|Home | About | Journals | Submit | Contact Us | Français|
The early antecedents of cerebral palsy (CP) are unknown but are suspected to be due to hypoxia-ischemia (H-I). In our rabbit model of CP, the MRI biomarker, ADC on diffusion-weighted imaging, predicted which fetuses will develop postnatal hypertonia. Surviving H-I fetuses experience reperfusion-reoxygenation but a sub-population manifested a continued decline of ADC during early reperfusion-reoxygenation, which possibly represented greater brain injury (RepReOx). We hypothesized that oxidative stress in reperfusion-reoxygenation is a critical trigger for postnatal hypertonia. We investigated if RepReOx predicted postnatal neurobehavior, indicated oxidative stress, and if targeting antioxidants at RepReOx ameliorated motor deficits, which included testing of a new superoxide dismutase mimic (MnTnHex-2-PyP).
Rabbit dams, 79% gestation (E25), were subjected to 40-min uterine ischemia. Fetal brain ADC was followed during H-I, immediate reperfusion-reoxygenation, and 4-72 hours after H-I. Endpoints were postnatal neurological outcome at E32, ADC at end of, ADC nadir during H-I and reperfusion-reoxygenation, and area under ADC during first 20-min reperfusion-reoxygenation. Antioxidants targeting RepReOx were administered before and/or after uterine ischemia.
The new MRI-ADC biomarker for RepReOx improved prediction of postnatal hypertonia. Greater superoxide production, mitochondrial injury and oligodendroglial loss occurred in fetal brains exhibiting RepReOx than in those without. The antioxidants, MnTnHex-2-PyP, and Ascorbate and Trolox combination, significantly decreased postnatal motor deficits and extent of RepReOx. The etiological link between early injury and later motor deficits can thus be investigated by MRI and allows us to distinguish between critical oxidative stress that causes motor deficits from non-critical oxidative stress that does not.
In adult stroke (Pan et al., 2007; Brouns and De Deyn, 2009), and myocardial ischemia (Yellon and Hausenloy, 2007; Ovize et al., 2010) significant injury is thought to occur during the reperfusion phase (Jung et al., 2010; Mazighi and Amarenco, 2011; Sanada et al., 2011). This has also been postulated in perinatal stroke (Gazzolo et al., 2009). However, most cases of perinatal injury involve global hypoxia and ischemia (H-I) and reperfusion rather than the focal injury of stroke. Reperfusion injury in fetal brain has been documented in cases of umbilical cord occlusion (Ikeda et al., 1999; Welin et al., 2005). Reperfusion injury in fetal brain has never been shown directly in cases of acute placental insufficiency such as abruptio placenta, possibly due to a lack of suitable biomarkers. Since antepartum hypoxia-ischemia (H-I) is a major cause of cerebral palsy (CP) (Gazzolo et al., 2009), we developed a model of CP involving global H-I and reperfusion-reoxygenation (Tan et al., 2005).
Using fetal MRI in our model, we showed that the nadir of the apparent diffusion coefficient (ADC) in fetal brains predicted postnatal motor deficits in survivors (Drobyshevsky et al., 2007). In some cases, the ADC began to recover towards the baseline immediately after the end of H-I. In others the ADC continued to decline further and reached nadir 5 -15 minutes after the end of H-I. In these cases, the ADC recovery was slower and was associated with a worse postnatal neurobehavioral outcome. Thus, the ADC drop after end of H-I could represent biological changes during the reperfusion-reoxygenation phase which most likely indicates injury to the brain. In this paper we define reperfusion-reoxygenation injury as RepReOx and our primary hypothesis is that RepReOx is an initial trigger for later motor deficits. Studies on white matter injury following cord occlusion in fetal sheep (Welin et al., 2005) suggest a role of free radicals in reperfusion injury. Indirect evidence of oxidative stress has also been found in human brains (Haynes et al., 2005). Hence, our secondary hypothesis is that RepReOx is caused by oxidative stress.
The systematic study of RepReOx represents advancement in elucidating early critical injury that results in adverse neurological outcome in CP. Previous studies did not have the benefit of an in vivo MRI biomarker to link early events to later motor deficits. To test our primary hypotheses, we first quantitatively assessed neurobehavioral outcome due to RepReOx. Then, we investigated if oxidative stress only occurred in RepReOx. Finally, to confirm the critical role of oxidative stress, we tested if subsequent motor deficits could be ameliorated by maternal administration of exogenous antioxidants targeting RepReOx, including a new potent manganese (Mn) porphyrin-based superoxide dismutase mimic.
The Institutional Animal Care and Use Committee of Evanston NorthShore University Healthcare approved all experimental procedures with animals.
The surgical procedure has been described in detail previously (Derrick et al., 2004). In vivo global H-I of fetuses was induced by sustained 40-min uterine ischemia at 25 days gestation (79% term, E25) in timed pregnant New Zealand white rabbits (Myrtle’s Rabbits, Thompson Station, TN). This procedure models acute placental insufficiency at a premature gestation. Briefly, dams were anesthetized with intravenous fentanyl (75 μg/kg/hr) and droperidol (3.75 mg/kg/hr), followed by spinal anesthesia using 0.75% bupivicaine. A balloon catheter was introduced into the left femoral artery and advanced into the descending aorta to above the uterine and below the renal arteries. The catheterized animal was placed inside MR scanner. Body core temperature was monitored with a rectal optic temperature probe (FOTS100, Biopac Systems, CA) and maintained at 37±0.3°C with a water blanket wrapped around the dam’s abdomen and connected to a temperature controlled heating pump. After the dam was positioned in the magnet, the balloon was inflated for 40 minutes causing uterine ischemia and subsequent global fetal H-I. At the end of H-I, the balloon was deflated, resulting in uterine reperfusion and reperfusion-reoxygenation period for fetal brains. After the imaging session, the catheter was removed, femoral artery repaired, and the dam was allowed to recover. MRI of the dam was repeated at 4, 24, and 72 hours after H-I. Six days after H-I, at E31 (normal term E31.5), all fetuses were delivered by hysterotomy to ensure identification of uterine position of each fetus in relation to the previous MRI scans.
Surviving kits at E32 underwent neurobehavioral assessment by two observers, masked to the treatment group assignment, to determine the presence of motor deficits and hypertonia as previously described (Derrick et al., 2004). The kits were categorized as follows:
Adverse outcome included stillbirths and presence of motor deficits. Since the reason of miscarriage was unknown and the exact time of death and presence of motor deficits could not be established in miscarried kits, those kits were excluded from final analysis.
Dams were imaged serially in utero on 3T Siemens Verio scanner using 8-channel knee coil. High resolution single shot fast spin echo T2-weighted images (HASTE, TR/TE 1000/111 ms, slice thickness 3 mm, matrix 256×157, field of view 20 cm with 30-45 axial slices covering all fetuses inside dam) were acquired in axial, coronal and sagittal planes to identify sequential position of each fetus in uterine horns. Additional T2-weighted images (HASTE, TR/TE 1000/111 ms) were taken for anatomical reference in axial plane, with 25-32 axial slices covering all fetal brains. Slice thickness was 4 mm, matrix 256×157, and field of view was 20 cm. Single shot EPI diffusion-weighted sequence with b=0 and 700 s/mm2, TR/TE 6000/119 ms, 1 NEX, were acquired with the same geometry and resolution as the anatomical reference scan. Continuous acquisition of the diffusion weighted images was performed before, during H-I and 20 min reperfusion with 40-sec time resolution. Dams underwent 3 follow up MRI sessions at 4, 24 and 72 hours after H-I to obtain anatomical and diffusion weighted images at these time points.
ADC maps were calculated and image analysis was performed using in-house software, written in Matlab (Natick, MA). The position of each fetus in each uterine horn and corresponding fetal brain was identified using orthogonal single shot anatomical data sets. Polygonal ROIs were placed on the whole fetal brains, and transferred to the corresponding ADC maps. The position and shape of ROI placement was manually corrected for each time point to account for fetal head movement. We were able to track the position of each fetal brain within each uterine horn during first and the follow up MRI sessions and subsequently linked the MRI images to the delivered kits following hysterotomy. The evolution of an individual fetus’ brain ADC response was followed from the initiation of H-I through fetal to the postnatal period with MRI during acute H-I, immediate reperfusion, at 4, 24 and 72 hours after, and associated with postnatal neurological outcome at E32. Outcome measures were ADC at the end of 40 min H-I, nadir of ADC during H-I and reperfusion-reoxygenation, area under the ADC curves during reperfusion-reoxygenation phase (0-20 min after H-I). Furthermore, if ADC declined further after the end of H-I and the nadir occurred in the reperfusion-reoxygenation period we diagnosed presence of RepReOx. We reasoned that the ADC change represented a biological change in the fetal brain, later shown to be associated with injury. A quantitative estimate of RepReOx was obtained from the area of the ADC curve during the first 20 min of reperfusion-reoxygenation period below the previously published threshold ADC value of 0.83×10−3 mm2/s.
Superoxide production in fetal brains was determined in vivo by the conversion rate of a redox-sensitive probe hydroethidine (HE) into the specific product 2-hydroxyethidium (2-OHE+). We defined superoxide production by the formation of 2-OHE+ detected in fetal brain homogenate. Since we expected the duration of free radical burst during reperfusion–reoxygenation to be short and access to fetuses in uterowas limited we chose to inject redox-sensitive probe HE intraperitoneally to fetuses before H-I. In this study we utilized HPLC-electrochemical detection. HE was administered to fetuses intraperitoneally 300 μg/fetus in a concentration of 1 mg/ml in 0.1 M phosphate buffer solution (PBS) containing 20% DMSO (Lewén et al., 2001) through a small laparotomy incision in E25 dams (n= 2 dams, 11 fetuses and 1 dam sham, 4 fetuses). The optimum dose and incubation time of HE to obtain signal was determined in optimization studies by varying HE dose and incubation times before H-I. Baseline fetal brain ADC after injection and before H-I was not different from typical ADC (1.25± 0.15 ×10−3 mm2/s) values without laparotomy. After closing the abdomen, the dams were transferred to 3T MRI scanner for imaging. After H-I, fetal brains were extirpated and tissue frozen. Even after optimization of HE dose and time of the drug delivery there was large variation between fetal brains in 2-OH-E+ concentration between 0 to 10.7 pmol/mg of protein in sham animals. To account for the difference between fetuses in HE accumulation in brains, and non-specific conversion of HE to E+, superoxide production was expressed as a ratio of 2-OH-E+ to the sum of all detectable products, 2OH-E+, E+ and HE.
Mitochondrial function was assessed by flow cytometry and staining for mitochondrial membrane potential. Fetal brains were extirpated after 20 min of reperfusion after H-I. A randomly chosen hemisphere was placed in ice cold Hank’s Balanced Salt Solution (HBSS, GibcoBRL) + 10mM HEPES, then the meninges were removed, and the cortex placed in 0.025% trypsin and incubated on a rotating shaker at 37°C for 45 minutes. The brain suspension was spun at 300 g for 10 minutes, the trypsin aspirated and the cells washed with HBSS before limited trituration (30 times) in Neurobasal Media (GibcoBRL). The brain suspension was passed through a sterile 70-μm filter to produce a single cell suspension. The cellular suspension was diluted to 1×106 cells per ml, incubated at 37°C for 15 minutes with cationic dye JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolocarbo cyanine iodide) (Invitrogen) and assessed immediately on the flow cytometer (FACSCalibur, Becton Dickinson, San Jose, CA) (Derrick et al., 2001). We defined good mitochondrial function in a cell if the cell showed increased mitochondrial membrane potential. JC-1 is a lipophilic, cationic dye that can selectively enter into mitochondria and reversibly change color from green (~529 nm) to red (~590 nm) as the membrane potential increases. Consequently, mitochondrial depolarization is indicated by a decrease in the red/green fluorescence intensity ratio. A separate aliquot of the brain cell suspension from each fetal brain was treated with valinomycin before incubation with JC-1 to validate the area on a scatter plot occupied by cells with low mitochondria membrane potential. The ionophore valinomycin effectively collapses mitochondria membrane potential and causes shift of probe fluorescence from red to green (Figure 1) (Cossarizza and Salvioli, 2001). The ratio of number of cells with high to low mitochondrial membrane potential provided an index of mitochondrial function.
Fetal brains were harvested from E26 fetal rabbits, 24 hours after H-I, and dissociated into single cell suspension using previously described method (Mayer-Proschel, 2001). Cell numbers were counted with BD Aria II with CountBrightTM absolute counting beads (Invitrogen, CA). For each brain, an aliquot of 2 × 105 of cells in 200 μl was stained with primary anti-O4 antibody (Back et al., 2007), identifying oligodendrocytes progenitors, at the concentration of 1:1000 at room temperature for 30 minutes, followed by two washes with PBS and 5% BSA. FITC-conjugated secondary antibody was added at 1:1000 and incubated for 30 minutes followed by two washes. Cells were finally re-suspended in 200 μl PBS solution and run through BD Aria II for analysis. 1 μl of propidium iodide (PI) solution (1mg/ml) was added immediately before analysis to assess cell death. The cut-off gait was set using unstained cells with only 1% above the threshold of the fluorescent intensity. We defined live oligodendrocyte progenitors as cells that stained for O4 and healthy oligodendrocyte progenitors as cells that stained for O4 but were negative for PI.
Since our preliminary studies with HE revealed superoxide generation in fetal brains during H-I, we chose to test the neuroprotective effect of superoxide dismutase mimics that were designed to mimic the action of native superoxide dismutase in a much more efficient manner.
Following 40 min uterine ischemia in pregnant rabbits at E22, newborn rabbit kits manifest motor deficits on a neurobehavioral battery of tests at P1 (E32), very similar to cerebral palsy. We first tested administration of 60 ml of 0.086 and 0.0086 mM Mn(III) meso-tetrakis(N-n-hexylpyridinium-2-yl) porphyrin (MnTnHex-2-PyP) solution (0.12 and 1.2 mg/kg) to rabbit dams or 1.05 mM MnTE-2-PyP, divided in two doses, 30 min before and 30 min after uterine ischemia and compared administration of 60 ml of saline. Both MnTnHex-2-PyP groups had 14 normal, 2 mild, 3 severely affected P1 kits and 0 stillbirths compared to saline with 4 normal, 3 mild, 5 severe and 6 stillbirths from 2 dams/group. Mn TE-2-PyP had 1 normal and 3 severe from one dam (Figure 2). It was not surprising that MnTnHex-2-PyP had better neuroprotection than MnTE-2-PyP because it is more lipophilic, accumulates in mitochondria more readily and is better able to cross the blood brain barrier (Sheng et al., 2011).
Having established feasibility, we chose MnTnHex-2-PyP infused intravenously (in an ear vein) to dam in the dose of 60 ml of 0.085 mM at a rate of 60 mL/hr. In the first group, the entire dose was given after onset of H-I. In the second group, half the dose, 30mL, was given before onset of H-I and the other half after end of H-I. During uterine ischemia, it is likely that the drug would not reach the fetus. The antioxidant MnTnHex-2-PyP possesses exceptional ability to catalyze dismutation of superoxide and reduce peroxynitrite, thus decreasing levels of the reactive species.
The antioxidants, ascorbic acid and Trolox (both from Sigma, MO), were intravenously (in an ear vein) administered to the dam as a bolus of 100 mg/kg Trolox (dissolved in 5 mL/kg 0.9% saline) and 1600 mg/kg ascorbic acid (dissolved in 5 mL/kg 0.9% saline). The bolus was infused after onset of and followed by a continuous infusion of 50 mg/kg/h Trolox (concentration, 5 mg/mL) and 60 mg/kg/h ascorbic acid (concentration, 300 mg/mL) for 45 min (Tan et al., 1996a) after the end of H-I. The control group got a saline infusion with volumes equivalent to the Ascorbate + Trolox experiment. As oxidative stress has been shown to occur relatively late after H-I, after the first 2 hr (Miller et al., 2005; Welin et al., 2005) or after 4-6 hr of reperfusion (Wang et al., 2009), we included a post-H-I treatment in all antioxidant groups. Since our primary end-point was hypertonia, measured 7 days after H-I, we did not want the later oxidative stress to be a factor in comparing hypertonia. The Sham group underwent the same procedure as saline controls but instead of aortic occlusion for uterine ischemia, the balloon catheter was not inflated and there was no H-I to fetuses. Number of dams and fetuses, stillbirths and miscarriages in each group are presented in Table 1.
We found that MnTnHex-2-PyP has paramagnetic properties and utilized its T1 relaxation - shortening effect to determine concentration of the compound in maternal and fetal tissues (n=2 dams and 8 fetuses). T1 relaxation maps were acquired during continuous intravenous infusion of MnTnHex-2-PyP, 30 mL, 0.085 mM, rate 60 mL/hr to dams. 3D VFA sequences (available as a Siemens works-in-progress package) were used to estimate T1 relaxation maps (Deoni et al., 2005). The parameters for the 3D VFA sequence included TR/TE = 15/1.7 ms, matrix size = 192×192×20, slice thickness = 5mm, bandwidth = 210 Hz/pixel. We used a two flip angles, 4° and 25°, acquisition and inline T1 mapping with an acquisition time of 2 min. For minimizing the effects of inherent B1 inhomogeneities, a B1 correction scheme was used. The method involves B1 mapping by comparing spin echo and corresponding stimulated echo (Akoka et al., 1993) and then applying a correction term for the regional flip angles. B1 mapping was performed at low spatial resolution and using a field of view (FOV) that was greater than that used for T1 mapping. The sequence parameters for the 2D multi-slice acquisition were TR/TE/TM=1368/14/14ms, matrix size = 64×64, FOV = 250 mm, slice thickness = 5mm.
T1 maps were acquired every 2 min for 80 min and 4 and 7 hours thereafter. Time course of T1 change relative to the baseline before compound infusion was determined in ROIs placed on dam’s vena cava, liver, placentas and fetal liver and fetal brains. Relaxivity of the MnTnHex-2-PyP was determined in saline and in fresh tissue homogenates from dam’s liver, blood, placentas and fetal liver and brain in a separate session. MnTnHex-2-PyP was added to the tissue in known concentrations, obtained by serial dilution of 0.085 mM of the compound from 1:1 to 1:256 and the samples were scanned within 30 min with the same sequence as for in vivo T1 determination. T1 relaxation times were obtained using standard inversion recovery spin-echo sequence with 12 inversion times from 23 to 2100 ms. The difference between T1 values obtained by the two methods were minimal.
Relaxivity of MnTnHex-2-PyP measured in fresh tissue homogenates at room temperature 20°C, in mM−1*sec−1, was 17.26 in maternal blood, 12.09 in placenta, 17.28 in fetal liver, 10.28 in fetal brain and 11.31 in saline. The relaxivity of the compound was relatively high, allowing determination of absolute concentration in living tissue with short acquisition and presence of motion. The method did not allow accurate determination of the compound concentration in big vessels with fast flow due to T2 shortening effect, so aortic and vena cava concentrations were not determined.
Total antioxidant activity was determined using a fixed time point TRAP assay with modification of the method proposed by Rice-Evans & Miller (Rice-Evans and Miller, 1994). Measurement of antioxidant activity is based on the inhibition by antioxidants of the absorbance of the radical cation of ABTS+. The ABTS+ radical is formed by the interaction of ABTS (600 μM) with ferrylmyoglobin radical species, generated by the activation of metmyoglobin (10 μM) with H2O2 (300 μM). Timed absorbance readings were taken on a spectrophotometer every 7 sec for 4 min, at 25°C. Samples were compared with a standard curve generated from similar absorbance readings obtained from known concentrations of Trolox in isotonic 5 mM PBS, pH 7.4 at 25°C. Total antioxidant activity in Trolox equivalents were determined in fetal blood and brain samples were taken from different fetuses before occlusion (no antioxidants), and at 5, 10 and 15 min of reperfusion.
Micro-vascular blood flow change in fetal brain was measured on a subset of dams (n=3), by microprobe needle (probe 415-230) attached to a laser Doppler flow meter (PeriFlux System 5000, Perimed Inc, OH). After the dam was instrumented, a small 2 cm incision was made over the abdomen and the uterus was palpated for a fetal head. The microprobe was inserted 2 mm into fetal cortex through the uterine wall and retained in place manually until the end of the experiment. The abdominal incision was closed with slight pressure. The relative blood flow in arbitrary units was recorded before, during H-I and during reperfusion-reoxygenation.
Imaging was performed on a subset of dams (n=4), twice for each dam, once before aortic balloon inflation (before H-I) and once 5 min after deflation (end of H-I), by quantifying image intensity changes after contrast injection. The perfusion sequence consisted of nine 6–mm slices with 5 mm gap, acquired in the transverse plane of the dam’s trunk using a fast T1-weighted SPGR sequence. Imaging parameters were: TE 1.1 ms, TR 3.3 ms, FA 25, bandwidth 83.3, 256×64 matrix, 18 cm FOV, 2 sec per time point, 50 time points, no ECG triggering. The acquisition started 20 sec before and continued during and after contrast intravenous injection of a bolus of 0.6 mL containing 0.30 mmol/kg Magnevist (Gadopentetae dimeglumine, Berlex, NJ) through the ear vein, followed by saline flush (3 mL). Bolus injection took about 15 sec. A separate SSFSE scan was taken with the same slice geometry for in-plane anatomical reference to identify fetal brains. The slice package was placed in the center of abdomen, occupied by fetuses. Four to six individual fetal brains were identified in each rabbit dam for perfusion measurements. An index of fetal brain perfusion was estimated as slope of T1 signal enhancement (Li et al., 2000) when the contrast reached fetal brains. Relative changes in fetal brain perfusion were estimated as a ratio of the enhancement slopes analyzed for values obtained before and after H-I. Perfusion scans took about 1.5 min each and were performed between DWI series so relative perfusion changes and ADC time course could be registered for the same fetuses. These perfusion values provide an indirect and non-invasive method of measuring fetal brain perfusion as Magnevist is known to cross placental and blood brain barriers (Taillieu et al., 2006).
To account for within litter data correlation, the differences between groups were tested using generalized estimating equations analysis, followed by post-hoc multiple comparisons with Bonferroni correction using statistical package SAS 9.1. Data presented as means ± standard error of means (SEM). Comparisons of hypertonia and stillbirth rate between group and controls was done using Fisher’s exact test with Bonferroni correction for multiple comparisons.
1) ADC during H-I and reperfusion that did not decline below 1.0 ×10−3 mm2/s which previously predicted a normal neurodevelopment (Drobyshevsky et al., 2007) and was confirmed, as none of fetuses in this group had an adverse neurological outcome. Fetuses in this group were not included in further analysis of outcomes.
2) ADC at end of H-I that immediately recovered towards the baseline. This group constituted the comparison group for the next group with RepReOx and indicated as No RepReOx group.
3) ADC that declined further after the end of H-I. The biological changes underlying this pattern of ADC during reperfusion –reoxygenation period indicate RepReOx:
4) ADC that declined below 0.65×10−3 mm2/s before the end of H-I and did not recover during the first 20 min after end of H-I.
RepReOx occurred in 60.4% of all fetuses and in 65.7% of the fetuses with an ADC decline below 1.0×10−3 mm2/s by the end of H-I (i.e. excluding group 1). RepReOx group accounted for 73.5% of all adverse outcomes (hypertonia or death), compared to 13.2% in fetuses without RepReOx (Fisher’s exact test, p<0.001, Table 2). Miscarriages were not included in the count so the adverse outcome rate may be underestimated. The combination of RepReOx and ADC nadir improved the predictive value of the MRI biomarkers: absence of both biomarkers predicted no adverse outcome, presence of both biomarkers predicted 93.3%, only ADC nadir - 86.2% and only RepReOx - 83% of adverse outcomes.
The rate of hypertonia, as determined by neurobehavioral testing (Derrick et al., 2004), in all surviving kits was 59.4%. The rate of hypertonia was 78.4% in kits with RepReOx and 38.1% in kits without RepReOx (Figure 4). Odds ratio of surviving kits with RepReOx to become hypertonic was 5.5 (confidence intervals 1.6-17.9, Fisher’s exact test, p<0.005). Further examination of the MRI biomarker in Figure 2 showed that hypertonia rate in the RepReOx group proportionally increased with lower ADC at the end of H-I period relative to fetuses in the No RepReOx group.
Fetal brains that exhibited RepReOx had a significant increase in superoxide production (greater 2-hydroxyethidium production) compared to fetal brains without RepReOx, or fetal brains that were taken at the end of H-I (before deflating aortic balloon) or sham control brains (Figure 5A). This is the first direct evidence of increased superoxide production in brain specifically in the immediate reperfusion-reoxygenation period.
The proportion of cells with good mitochondrial function was significantly less in the RepReOx group that in those in the No RepReOx group. This supports the reasoning that the ADC change showing RepReOx represents fetal brain injury (Figure 5B). Mitochondrial dysfunction was greater in fetuses with the most severe H-I Injury (Group 4, 61.6±1.9% cells with intact mitochondrial membrane potential). In view of previous studies linking white matter injury to fetal sheep H-I (Ikeda et al., 1999; Welin et al., 2005), we next investigated oligodendrocyte injury.
The decrease in number of oligodendroglial precursors and immature oligodendrocytes 24 h after H-I further supports the concept of brain injury in RepReOx (Figure 5C).
The SOD mimic, MnTnHex-2-PyP, when given before H-I decreased hypertonia rate to 21.7% (Figure 6, p<0.05 on Fisher’s exact test with Bonferroni correction for multiple comparisons). Similarly administration of Ascorbate and Trolox as a combination (A+T) at reperfusion significantly decreased hypertonia to 20.0% in surviving kits compared to 59.4% in saline control group. In contrast, when MnTnHex-2-PyP was administered at onset of reperfusion (post-treatment), there was no protection.
MnTnHex-2-PyP pretreatment decreased the area under the ADC cutoff during reperfusion-reoxygenation period by 30.0% and Ascorbate and Trolox decreased the area by 65.7% (Figure 7), unlike MnTnHex-2-PyP post treatment.
In attempt to explain the difference in neuroprotection between the antioxidants treatments, we subsequently measured the relative concentration of MnTnHex-2-PyP in fetal tissues, including fetal brain, non-invasively by MRI, exploiting the T1-shortening properties of the manganese compound. Concentration of the compound in maternal blood rose rapidly after beginning of infusion (not shown), as indicated by T1 signal change. Concentration of the compound in placenta rose slowly (Figure 8A) and appreciable accumulation of the compound in fetal brain >0.002 mM was measured at 30-35 min time point. Concentration of the compound in fetal and maternal tissues decreased to almost to baseline levels within 3 hours (Figure 8B), except in maternal and fetal liver. Maternal liver was significantly enhanced on T1 weighted images at 4 hours and even 24 hours after infusion. The delay of the appearance of the compound in fetal brains from the beginning of intravenous maternal infusion can be attributed to partial clearance of the contrast during passage though maternal lungs until it reaches certain saturation level and also placental and fetal blood-brain barriers. Despite lipophilic properties of MnTnHex-2-PyP, accumulation of the compound was not fast enough to prevent free radical damage immediately after reperfusion of ischemic fetuses. The delayed entry into fetal brain explains the lack of effect on reactive oxygen species and the difference between the protection between pre and post administration of MnTnHex-2-PyP.
Ascorbate and Trolox rapidly increased the total antioxidant activity in blood (2 fold at 5 minutes) and in the fetal brain, an increase from the baseline was observed at 5 min and at 15 min of reperfusion using modified TRAP assay (Figure 9).
At 24 hours after H-I there was a significant decrease in ADC in fetuses that became hypertonic compared with non-hypertonic and sham fetuses. No differences were seen at 4 or 72 hours after H-I. However, the ADC at 24 hours was not as good a biomarker for prediction of hypertonia as the earlier biomarkers because the absolute difference between hypertonic and non-hypertonic kits was small, less than 0.02×10−3 mm2/s, and was not correlated with outcome if antioxidants were administered. ADC recovery was significantly higher for MnTnHex-2-PyP groups but not for Ascorbate+Trolox compared to saline (Figure 10).
To determine whether ADC decline in reperfusion phase in fetuses with RepReOx is truly related to reperfusion or is actually due to lack of reperfusion we performed direct measurements of fetal cortical microvascular blood flow with laser Doppler probe (figure 11A). Microvascular perfusion started immediately after deflation of aortic balloon and end of uterine ischemia, rising to the levels of the pre-ischemic state, and followed by a hyperperfusion period less than 10 min after the end of H-I.
We estimated an index of fetal brain perfusion as a slope of T1 signal enhancement in fetal brains after entry of contrast agent into fetal brain. The perfusion estimation was performed before H-I and then again at 5 minutes after H-I. We did not observe any evidence a lack of reperfusion at any ADC value around that time. There was no apparent relationship to the extent of ADC fall at 4-6 min after H-I (figure 11B).
We have determined that certain fetuses undergo critical fetal brain injury following H-I. Using novel MRI biomarkers, RepReOx can be non-invasively identified in vivo, representing injury that is separate from injury that occurs during H-I. A key finding of this study is that the time of occurrence of the nadir of ADC (lowest ADC value) differed in those fetuses undergoing RepReOx (nadir during reperfusion-reoxygenation phase) and those that did not (before the onset of reperfusion-reoxygenation phase). The time of nadir thus, determined our definition of whether the fetuses undergo RepReOx or not. All surviving fetuses undergo the same time of reperfusion-reoxygenation after H-I but may undergo some variable degree of injury during the reperfusion-reoxygenation phase. Only a subset of these fetuses will manifest injury to the extent that it can be detected by ADC as an added drop (defining RepReOx). Whether the injury in this definition of RepReOx is due to only to reperfusion-reoxygenation or due to accumulated injury from H-I and reperfusion-reoxygenation is illustrated in Figure 4. There is a small percentage of fetuses that manifest RepReOx at even high ADC levels at end of H-I. These fetuses could be part of a group that predominantly suffers injury from events in the reperfusion-reoxygenation phase. Most of the fetuses that undergo RepReOx will manifest cumulative injury from both injury in H-I and reperfusion-reoxygenation periods.
The translational implication is that the fetus at high risk for having subsequent adverse outcomes including motor deficits, hence critical brain injury can be identified. To the best of our knowledge, this is the first study to quantify the immediate brain ADC changes after H-I and its effect on postnatal outcome. The current understanding of mechanisms of reperfusion brain injury originates from adult stroke models (Brouns and De Deyn, 2009). These may not necessarily be applicable to antenatal brain injury as mechanisms of cell death are quite distinctive (Ditelberg et al., 1996; McQuillen and Ferriero, 2004; Wang et al., 2009). Previous studies were focused on the delayed changes of ADC (>30 min) of focal H-I in adult (Neumann-Haefelin et al., 2000; Qiao et al., 2002), and postnatal animals (Nedelcu et al., 1999; Meng et al., 2005; Munkeby et al., 2008; Bjorkman et al., 2010) and thus on secondary energy failure.
Reperfusion after ischemia results in the restoration of oxygen and glucose, resumption of oxidative phosphorylation, and normalizes energy demanding physiologic processes (Pulsinelli and Duffy, 1983). Paradoxically it may also result in deleterious biochemical processes in various organs, particularly the heart, and these constitute “reperfusion injury” (Jennings et al., 1960). Permanent blood flow occlusion results in smaller brain infarcts than in those with reversible occlusion (Aronowski et al., 1997; Pan et al., 2007; Yellon and Hausenloy, 2007). However, stroke constitutes a small fraction of perinatal H-I brain injury (Wu et al., 2004), which is mostly due to global H-I. Hippocampal reperfusion injury has been demonstrated in global antenatal H-I from transient umbilical cord occlusion in fetal sheep (Fujii et al., 2003). This model usually results in delayed biochemical changes in reperfusion that constitute secondary energy failure (Tan et al., 1996b; Dean et al., 2006).
Oxidative stress is the most likely explanation for RepReOx, especially in the first 20 min after H-I. Fetuses exhibiting RepReOx have greater oxidative stress (Figure 5A) compared to fetuses with no RepReOx in the same litter (hence, same insult). Targeted administration of antioxidants and MnTE-2-PyP attenuate postnatal hypertonia (Figure 4). Importantly, changes in the first 20 min initiates an irreversible chain of events culminating in motor deficits. Later restorative systems for cellular homeostasis do not completely compensate for the injury marked by RepReOx. Following umbilical cord occlusion of fetal sheep, the major increase in free radical production is delayed and occurs after the first 2 hr (Miller et al., 2005; Welin et al., 2005) although an early phase is also noticed in the first 60-90 min in one study (Miller et al., 2005). Even in umbilical cord occlusion in fetal sheep, only 4 out of 7 fetuses showed an increase in free radicals after 4-6 hr of reperfusion, this increase was associated with greater brain injury (Wang et al., 2009). The new finding of the present study is that for the first time there is evidence of early oxidative stress immediately after global H-I in fetuses and this oxidative stress plays a critical etiological role in the generation of later motor deficits. The failure of post-MnTnHex-2-PyP to prevent postnatal motor deficits suggests that later oxidative stress may not likely play a major etiological role. Mitochondrial injury can lead to bioenergetics’ failure and inability of cells to restore ionic and water balance during reperfusion. MnTnHex-2-PyP, accumulates in mitochondria where it can mimic mitochondrial matrix SOD (Batinic-Haberle et al., 2011). Mechanisms of mitochondrial and metabolic dysfunction include inhibition of the pyruvate dehydrogenase complex by free radicals, loss of mitochondrial NAD(H) through the permeability transition pore, consumption of NAD+ by poly-ADP ribose polymerase 1, and release of cytochrome c (Robertson et al., 2009). Early mitochondrial injury as evidenced by changes in mitochondrial membrane potential was also observed in RepReOx (Figure 5B). The exact biophysical nature of ADC changes in RepReOx is still unclear. ADC changes are linked to the bioenergetics’ status of cells and ability to maintain water and ion homeostasis (van der Toorn et al., 1996; van Pul et al., 2005). However, mitochondrial dysfunction was the greatest in the severe H-I injury group (Group 4) indicating that if anything, the smaller mitochondrial dysfunction in RepReOx was associated with oxidative stress rather than from energy failure alone.
In fetal sheep umbilical cord occlusion there is an initial cerebral hyper-perfusion that peaks at 6 min after end of occlusion followed by hypo-perfusion that reaches nadir at 90 min (Dean et al., 2006). The late hypo-perfusion was not observed with shorter occlusion in later gestation sheep (Fujii et al., 2003). If there was delayed hypo-perfusion in our rabbit fetuses, it should have resulted in a delayed ADC fall but ADC at 4 h was not different. The point is that RepReOx is not likely from either hypo-perfusion or “no-reflow” phenomenon observed after stroke (del Zoppo and Mabuchi, 2003) because the timing is incongruent (Davis et al., 1994).
In our previous study ADC nadir was highly predictive of adverse postnatal neurological outcome (Drobyshevsky et al., 2007). By adding the ADC decline in reperfusion the positive predictive value of early ADC markers for adverse postnatal neurological outcome improves to 93.3% from 83.3%. Remarkably, positive predictive value for normal postnatal neurological outcome with the absence of the both ADC biomarkers was 100%. As a marker, diffusion weighted imaging complements existing MRI and electron paramagnetic resonance (EPR) for redox-imaging and does not require custom-built variable field scanners (Yamato et al., 2009) or injection of blood–brain barrier-permeable redox sensitive probes for EPR (Yokoyama et al., 2004) or MRI (Hyodo et al., 2008). The non-invasive marker with high temporal and spatial resolution enables the investigation of early oxidative stress without the need to sacrifice animals and thus lose the ability to associate delayed neurological sequelae. This allows a direct link to be made from early events to multiple pathways including cell death continuum and inflammation that result in neonatal brain damage. MRI is most useful for investigating not only the early triggers but the critical delayed cellular events as shown in Figure 5C. A detailed study of gray and white matter injury is a future endeavor outside the scope of this manuscript. It is possible that, like the case of oxidative stress, not all delayed events, result in motor deficits.
Postnatal hypertonia increased significantly as the severity of initial H-I injury increased, as indicated by ADC at the end of H-I (Figure 4). Added injury, measured as increased postnatal hypertonia rate with RepReOx given the same ADC at the end of H-I, was also small when severity of insult during H-I, determined by ADC at end of H-I period, was small (Figure 4) and gradually increased with increased severity of H-I insult, suggesting that H-I injury is necessary for RepReOx to occur. The extent of reperfusion injury depends on the severity of initial insult during myocardial ischemia (Ovize et al., 2010). In a piglet model of global H-I the significant correlation between reduction in cerebral perfusion and cerebral ADC during H-I was lost during the reperfusion period (Munkeby et al., 2004). It implies that ADC changes and the corresponding cellular damage during reperfusion is not a yes/no condition but may be a function of degree of and duration of the initial hypo-perfusion in uterine ischemia. Fast recovery of ADC indicates that the initial injury is probably reversible, but delayed ADC recovery is associated with adverse outcome (Miyasaka et al., 2000).
The similar decrease in the postnatal hypertonia rate after pretreatment with MnTnHex-2-PyP and post-treatment with Ascorbate+Trolox (Figure 6) suggests that the neuroprotection depends on presence of antioxidants in the fetal brain when free radical production is highest and less on the therapy’s antioxidant efficiency. The lack of effect with post-treatment with MnTnHex-2-PyP suggests that delayed oxidant production was not a major etiological factor in motor deficits due to antenatal H-I. The development of the future antioxidant agents and therapeutic strategies should target rapid delivery to the perinatal brain and efficacy can be assessed with MRI biomarkers.
Differences in accumulation and retention of antioxidants in fetal brains may explain the delayed recovery of ADC at 24 hr of MnTnHex-2-PyP post-treatment compared to the insignificant effect of Ascorbate+Trolox (Figure 10). The delayed effect of MnTnHex-2-PyP raises the possibility of delayed oxidative injury playing a role in the development of later brain injury.
In conclusion, we provide evidence that the diagnosis of brain injury, occurring during early in the reperfusion-reoxygenation period, can be made using MRI with a further drop in ADC, indicating RepReOx. The brain injury in fetuses with RepReOx adversely contributes to the eventual hypertonia. ADC change is influenced by both oxidative stress and mitochondrial dysfunction. Antioxidants are viable candidates for neuroprotection when given either before or after the onset of H-I but adequate levels need to obtained early in reperfusion otherwise the protective effects are lost. The novel non-invasive RepReOx index, indicative of oxidative and mitochondrial injury, can guide development of drugs targeting critical period of early reperfusion–reoxygenation after H-I.
statement (including conflict of interest and funding sources):
Funding sources: UCP/Hearst foundation (AD), NIH NINDS grants NS062367 (AD), NS063141, NS051402, NS043285 (ST), and NIH NICHD grant HD057307 (MD).