Hypothermia Treatment after Hypoxia-Ischemia in Glutathione Peroxidase-1 Overexpressing Mice

Abstract The developing brain is uniquely susceptible to oxidative stress, and endogenous antioxidant mechanisms are not sufficient to prevent injury from a hypoxic-ischemic challenge. Glutathione peroxidase (GPX1) activity reduces hypoxic-ischemic injury. Therapeutic hypothermia (HT) also reduces hypoxic-ischemic injury, in the rodent and the human brain, but the benefit is limited. Here, we combined GPX1 overexpression with HT in a P9 mouse model of hypoxia-ischemia (HI) to test the effectiveness of both treatments together. Histological analysis showed that wild-type (WT) mice with HT were less injured than WT with normothermia. In the GPX1-tg mice, however, despite a lower median score in the HT-treated mice, there was no significant difference between HT and normothermia. GPX1 protein expression was higher in the cortex of all transgenic groups at 30 min and 24 h, as well as in WT 30 min after HI, with and without HT. GPX1 was higher in the hippocampus of all transgenic groups and WT with HI and normothermia, at 24 h, but not at 30 min. Spectrin 150 was higher in all groups with HI, while spectrin 120 was higher in HI groups only at 24 h. There was reduced ERK1/2 activation in both WT and GPX1-tg HI at 30 min. Thus, with a relatively moderate insult, we see a benefit with cooling in the WT but not the GPX1-tg mouse brain. The fact that we see no benefit with increased GPx1 here in the P9 model (unlike in the P7 model) may indicate that oxidative stress in these older mice is elevated to an extent that increased GPx1 is insufficient for reducing injury. The lack of benefit of overexpressing GPX1 in conjunction with HT after HI indicates that pathways triggered by GPX1 overexpression may interfere with the neuroprotective mechanisms provided by HT.

normothermia.GPX1 protein expression was higher in the cortex of all transgenic groups at 30 min and 24 h, as well as in WT 30 min after HI, with and without HT.GPX1 was higher in the hippocampus of all transgenic groups and WT with HI and normothermia, at 24 h, but not at 30 min.Spectrin 150 was higher in all groups with HI, while spectrin 120 was higher in HI groups only at 24 h.There was reduced ERK1/2 activation in both WT and GPX1-tg HI at 30 min.Thus, with a relatively moderate insult, we see a benefit with cooling in the WT but not the GPX1-tg mouse brain.The fact that we see no benefit with increased GPx1 here in the P9 model (unlike in the P7 model) may indicate that oxidative stress in these older mice is elevated to an extent that increased GPx1 is insufficient for reducing injury.The lack of benefit of overexpressing GPX1 in conjunction with HT after HI indicates that pathways triggered by GPX1 overexpression may interfere with the neuroprotective mechanisms provided by HT.

Introduction
The developing brain is uniquely susceptible to oxidative stress and subsequent injury [1][2][3].Hydrogen peroxide, in particular, accumulates in the neonatal, but not the adult, brain after hypoxia-ischemia (HI) [4].Endogenous antioxidant mechanisms, such as catalase and glutathione peroxidase (GPX1), are not sufficient to prevent injury from a hypoxic-ischemic challenge.GPX1 activity has been shown to peak at day 1 (P1) of life in the mouse brain, decrease at P4, again at P7, and decline to adult levels by P14 [5].Increased GPX1 at P7, however (via GPX1 overexpression in the mouse), reduces HI injury [6,7].Hypoxia preconditioning has been shown to protect the brain from subsequent insults, but we found that the GPX1-tg mice lost the protection afforded by GPX1 overexpression when given hypoxia preconditioning.WT littermates of these mice were protected by hypoxia preconditioning, however [7].We also showed that ERK1/2 is transiently activated by hypoxia preconditioning in WT P7 mice, but this activation is blocked by GPX1 overexpression [8].This indicates that redox balance depends on ERK1/2 activity, which subsequently protects against oxidative injury.
We previously determined GPX1 activity at P7 to be approximately 1.2-fold higher in the cortex of these GPX1-tg mice than in wild-type (WT) mice [6].In addition, 24 h after HI, GPX1 activity rose 1.3-fold over naïve GPX1-tg cortex [6].The P9-10 mouse brain is now considered by many to be more representative of the term-gestation human brain [9].In the P9 GPX1-tg mouse, GPX1 protein expression was several times higher than WT in the cortex and hippocampus [10].Since GPX1 targets hydrogen peroxide, its mechanism of action is presumably in the early, or acute, phase of HI injury, while hypothermia (HT), applied after injury onset, acts on the subsequent phase.
As a therapy, HT acts after injury is underway and has been shown to reduce HI injury, in the rodent [11] and the human [12] brain, but the benefit is limited and many questions remain regarding optimal treatment of newborns with HIE [13,14].The mechanisms of protection are not fully understood but are wide-ranging and include reduced metabolic rate, decreased production of free radicals, suppression of inflammation, and inhibition of excitotoxicity [14].Consequently, additional treatments to HT may prove more effective, but they must be synergistic.Indeed, the challenge of inhibiting cell death without interfering with overall recovery indicates therapies should be timed to target the phases of injury and, ultimately, repair [15].Here, we combined GPX1 overexpression with HT in a P9 mouse model of HI to test the effectiveness of enhanced antioxidant capacity with therapeutic cooling.

Mice
The GPX1-overexpressing mice used here (GPX1-tg) were developed with 1 additional copy of the transgene, resulting in expression in all brain regions analyzed in adult mice, including the cortex and striatum, as previously described [16].The GPX1 activity ratio in these mice was approximately 1.5 for mesencephalon, and protein expression was 2.4-fold in cortex and 1.9-fold in striatum compared to WT [16].GPX1-tg mice were bred and maintained at the UCSF Laboratory Animal Resource Center.Male mice heterozygous for GPX1 were bred with female WT (CD1) mice, and the genotype of resulting litters was determined by PCR, using standard methods as previously described [6,16].While we have used these mice in several previous studies [6,7,17] and have confirmed increased GPX activity and expression in the P7 mouse [6], we felt that the brain of the P9 mouse was sufficiently more mature to warrant confirmation of degree of increased GPX activity.

GPX Activity
To confirm overexpression in P9 GPX1-tg mice, seleniumdependent GPX1 enzymatic activity was measured in naïve cortex and hippocampus as previously described [6,18].Briefly, brains were removed from anesthetized GPX1-tg (n = 10) and WT mice (n = 13); cortices and hippocampi were quickly dissected on a cold surface and immediately frozen in methylbutane cooled by dry ice.Brain samples were stored at −80°C until assay, at which time they were homogenized in 50 mM potassium phosphate buffer with 1 mM EDTA (pH 7.0).GPX activity was measured in a coupled test system in which reduced glutathione and tert-butyl hydroperoxide were used as the substrates, and oxidized glutathione produced by GPX activity was measured by kinetic spectrophotometry (340 nm) of glutathione reductase-mediated NADPH oxidation.Units of GPX activity were determined by a standard curve of GPX expressed as units GPX/mg/min, where 1 unit (U) is defined as 1 nmol NADPH oxidized per minute.Protein was determined by Pierce BCA spectrophotometric assay (Pierce, Rockford, IL, USA).

HI/HT
Neonatal mice overexpressing GPX1 and their WT littermates (CD1 background) underwent HI on postnatal day 9 (P9) [11,19,20].Briefly, under isoflurane anesthesia, the left common carotid artery was dissected and coagulated until severed.After recovery with the dam for 1 h, mice were exposed to hypoxia by exposure to 10% oxygen (balance nitrogen) for 40 min.After another 1 h period with the dam, mice were placed in chambers maintained at 36.5°C (normothermia [NT]) or 32°C (HT) for 3.5 h and gradually rewarmed over 30 min before returning to the dam.The temperature of each mouse was measured at the nape of the neck with a hand-held laser-guided remote thermometer which we have compared to a rectal probe and found this measurement to be 0.5°C less than the rectal temperature.Temperature was recorded every 30 min.Histopathologic Analysis of Injury For determination of degree of injury, 18 weeks after HI and HT or NT (a timepoint when injury is fully resolved), mice were anesthetized with Euthasol (Virbac AH, Fort Worth, TX, USA), perfused with 4% paraformaldehyde, brains were cut on a vibratome (50 μm), and alternate sections stained with cresyl violet and Perl's iron stain.Brain injury in the Vannucci model is variable, from large areas of infarct to scattered areas of pyknosis.The Perl's iron stain reveals these smaller areas of injury and thus provides a complement to the Nissl stain of the cresyl violet.Consequently, all sections are examined, and 11 regions are scored on a scale of 0-3: anterior, middle, and posterior cortex; anterior, middle, and posterior striatum; CA1, CA2, CA3, and dentate gyrus of the hippocampus; and thalamus, with 0 = no injury and 3 = severe cystic infarction, for a cumulative score of 0-33 [21].In addition, injury volume was measured with the cresyl violet-stained sections using ImageJ software (NIH).

Protein Expression by Western Blot
At 30 min or 24 h after sham surgery, HI and NT, or HI and HT, brains were removed from anesthetized mice, and cortices and hippocampi were quickly dissected on a cold surface and immediately frozen in methylbutane cooled by dry ice.Brain samples were stored at −80°C.For 30 min: WT sham NT (n = 6), WT sham HT (n = 4), WT with HI and NT (n = 9), WT with HI and HT (n = 11), GPX1-tg sham NT (n = 6), GPX1-tg sham HT (n = 4), GPX1tg with HI and NT (n = 10), GPX1-tg with HI and HT (n = 10).For 24 h: WT sham NT (n = 10), WT sham HT (n = 14), WT with HI and NT (n = 16), WT with HI and HT (n = 16), GPX1-tg sham NT (n = 10), GPX1-tg sham HT (n = 11), GPX1-tg with HI and NT (n = 18), GPX1-tg with HI and HT (n = 17).It should be noted that the timepoints used for Western blots here, 30 min and 24 h after NT or HT exposure, are 5 and 29 h after hypoxia, respectively.Thirty minutes is the point of recovery from HT and is the early timepoint for analysis.
Frozen brain tissue was homogenized in RIPA buffer (Sigma, St. Louis, MO, USA) with protease and phosphatase inhibitors (Thermo Fisher, Rockford, IL, USA) using dounce homogenizers.The homogenate was transferred to chilled Eppendorf tubes and left on ice for 30 min before centrifugation at 14,000 rpm for 15 min at 4°C.The supernatant was transferred to clean chilled tubes, and an aliquot was removed for determination of total protein by BCA assay (Thermo Fisher).30 μg protein were separated by SDS-PAGE and transferred to PVDF membranes.After blocking for 1 h in 5% non-fat dry milk in TBS with 0.5% TWEEN, membranes were incubated in the following antibodies: goat-actin 1:2,000 (Abcam, Cambridge, MA, USA); rabbit-GPX1, 1:2,000 (Abcam, Cambridge, MA, USA); mouse-spectrin 1:4,000 (Millipore, Temecula, CA, USA); mouse-ERK1/2 1:4,000, Invitrogen; rabbit-phospho-ERK1/2 (Cell Signaling, Danvers, MA, USA); and corresponding secondary antibodies, all 1:2,000.The signal was visualized with enhanced chemiluminescence (Thermo Fisher) and blots exposed to film.Film was scanned, and mean optical densities were determined with ImageJ (NIH).

Statistical Analysis
GPX1 enzymatic assay results were analyzed by unpaired t-test and described as mean U GPX/mg protein ± SEM.Injury scores were evaluated by one-way ANOVA and Kruskal-Wallis with Dunn's multiple comparison test and are shown as scatter plots with the median value of each group a horizontal line.Injury volumes were compared by t-test and are described as percent injured hemisphere compared to contralateral hemisphere ± S.D. Western blots were normalized to β-actin, evaluated with unpaired t-test, and presented as fold change relative to WT sham NT.Differences were considered significant at p < 0.05.Mortality was analyzed by contingency test and χ 2 .Graphpad Prism 7.0 (Carlsbad, CA) was used for all analysis except injury volumes which were by Excel.

Brain Injury after HI
WT mice with HT were less injured than WT with NT (median scores 6 [n = 13] and 29 [n = 15], respectively; p < 0.03; Fig. 2a).In the GPX1-tg, however, despite a lower median score in the HT-treated mice, there was no significant difference between HT and NT (median scores 19 [n = 12] and 31 [n = 10], respectively; Fig. 2a).The pattern is similar in the brain regions analyzed separately: In the cortex, WT mice with HT were less injured than WT with NT (median scores 3 and 9, respectively; p < 0.03; Fig. 2b).In the GPX1-tg, there was no significant difference between HT and NT (median scores 5.5 and 9, respectively; Fig. 2b).In the hippocampus, WT mice with HT were less injured than WT with NT (median scores 3 and 10, respectively; p = 0.05; Fig. 2c).In the GPX1-tg, there was no significant difference between HT and NT (median scores 7 and 11, respectively; Fig. 2c).In the striatum, WT mice with HT were less injured than WT  with NT (median scores 1 and 8, respectively; p < 0.02; Fig. 2d).In the GPX1-tg, there was no significant difference between HT and NT (median scores 5 and 8.5, respectively; Fig. 2d).In the thalamus, WT mice with HT were less injured than WT with NT (median scores 0 and 2, respectively; p < 0.006; Fig. 2e).In the GPX1-tg, the median scores were the same (2) for HT and NT (Fig. 2e).There were no differences in injury scores when comparing male and female of each treatment group (Fig. 2f).However, there were a small number of female mice in the WT with HT (n = 3) and male mice in the GPX1-tg with HT (n = 4) groups.
Representative photomicrographs of a brain with median injury for each experimental group, cortex and hippocampus stained with cresyl violet and Perl's iron stain, respectively, for WT with HI and NT (Fig. 3a), WT with HI and HT (Fig. 3b), GPX1-tg with HI and NT (Fig. 3c), and GPX1-tg with HI and HT (Fig. 3d).

Mortality
There was no difference in mortality between the groups.Four mice died in WT NT, five in WT HT, four in GPX1-tg NT, and five in GPX1-tg HT.One was euthanized due to hydrocephalus (WT HT), three died before weaning (1 GPX1-tg NT, 2 GPX1-tg HT), and the remainder prior to scheduled perfusion.

Spectrin Protein Expression
Spectrin expression was measured to determine the effect of HT and GPX1 overexpression on cell death mechanisms, as seen by the 145/150 kD fragments indicating the calpain-specific action of necrosis or the 120 kD fragments indicating the caspase-specific action of apoptosis, at an early (30 min) and late (24 h) stage of injury.
Since spectrin 120 is a marker for apoptosis, which occurs in the later stages of HI injury, it is not surprising that there was no change in the cortex at 30 min (Fig. 6a).In the cortex at 24 h, however, all HI groups showed about a 20-fold increase over WT sham: WT with HI and NT was 27.1-fold (p < 0.02), WT with HI and HT was 24.6-fold (p = 0.05), GPX1-tg with HI and NT was 16.1-fold (p < 0.02), and GPX1-tg with HI and HT was 25.1-fold (p < 0.04) (Fig. 6b).Similar to the cortex, there was no change in the hippocampus at 30 min (Fig. 6c), but there was at 24 h: WT with HI and NT increased by 21.6-fold (p < 0.02), WT with HI and HT by 13.0-fold (p = 0.05), and GPX1-tg with HI and NT by 14.1-fold (p < 0.03) (Fig. 6d).In GPX1-tg with HI and HT, although the mean was more than 10 times higher than WT sham, it did not reach significance (11.5-fold, p = 0.07) (Fig. 6d).

ERK1/2 Protein Expression
ERK1/2 is of interest primarily in comparison to its activated state.Thus, we do not expect to see significant differences between the groups.Indeed, there were no changes in ERK1/2 expression in the cortex at 30 min (Fig. 7a), the cortex at 24 h (Fig. 7b), the hippocampus at 30 min (Fig. 7c), or the hippocampus at 24 h (Fig. 7d).

Phosphorylated ERK1/2 Protein Expression
In the cortex at 30 min, phosphorylation of ERK1/2 was lower in WT with HI and HT by −0.57-fold (p < 0.04), and while there was a trend toward lower phosphorylation of ERK1/2 in GPX1-tg with HI and HT, it did not reach significance (p = 0.054) (Fig. 8a).There were no differences in the levels of phosphorylation of ERK1/2 in the cortex at 24 h (Fig. 8b), the hippocampus at 30 min (Fig. 8c), or the hippocampus at 24 h (Fig. 8d).

Discussion
This study expands upon our previous work on oxidative mechanisms in neonatal HI via GPX1 overexpression and with HT.The combination of two potentially protective interventions, GPX1 overexpression and HT, is a novel way of exploring the impact of HT on oxidative injury in the neonatal brain.This is the first study to show that there is no synergism of HT with GPX1 overexpression.
With a relatively moderate insult (40 min of 10% oxygen), we see a benefit with cooling in the WT but not in the GPX1-tg when the insult occurs at P9.The P9 mouse brain is now considered to more closely mimic the human brain at term birth, and the fact that we do not see a significant benefit with increased GPX1 here in the P9 model may be specific to this stage of development.It may indicate that oxidative stress in these slightly older mice is elevated to an extent that the degree of increase in the amount of GPX1 due to the transgene is insufficient for reducing injury, presumably due to an excess of hydrogen peroxide.It may also be a consequence of a surprisingly severe degree of injury overall, as demonstrated by the WT with NT group.We have shown the benefits of cooling in CD1 WT mice previously with a more severe insult (50 min 10% oxygen), in which a large number of NT-treated mice had severe injury [22].Consequently, we reduced the duration of hypoxia to 40 min, aiming for a more moderate insult compared to 50 min of hypoxia.
We have previously shown that GPX1 activity is higher in the naïve P7 GPX1-tg mouse cortex and hippocampus compared to WT and have now confirmed higher activity in the P9 GPX1-tg cortex and hippocampus.In the P7 model of HI, we previously showed decreased injury in GPX1-tg mice compared to WT littermates as a consequence of reduced hydrogen peroxide accumulation.The comparison to the P7 mouse may also suggest that increased antioxidant availability is more likely to reduce injury in the pre-term brain than the term brain.While there are eight known forms of GPX [23], GPX1 (along with catalase) is considered to be the primary detoxifier of hydrogen peroxide in the brain [24].Also, hydrogen peroxide accumulates in neonatal, but not adult, mouse brain after HI [4].In recent years, however, there has been much interest in the phospholipid hydroperoxide GPX4.Also found in the brain, GPX4 reduces hydrogen peroxide at a much slower rate than GPX1 but, importantly, reduces hydroperoxides in cell membranes [25], thereby inhibiting ferroptosis [26].The roles of GPX1 and GPX4 in HI, whether distinct or complementary, have yet to be fully determined.Some previous studies using the Vannucci model of HI have found differences in injury severity associated with sex of the mice, with males being more injured than females [27,28].However, one study using P10 C57Bl/6 mice and 4 h of HT found no reduction in injury or seizure susceptibility overall with HT, but injury did correlate with seizure susceptibility in male mice only [29].The fact that we did not find sex differences here could be attributable to low numbers in some groups, the relatively high degree of injury severity overall, strain of mouse used, or other factors yet to be understood.Regardless, sex differences in HI merit continued attention.
Whereas HT likely acts on a number of physiological mechanisms, delaying or diminishing secondary energy failure, edema, and inflammation during the hours after the initial injury, GPX1 is more specific in its action.It reduces the hydrogen peroxide that is produced by superoxide dismutase during the acute phase of HI injury as a consequence of excitotoxicity and oxidative stress, yet in GPX1-tg mice, the overexpression is ongoing.It is conceivable that HT applied during the later stages of the acute phase (beginning 1 h after mice resume breathing room air) negates the potential beneficial effects of ongoing overexpression of GPX1 in the brain.The increase in GPX1 protein seen in the WT cortex 30 min after HI and either NT or HT treatment (which is absent at 24 h) supports the idea that in this late acute phase, the brain is increasing its endogenous antioxidant defenses.HT may be synergistic to this increased GPX1 in the WT cortex but not the GPX1-tg.Regional differences in the timing of GPX1 expression, and perhaps antioxidant defense mechanisms overall, are suggested by the lack of increased GPX1 in the hippocampus in any group at 30 min, yet increased GPX1 is seen at 24 h.Spectrin expression could provide a glimpse of the effects of HT and GPX1 overexpression on cell death mechanisms, whether necrotic or apoptotic.It is not surprising that spectrin 145/150, as a marker of necrosis, was present in all HI groups at 30 min and persists for 24 h, but it is notable that it was particularly high in the hippocampus.It is also not surprising that there is a lack of significant spectrin 120, a marker of apoptosis, at 30 min.The high levels of spectrin 120 at 24 h (approximately 10-25-fold) suggest ongoing apoptotic cell death which is not diminished by HT treatment or GPX overexpression.
We chose 30 min after HT or NT exposure as a timepoint for Western blot experiments in part due to our previous finding in the P7 mouse that hypoxia alone activates ERK1/2 in the WT cortex at 30 min, and GPX1 overexpression prevents this activation.In this study, HI transiently reduced ERK activation at 30 min in both the cortex and hippocampus in both WT and GPX1-tg, with and without HT.Since the ERK pathway is involved in many cellular processes, both pro-survival and pro-death, it is difficult to know what the consequence of this reduced activation is; however, this decrease mirrors the increase in spectrin 145/150 in the cortex at 30 min, suggesting that reduced ERK activation plays a role in necrotic cell death in HI.
In summary, this work contributes to our understanding of antioxidant mechanisms in neonatal HI, as well as the limitations and promise of HT in combination with genetic or pharmacological interventions.

Statement of Ethics
All animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee at UCSF under protocol AN187224 (D.M.F.) and carried out with standards of care in accordance with the NIH Guide for the Care and Use of Laboratory Animals.

Fig. 3 .
Fig. 3. Histological injury of WT and GPX1-tg cortex and hippocampus after HI and either NT or HT.Brains shown represent median injury score of each group.Alternate sections stained with cresyl violet (left) or Perl's iron (right).a WT NT. b WT HT. c GPX-tg NT. d GPX-tg HT.Arrows depict focal areas of cell loss.Scale bar = 500 μm.

Fig. 7 .
Fig.7.ERK1/2 protein expression in WT and GPX1-tg after HI and either HT.Shown as fold change compared to WT sham NT. a Cortex 30 min.There were no differences in ERK1/2.b Cortex 24 h.There were no differences in ERK1/2.c Hippocampus 30 min.There were no differences in ERK1/2.d Hippocampus 24 h.There were no differences in ERK1/2.

Fig. 8 .
Fig. 8. Phospho-ERK1/2 (p-ERK) protein expression in WT and GPX1-tg after HI and either NT or HT.Shown as fold change compared to WT sham NT. a Cortex 30 min p-ERK is lower in WT HI HT (*p < 0.04).There was a trend to lower p-ERK in GPX1-tg HI HT which was not significant (p = 0.054).b Cortex 24 h.There were no differences in p-ERK.c Hippocampus 30 min.There were no differences in p-ERK.d Hippocampus 24 h.There were no differences in p-ERK.