

Experimental protocols

Fetuses were randomized to sham control (n = 8), ischemia–normothermia (n = 8), ischemia–48 h hypothermia (n = 8), or ischemia–72 h hypothermia (n = 8). At 128 ± 1 day gestation, ischemia was induced by reversible inflation of the carotid artery occluders for 30 min with sterile saline. In the sham control group, carotid artery occluders were not inflated. Fetal blood samples were taken before the occlusion and at 2, 4, and 6 h after occlusion, then daily for the remainder of the experiment.

Cooling was initiated in the ischemia–hypothermia groups 3 h after the end of ischemia and continued until either 48 or 72 h. Cooling was achieved by attaching the cooling coil placed on the fetal head to a pump, which circulated cold water through the cooling coil. Water temperatures were titrated to achieve extradural temperatures between 31 and 33 °C, which are associated with optimal protection in this setting.16 After the cooling period, the cooling machine was turned off, and fetuses were allowed to rewarm spontaneously. In the sham control and ischemia–normothermia groups, the water was not circulated through the cooling coil, which remained in equilibrium with the fetal temperature. Ewes and fetuses were killed 7 days after ischemia with an overdose of sodium pentobarbitone (9 g i.v. to the ewe; Pentobarb 300; Chemstock, Christchurch, New Zealand). Fetal brains were perfusion fixed with 10% phosphate-buffered formalin and embedded in paraffin for immunohistochemistry.

Immunohistochemistry

Diaminobenzidine tetrachloride (DAB) immunohistochemistry

Coronal sections (10-µm thick) were cut starting at the level of the dorsal hippocampus with a microtome (Leica Jung RM2035, Wetzlar, Germany). Two sections from each animal were used. Slides were dewaxed in xylene and rehydrated in decreasing concentrations of ethanol, then washed in 0.1 mol/L phosphate-buffered saline (PBS). Antigen retrieval was performed using a pressure cooker (2100 Antigen Retriever, Aptum Biologics Ltd., Southampton, England) in citrate buffer. DAB immunohistochemical sections were incubated in 1% H2O2 in methanol for 30 min to block endogenous peroxidase activity. Blocking was performed in 3% normal goat serum for 1 h at room temperature. Sections were labeled with 1:200 mouse anti-myelin basic protein (anti-MBP; Cat# MAB381, Lot# NG1726107, Millipore, Burlington, MA) or 1:50 mouse anti-SMI-312 (Cat# BIO837904, Lot# B253553, Biolegend, San Diego, CA) overnight at 4 °C. Sections were incubated in biotin-conjugated 1:200 anti-mouse (Cat# BA-9200, Lot# 2B0324, Vector Laboratories, Burlingame, CA) for 3 h, then incubated in 1:200 ExtrAvidin (Cat# E2886, Lot# 047M4805V, Sigma-Aldrich Pty. Ltd, St Louis, Missouri) for 2 h at room temperature. DAB was added to the sections and the reaction was stopped by washing in PBS, then dehydrated and mounted.

Images were obtained from the intragyral white matter of the first and second parasagittal gyri using light microscopy at ×20 magnification (Nikon eclipse 80i, Scitech Ltd, Preston, VIC, Australia) and at ×63 magnification (Zeiss Axio Imager, Carl Zeiss AG, Oberkonchen, Germany).

Fluorescent immunohistochemistry

The dewaxing, rehydration, and antigen retrieval procedure was carried out as above. Sections were incubated in 0.1 M glycine solution for 1 h at 4 °C to reduce background fluorescence. Sections were labeled with 1:50 mouse anti-SMI-312 (Cat# BIO837904, Lot# B253553, Biolegend) and 1:250 rabbit anti-Iba1 (Cat# ab178846, Lot# GR3185035–3, Abcam, Cambridge, England) or 1:250 rabbit anti-glial fibrillary acidic protein (anti-GFAP; Cat# ab68428, Lot# GR257920–24, Abcam), overnight at 4 °C. Alexa Fluor 488 goat anti-rabbit (Cat# A11008, Lot# 1853312, Invitrogen, Carlsbad, CA) and Alexa Fluor 568 goat anti-mouse (Cat# A11004, Lot# 1862187, Invitrogen) (1:250) were added and incubated for 3 h at room temperature. Hoechst staining 1:100 was applied to sections for 15 s, then mounted.

The sections were imaged from the intragyral white matter of the first parasagittal gyri using the Zeiss Axio Imager M2 at ×40 magnification (Carl Zeiss AG).

Area fraction analysis

Area fraction of MBP and SMI-312 were quantified as the percentage of each image showing positive labeling using Image J (National Institutes of Health, Bethesda, MD). Images were converted into 8-bit; the default auto-threshold was used to quantify MBP and SMI-312 area fraction.

Directionality analysis

The directionality of the MBP and SMI-312 labeling was assessed using the Image J plugin OrientationJ (Distribution).17 This plugin evaluates the orientation of each pixel of the image based on the structure tensor of a Gaussian weighted local area of interest with the Gaussian window size set to σ = 5 pixels (Supplemental Figure S1). This window size allowed the broad directionality of each image to be determined. A frequency histogram of the local orientations (−89 to 89°) of each image was generated. Pixel counts per angle of orientation were expressed as percentages of the area under the curve, and the minimum percentage was subtracted to reduce the contribution of noise in the angle estimations. The angle with the highest percentage was set to 0°. The proportion of labeled structures aligned to this direction was calculated by summing the percentages from −5 to +5°, based on the predominant directions seen in sham controls.

Statistical analysis

Data were analyzed by one-way analysis of variance, followed by Least Significant Difference post hoc test when statistical significance was found (IBM SPSS Statistics 24). Linear regression was used to analyze correlations (GraphPad Prism 7). Statistical significance was accepted when p < 0.05. The sample size was based on the power calculations for the original study.13 In the present study, with population of standard deviation of 6 and alpha of 0.05, this provided at least 90% power to detect a change of ≥10% in axonal directionality.

Results

Blood gas analysis and temperature

There were no significant differences in blood gas, pH, glucose, and lactate measurements during the baseline or recovery period.13 Extradural temperature in the sham control and ischemia–normothermia groups was 39.5 ± 0.1 °C. During cooling, the extradural temperature was significantly decreased to similar nadirs of 32.3 ± 0.3 °C in the ischemia–48 h hypothermia group and 31.3 ± 0.2 °C in the ischemia–72 h hypothermia group (N.S.). Esophageal temperature was reduced to between 37 and 38 °C during cooling for both the hypothermia groups.

Area fraction of MBP and axon labeling

Ischemia was associated with reduced MBP and SMI-312 labeling in the intragyral white matter of the first and second parasagittal gyri compared to sham controls (p < 0.05, Fig. 1). Ischemia–48 h hypothermia and ischemia 72 h hypothermia were associated with a significant increase in both MBP and SMI-312 area fraction compared to ischemia–normothermia (p < 0.05) to levels that were not significantly different to sham control (p < 0.05).

Fig. 1
figure 1

Area fraction of myelin basic protein (MBP) (a) and SMI-312 (b) labeling in the intragyral white matter of the first (IGWM1) and second (IGWM2) parasagittal gyri in the sham control, ischemia–normothermia, ischemia–48 h hypothermia, and ischemia–72 h hypothermia groups. Area fractions of MBP and SMI-312 labeling in the IGWM1 and IGWM2 were significantly lower in ischemia–normothermia compared to sham control. Ischemia–48 h hypothermia and ischemia–72 h hypothermia were associated with significantly higher area fractions of MBP and SMI-312 labeling (IGWM1 and IGWM2) compared to ischemia–normothermia. Data are mean ± SEM. *p < 0.05 vs sham control and ^p < 0.05 vs ischemia–normothermia. Representative photomicrographs of MBP (c) and SMI-312 (d) labeling in the IGWM1 and IGWM2 in sham control, ischemia–normothermia, ischemia–48 h hypothermia, and ischemia–72 h hypothermia groups. Scale bar = 200 µm

Full size image

Directionality

The angles of orientation of MBP and SMI-312 labeling were tightly distributed within a narrow range in the sham control, ischemia–48 h hypothermia, and ischemia–72 h hypothermia groups. Ischemia–normothermia was associated with a broader distribution (Fig. 2). In sham controls, a large proportion of MBP and SMI-312 labeling was aligned to the peak directions in the IGWM1 and IGWM2. There was a significant reduction in MBP and SMI-312 linearity in the ischemia–normothermia group (p < 0.05), whereas both were markedly improved after ischemia–48 h hypothermia and ischemia–72 h hypothermia compared to ischemia–normothermia (p < 0.05), to levels that were not significantly different to sham controls.

Fig. 2
figure 2

Directionality analysis of myelin basic protein (MBP)-positive myelin (top) and SMI-312-positive axons (bottom) showing the distribution of the directionality of the labeling represented as the percentage of the area under the curve per angle for the intragyral white matter of the first (1a, 2a) and second (1b, 2b) parasagittal gyri in the sham control, ischemia–normothermia, ischemia–48 h hypothermia, and ischemia–72 h hypothermia groups. The distribution of MBP and SMI-312 directionality for the sham control and ischemia–hypothermia groups was tightly distributed within a narrow range but was broader for the ischemia–normothermia group. The percentage of the MBP and SMI-312 labeling aligned in the peak directions (−5 to +5 °) (1c, 2c). A large proportion of MBP and SMI-312 labeling was aligned in the peak directions in the sham control, ischemia–48 h hypothermia, and ischemia–72 h hypothermia groups, which was significantly reduced in the ischemia–normothermia group. Data are mean ± SEM. *p < 0.05 vs sham control and ^p < 0.05 vs ischemia–normothermia

Full size image

Correlations with axonal directionality

There was a strong positive relationship between the percentage of SMI-312-positive axons in peak directions and percentage of MBP-positive myelin in peak directions (r2 = 0.78, Fig. 3). Further, the percentage of SMI-312-positive axons in peak directions was significantly associated with the numbers of surviving NeuN-positive neurons in the overlying parasagittal gyri (r2 = 0.42); neuron counts have been previously reported.13

Fig. 3
figure 3

Correlation between the percentage of SMI-312+ axons in peak directions and the percentage of MBP+ myelin in peak directions. p < 0.0001, r2 = 0.78 (a). Correlation between the percentage of SMI-312+ axons in peak directions and the numbers of NeuN+ neurons. p < 0.0001, r2 = 0.42 (b)

Full size image

Qualitative assessment of axonal morphology

The SMI-312-positive axons in the sham control group appeared dense, linear, and uniformly distributed (Fig. 4a). By contrast, the axonal morphology in the ischemia–normothermia group was fine and sparse (Fig. 4b) or discontinuous, with a beaded or spheroid appearance (Fig. 4c). Axonal morphology in the ischemia–48 h hypothermia and ischemia–72 h hypothermia groups was similar to sham control (Fig. 4d, g), but in some animals, subtle abnormalities were still present where the axons appeared sparse or discontinuous (Fig. 4e, f, h, i).

Fig. 4
figure 4

Photomicrographs showing examples of axonal morphology in the first intragyral white matter of the sham control (a), ischemia–normothermia (b, c), ischemia–48 h hypothermia (d–f), and ischemia–72 h hypothermia (g–i) groups. Dense and linear axonal labeling is shown in a, d. Fine and sparse axonal labeling is shown in b, e, h. Discontinuous axons are shown in c, f, i, and the arrows indicate axonal spheroids. Scale bar = 50 µm

Full size image

Axons and inflammation

The dense SMI-312 labeling in the sham control group was associated with abundant GFAP-positive astrocytes with fine processes (Fig. 5). The reduction in SMI-312-positive axonal labeling seen after ischemia–normothermia was associated with dysmorphic astrocytes with thicker processes. However, in the ischemia–48 h hypothermia and ischemia–72 h hypothermia groups, both axonal labeling and astrocyte morphology appeared similar to sham controls.

Fig. 5
figure 5

Representative photomicrographs of glial fibrillary acidic protein (GFAP, green), SMI-312 (red), and Hoechst (blue) in the intragyral white matter of the first parasagittal gyrus in the sham control, ischemia–normothermia, ischemia–48 h hypothermia, and ischemia–72 h hypothermia groups. Scale bar = 50 µm

Full size image

In the sham control group, dense axonal labeling was associated with microglia with ramified morphology (Fig. 6). The reduction in axonal labeling in the ischemia–normothermia group was associated with a preponderance of microglia with amoeboid morphology. By contrast, in the ischemia–48 h hypothermia and ischemia–72 h groups, axonal labeling and microglia morphology were similar to sham controls.

Fig. 6
figure 6

Representative photomicrographs of Iba1 (green), SMI-312 (red), and Hoechst (blue) in the intragyral white matter of the first parasagittal gyrus in sham control, ischemia–normothermia, ischemia–48 h hypothermia, and ischemia–72 h hypothermia groups. Scale bar = 50 µm

Full size image

Discussion

This study demonstrated marked loss of area fraction and the linearity of SMI-312-labeled axons of the intragyral white matter tracts 7 days after global cerebral ischemia in near-term fetal sheep associated with dysmorphic astrocytes around areas of injury and induction of microglia in an amoeboid morphology. We report for the first time that delayed head cooling initiated 3 h after ischemia and then continued until either 48 or 72 h dramatically improved area fraction of axonal labeling and linearity to sham control values, although a subset of fetuses showed some residual axonal morphological abnormalities. Further, hypothermia was associated with suppression of microglial induction and restoration of the morphology of astrocytes and microglia to very similar to sham controls. Interestingly, although there was much greater loss of axonal area than myelin area after cerebral ischemia, myelin linearity was tightly correlated with axonal linearity across all groups. Overall, these data show that hypothermia effectively protects axons from post-ischemic injury, likely, at least in part, by suppressing secondary sterile inflammation and adverse glial responses.

Disruption of myelin and axonal integrity appears to be an important contributor to long-term cognitive and motor disability after HIE, as shown by the consistent correlation between damage in the PLIC and the frontal and parietal white matter tracts on MRI and subsequent poor neurodevelopmental outcomes after neonatal HI.6,7,18 Experimentally, in rabbit kits, antenatal HI at 0.7 gestation was associated with decreased fractional anisotropy in the internal capsule, corona radiata, and corpus callosum and subsequent hypertonia after birth.19

Postmortem studies in human preterm infants with periventricular leukomalacia and in preterm and term infants with neonatal encephalopathy have confirmed severe axonal injury including axonal spheroids.8,9,20 Axonal spheroids are focal swellings containing disorganized cytoskeleton and organelles.21 In ex vivo optic and spinal cord studies, nerve transection was associated with spheroids appearing near the injury within approximately 1 h, that then progresses in an anterograde fashion.22

In this study, we used SMI-312, a pan-marker of neurofilaments, which labels both healthy and damaged axons.23,24 Ischemia was associated with substantial loss of axonal area and gross changes in axonal morphology after 7 days of recovery, including axonal spheroids, consistent with previous studies.10,11 Interestingly, in this study we found diffuse axonal damage throughout the intragyral white matter tracts of the parasagittal gyri in the absence of focal necrosis. Consistent with these findings, Haynes et al. also reported axonal injury in non-necrotic white matter regions in postmortem brain tissue of preterm infants with periventricular leukomalacia.8

Potentially, axons may be injured indirectly, due to ischemic injury of the neuronal cell body, leading to secondary axonal degeneration. Consistent with this hypothesis, we have previously reported considerable neuronal loss after cerebral ischemia in near-term fetal sheep.13 Alternatively, there can be direct injury as a result of mitochondrial dysfunction after reperfusion, and the increase of intra-axonal calcium, oxidative stress, and impaired axonal transport.21,25 This in turn may lead to neuronal cell body injury through Wallerian degeneration, as seen in diseases such as spinocerebellar ataxia and amyotrophic lateral sclerosis.22,26 This possibility is supported, at least in part, by the finding in the present study that axonal linearity was correlated with numbers of surviving neurons in the overlying parasagittal cortex. Nevertheless, the essentially complete normalization of axonal area fraction and linearity after both hypothermia protocols, compared with partial neuroprotection in the overlying cortex, particularly with the 48-h protocol,13 suggests that there is a significant element of direct axonal damage that can be effectively attenuated by therapeutic hypothermia.

In the present study, we showed a tight association between axonal and myelin linearity. Each oligodendrocyte myelinates several axons, and so demyelination due to loss of oligodendrocytes could contribute to axonal dysfunction and potentially degeneration.27 Conversely, axonal injury can lead to degeneration of the surrounding myelin sheath, as shown in the peripheral nervous system where axonal degeneration leads to Schwann cell demyelination.28 It is not possible to be certain which develops first after neonatal HI. Nevertheless, the much greater total loss of axonal area fraction than that of myelin in the present study is highly suggestive that post-ischemic axonal damage plays a critical role.

An inflammatory environment can contribute to delayed axonal injury, through release of pro-inflammatory cytokines and free radicals by activated microglia. For example, in cerebellar organotypic cultures lipopolysaccharide-stimulated microglial activation was associated with demyelination and axonal damage.29 Cerebral ischemia consistently induces sterile inflammation,30,31 as shown in the present study by abundant microglia with an amoeboid morphology. Hypothermia consistently suppresses induction of microglia, as recently reviewed.32 Of note, in the same experimental cohort as the present study, we found that, compared to hypothermia for 72 h, cooling for 48 h was associated with impaired suppression of microglia.13 Given that both protocols similarly attenuated axonopathy, this suggests that the residual microglial induction was not injurious for axons. In support of this concept, in the present study microglia showed an amoeboid, activated morphology 7 days after ischemia, whereas after both hypothermia protocols microglia exhibited a preponderance of ramified morphology that was highly similar to sham controls.

Further, astrocytic responses may also contribute to axonal injury.30,31 Consistent with this hypothesis, in the present study, dysmorphic astrocytes tightly co-localized with damaged axons in the intragyral white matter tracts. We have previously reported that, in the same experimental paradigm, ischemia was not associated with any change in astrocyte number or area fraction after 7 days of recovery.14 However, this presumptively reflects that astrocyte responses change very rapidly after ischemia and so precede frank histological hypertrophy.33

In the present study, both 48 and 72 h of hypothermia improved the area fraction and linearity of myelin and axons and helped to normalize the morphology of reactive microglia and astrocytes. Hypothermia after HI has well-characterized neuroprotective effects, including anti-apoptotic and anti-inflammatory mechanisms.32 As above, axons can be damaged by apoptosis of the entire neuron or axon-specific degeneration, but there is some overlap between these two mechanisms, involving apoptotic proteins such as caspase-9 and caspase-3.34 Hypothermia treatment after ischemia reduces cytochrome c release and activation of caspase-3.35,36 Taken together, the attenuation of apoptotic pathways, reduced inflammation, and sparing of neuronal death may have contributed to reduced axonal injury. In turn, the protection of both axons and oligodendrocytes with both 48 and 72 h of hypothermia likely contributed to improved myelination.13

These data show that 48 h of hypothermia provides maximal axonal protection that is not further improved by longer cooling for 72 h, and so additional interventions will be needed to improve axonal protection. It is important to appreciate that, as previously reported, 48 h of hypothermia was associated with impaired EEG recovery, neuronal survival, and suppression of microglial induction compared with 72 h of hypothermia,13 and so 72 h of hypothermia is required for optimal neuroprotection.37

The present study used OrientationJ17 to quantify changes in the linearity of the axonal and myelin fibers in the first and second parasagittal gyri in a two-dimensional coronal plane. One limitation of this approach is that the technique can only be applied in longitudinal sections of linear white matter tracts. Thus, for example, we were not able to analyze linearity of the periventricular white matter. Diffusion tensor imaging of ex vivo brains would provide greater coverage of the white matter tracts of the brain in three dimensions10 but is more expensive and less accessible. Alternatively, electron microscopy is a well-validated method of assessing ultra-structural integrity of axons in fetal sheep brain tissue10; however, it requires different fixation, and in the present study, gross morphological changes in axonal integrity were able to be detected by immunohistochemistry.

Some fetal sheep showed persistent elements of sparse axonal labeling and axonal spheroids despite a dramatic overall improvement in axonal morphology and area fraction. It is important to appreciate that injury can continue to evolve for weeks or months after the original insult.38 Thus, although it is reassuring that there was normalization of microglial and astrocytic morphology, it is possible that these persisting morphological axonal changes in the present study could be associated with longer-term loss of axonal integrity. Potentially then, these persisting abnormalities may contribute to the long-term cognitive and motor deficits seen in some infants after HI despite treatment with hypothermia.39,40,41,42 It is unclear why some animals in this study showed better preservation of axonal morphology after hypothermia. In principle, it could be due to genetic predispositions, fetal weight or sex, inflammatory load, and variations in neuronal or oligodendrocyte survival. The present study was not statistically powered to disentangle the relationship between such factors and the presence of abnormal axonal morphology.

In summary, the present study demonstrated that delayed treatment with hypothermia continued until 48 or 72 h was associated with preservation of myelin and axonal integrity and normalization of microglia and astrocyte morphology. A subset of animals showed persisting abnormal axonal morphology. Further research is required to elucidate the timing and relationship between oligodendrocyte loss and axonal and myelin disruption, but the present data suggest that both 48 and 72 h of hypothermia protocols effectively protect axons after severe cerebral ischemia in near-term fetal sheep, at least in part, by suppressing secondary sterile inflammation and glial activation.