Hypothermia treatment in Hypoxia-Inducible Factor-1α Knockout Mice with Hypoxia-Ischemia Open Access

Hypoxia-inducible factor-1α (HIF-1α) is a key regulator of the cellular responses to hypoxia via induction of a wide range of genes [1], many of which are protective to the neonatal brain under hypoxic-ischemic (HI) conditions [2]. However, the complexity of how HIF affects various metabolic pathways cannot be underestimated and alterations to HIF may have unintended consequences [3]. We previously found that mice with a neuron-specific HIF-1α deficiency that underwent HI at postnatal day 7 (P7) had increased brain injury [4]. These mice also demonstrated resistance to hypoxia preconditioning protection, further indicating a protective role for HIF-1α to the neonatal brain under oxidative stress [5]. The P7 mouse modification of the Vannucci rat procedure of HI, first employed by us [6], has been modified for the P9 mouse [7, 8] as the P9–10 rodent brain is now widely considered to be a more appropriate model of the term newborn human brain [9, 10]. We recently showed that HIF-1α reduction in P9 mice increases injury after HI, confirming our results with the P7 mouse [11].

After more than 2 decades of research using animal models of hypothermia (HT) as a treatment for HI injury, as well as successful clinical trials, therapeutic HT has become established as standard treatment for HI injury in human newborns [12]. However, the benefit in terms of reduced brain damage is incomplete for many infants, who may experience lifelong mental and physical deficits [13]. The mechanisms of protection by HT are not fully understood but likely involve reduced metabolic activity, decreased inflammation and acidosis and subsequent activation of protective genes. Thus, there remains a need to explore the mechanisms of HI injury with HT for potential protective avenues, as well as mechanisms of repair. Activation of HIF-1α and its downstream genes may be involved but has not been investigated. Thus, we utilized the same strain of HIF-KO mouse as we used previously, subjecting them to HI at P9 followed by HT. Degree of injury was determined histologically and was also demonstrated by protein expression for the cell death markers spectrin 145/150 and spectrin 120. To explore the pathways that may be associated with HIF-1α activation, we determined protein expression of extracellular signal-related kinase 1/2 (ERK1/2), which is necessary for brain development [14]. It also phosphorylates HIF-1α [15] and is activated after HI in the neonatal rat [16, 17]. Also, HT increased ERK phosphorylation, but whether ERK signaling accounts for the protection of HT is controversial [18]. The brain’s response to HT likely involves upregulation of “cold stress” proteins such as the RNA-binding motif 3 (RBM3) and cold-inducible protein-1 (CIRP1). Here, we measured protein expression of RBM3 to determine if it was associated with protection or injury. This work furthers our understanding of the role of HIF-1α in neonatal HI injury and the effect of HT on the immature brain.

Mice with conditional neuron-specific inactivation of HIF-1α were generated using CRE/Lox technology as previously described [19]. Briefly, mice heterozygous for CAMCRE (expressing Cre recombinase under the control of the CaMKIIα promoter) were bred with homozygous “floxed” HIF-1α transgenic mice. The resulting litters produced mice with a forebrain predominant deletion of HIF-1α (HIF-KO), as well as littermates without the deletion. All mice negative for the CAMCRE gene were considered “wild-type (WT).” Genotyping was carried out by PCR on tail DNA samples.

The Vannucci procedure of HI [20] was carried out on mice at P9 as previously described [7, 8] (unilateral carotid artery ligation and 50 min of 10% oxygen). Following a 1-h recovery period with the dam, mice were either cooled to 32°C or maintained at 36.5°C for 3.5 h. Cooled mice were gradually rewarmed over 30 min. Sham-operated control mice received anesthesia and exposure of the carotid artery without ligation and hypoxia. For histological determination of injury severity, mice were perfused with 4% paraformaldehyde 5–7 days later. For Western blots, mice were anesthetized and decapitated 4 h or 24 h after HI with either HT or normothermia (NT). The brain was removed, and ipsilateral cortices and hippocampi were dissected and flash frozen in methyl butane on dry ice.

Brains were cut with a Vibratome and 50 μm sections collected for Cresyl Violet and Perl’s iron stain. Brain injury was determined by microscopic examination of all sections by an observer blinded to the mouse groups as previously described [8]. Briefly, eleven regions were each given a score from 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, 1 = mild, focal injury, 2 = moderate and 3 = severe cystic infarction, for a cumulative score of 0–33. In addition, the ipsilateral and contralateral hemispheres from each brain from six cresyl violet-stained sections were measured for injury volume using ImageJ software (NIH).

Frozen cortices and hippocampi were homogenized with dounce homogenizers and separated into cytoplasmic and nuclear fractions using NE-PER kits, with the addition of protease and phosphatase inhibitors, according to the manufacturer’s protocol (Thermo Scientific, Rockford, IL, USA). Protein concentration was determined by BCA assay (Thermo Scientific). Cytoplasmic (30 μg) or nuclear (10 μg) proteins were separated by SDS-PAGE and transferred to PVDF membranes. After blocking in 5% nonfat dry milk in TBS with 0.05% Tween for 1 h, cytoplasmic proteins were incubated with the following antibodies: mouse β-actin-HRP (1:4,000, sc-47778 HRP, Santa Cruz Biotechnology, Dallas, TX, USA), mouse spectrin (1:4,000, MAB-1622, Millipore, Burlington, MA, USA), mouse ERK1/2 (1:3,000, 13-6200, Fisher, Waltham, MA, USA), rabbit phosphorylated-ERK1/2 (p-ERK) (1:1,000, 9101S, Cell Signaling Technology, Danvers, MA, USA), and rabbit RBM3 (1:500, HPA003624, Sigma, St. Louis, MO, USA). Nuclear proteins were probed for actin (as above) and HIF-1α (1:1,000, NB100-479, Novus Biologicals, Centennial, CO, USA). Corresponding secondary antibodies were all 1:2,000. The signal was visualized with ECL (Thermo Fisher) and blots exposed to film which were scanned. Optical densities were determined with ImageJ (NIH).

Injury scores were evaluated by Kruskal-Wallis with Dunn’s multiple comparisons test. They are shown as scatter plots with the median value of each group a horizontal line. Injury volumes were compared by Mann-Whitney and are shown as percent of remaining injured hemisphere compared to contralateral hemisphere +/− SD. The OD values of Western blots bands were normalized to β-actin, evaluated with Mann-Whitney and presented as fold change of WT sham, with the exception of spectrin 120, for which all sham values were 0 and thus are shown as normalized to β-actin. Sex differences were evaluated by Kruskal-Wallis with Dunn’s multiple comparisons test. Differences were considered significant at p < 0.05.

WT mice had reduced histologic brain injury with HT compared to WT mice with NT (median score = 6 [range 2–33, n = 12] vs. 27 [range 9–33, n = 9], respectively, p < 0.04; Fig. 1a). In the HIF-KO mice, however, injury was not reduced with HT treatment compared to NT (median score = 10 [range 1–33, n = 7] vs. 24 [range 4–33, n = 13], respectively, p < 0.50). Injury was not different between WT with NT compared to HIF-KO with NT nor WT with HT compared to HIF-KO with HT.

Evaluated by brain regions separately, injury in the cortex was not different between WT with HT and WT with NT, despite the median score of WT with HT = 3 (range 1–9) vs. WT with NT = 8 (range 2–9) (Fig. 1b). There was also no difference between HIF-KO with HT and HIF-KO with NT (median score = 4 [range 1–9] vs. 6 [range 0–9]). There was also no difference in injury between WT with NT compared to HIF-KO with NT nor WT with HT compared to HIF-KO with HT.

In the hippocampus, however, WT mice had reduced histologic brain injury with HT compared to WT mice with NT (median score = 3 [range 0–12] vs. 10 [range 4–12], respectively, p = 0.02; Fig. 1c). In the HIF-KO mice, however, injury was not reduced with HT treatment compared to NT (median score = 0 [range 0–12] vs. 7 [range 2–12], respectively). There was also no difference in injury between WT with NT compared to HIF-KO with NT nor WT with HT compared to HIF-KO with HT.

Volume measurements of the injured hemisphere correspond with the injury scores. WT mice with HT had reduced injury compared to WT mice with NT (mean remaining hemisphere = 84.1 ± 24.5% vs. 55.8 ± 35.3%, respectively, p < 0.002; Fig. 2). There was no reduction in injury in HIF-KO with HT compared to HIF-KO with NT (mean remaining hemisphere = 78.2 ± 31.1% vs. 67.0 ± 28.3%, respectively).

There was no postoperative mortality. There were no differences based on sex (WT with NT female n = 4, male n = 5; WT with HT female n = 7, male = 6; HIF-KO with NT female = 8, male = 5; HIF-KO with HT female = 6, male = 1).

In the cortex 4 h after the end of HI with NT or HT exposure, HIF-1α was lower in both HIF-KO with NT and HIF-KO with HT compared to WT sham (p < 0.01 and p < 0.05, respectively; Fig. 3a). HIF-1α was also lower in HIF-KO with NT compared to WT with NT or with HT (p < 0.04 and p < 0.01, respectively). In the cortex 24 h after the end of NT or HT exposure, HIF-1α increased in WT with NT compared to WT sham (p < 0.03) and compared to HIF-KO with NT (p < 0.03). HIF-KO with NT was also lower than WT with HT (p < 0.05). In the hippocampus 4 h after the end of HI with NT or HT exposure, there were no differences in HIF-1α expression, but there was a trend for reduced HIF-1α in HIF-KO with NT and HIF-KO with HT compared to WT sham (p = 0.06 and p = 0.11, respectively; Fig. 3b). In the hippocampus at 24 h, HIF-1α was lower in HIF-KO with NT compared to sham (p = 0.02), but again there was only a trend toward reduced HIF-1α in HIF-KO with HT (p = 0.07).

In the cortex 4 h after the end of HI with NT or HT exposure, spectrin 145/150 (a marker of necrotic cell death) increased in WT with NT compared to sham (p = 0.001; Fig. 4a) and compared to WT with HT (p < 0.02). Spectrin 145/150 also increased compared to sham in HIF-KO with NT or with HT (p < 0.002 and p < 0.03, respectively), but there was no difference between these two conditions. Also, HIF-KO with HT was higher than WT with HT at 4 h (p < 0.05). In the cortex at 24 h, spectrin 145/150 increased compared to sham in WT with NT or with HT (p < 0.002 and p < 0.01, respectively) and in HIF-KO with NT or with HT (p = 0.001 and p < 0.05, respectively). In the hippocampus, spectrin 145/150 increased 10-fold or more in all groups compared to sham (all p < 0.05, Fig. 4b). There were no differences between groups.

In the cortex 4 h or 24 h after the end of HI with NT or HT exposure, spectrin 120 (a marker of apoptotic cell death) was not different between groups (Fig. 4c). In the hippocampus at 4 h, spectrin 120 was also unchanged (Fig. 4d). In the hippocampus at 24 h, spectrin 120 increased in all groups compared to WT with NT at 4 h (all p < 0.05, Fig. 4d).

ERK activation following HI was determined as the ratio of p-ERK/ERK. In the cortex 4 h after the end of HI with NT or HT exposure, p-ERK/ERK was not different compared to sham or between groups hanging (Fig. 5a). In the cortex at 24 h, however, p-ERK/ERK was higher compared to sham in WT with NT or with HT (both p < 0.03) and in HIF-KO with NT or HT (p < 0.002 and p < 0.01, respectively). Although not statistically different, ERK activity in the HIF-KO mice tended to be higher than the WT mice, with NT and HT. In addition, p-ERK/ERK levels at 24 h were higher overall than those at 4 h, in both WT and HIF-KO mice, with either NT or HT (WT with NT, 24 h vs. 4 h, p < 0.01; WT with HT, 24 h vs. 4 h, p < 0.002; HIF-KO with NT, 24 h vs. 4 h, p < 0.02; HIF-KO with HT, 24 h vs. 4 h, p = 0.01). In the hippocampus at 4 h or 24 h, there were no differences compared to sham or between groups (Fig. 5b).

In the cortex 4 h after the end of HI with NT or HT exposure, RBM3 increased compared to sham in the WT with HT (p < 0.05; Fig. 6a) and in HIF-KO with HT (p < 0.04). In addition, RBM3 increased in WT with HT compared to WT with NT (p < 0.02). The difference between HIF-KO with HT and HIF-KO with NT is only a trend (p = 0.11). In the cortex 24 h after the end of HI with NT or HT exposure, RBM3 increased compared to sham in both WT with HT and HIF-KO with HT (p < 0.05 for both). In the hippocampus at 4 h or 24 h, there were no differences compared to sham or between groups (Fig. 6b).

The novelty in this study is the combination of an established therapy (HT) with diminished neuronal HIF-1α. The results support a critical role for HIF-1α in the mechanisms of protection against neonatal HI injury in hypothermic conditions since the HIF-KO mice were not protected with HT, unlike their WT littermates. We confirmed our previous results showing that injury after HI in mice with diminished neuronal HIF-1α is comparable or worse than WT [4, 5, 11] and also confirmed an increase in HIF-1α in WT mouse cortices 24 h after HI [4]. Regional variations in injury can have a profound effect on clinical outcome and are different here than in our earlier study, which may reflect differences between the P7 and P9 model. Previously, the cortex displayed greater injury in the HIF-KO compared to WT, while in the hippocampus, injury was equally severe [4]. Here, WT and HIF-KO with NT do not show a difference in injury in the cortex or hippocampus when analyzed separately. However, while HT reduced injury in the WT overall, regionally, there was a greater reduction in the hippocampus than the cortex. The HIF-KO brains were not protected with HT overall nor in the cortex or hippocampus alone despite the hippocampus with HT having a medium injury score of 0. The small numbers of HIF-KO with HT and the wide variation seen with this model are limitations and may influence this result.

Assessing the expression of spectrin breakdown products at 145/150 KD and at 120 KD is another way of analyzing acute cell death. Only the WT cortex with HT at 4 h has spectrin 145/150 protein levels similar to sham. This corresponds with the lower injury scores and better hemispheric volume preservation in the WT hemisphere with HT. However, it does not correspond with the injury scores in the cortex alone, which, while comparatively low with HT, were not different than WT with NT. The high levels of spectrin 145/150 in the hippocampus of the WT with HT at 4 h also do not correspond with the low injury scores in the hippocampus. Apoptotic cell death often follows necrotic cell death, measurable by spectrin 120. This is seen in the hippocampus in all groups at 24 h but not in the cortex. Since the in vivo brains were acquired 5–7 days after HI, the 4 h and 24 h timepoints of the Western blots render them an imperfect reflection of cell death mechanisms in the overall recovery period. They are rather one marker of a very complex process. These disparities are also likely to reflect the variation in injury in the HI model as well as variations in the brain’s response to therapeutic HT, both regionally and over time.

ERK signaling pathway is activated upon hypoxia or ischemia, including neonatal brain HI [17, 21, 22], and it has been suggested that ERK1 is necessary for the activation of HIF-1α by hypoxia [15]. It has been implicated in cell proliferation and differentiation, as well as cell survival or death depending on time, intensity and duration of activation, ERK isoform, as well as cell type [23]. P-ERK/ERK is increased in the cortex of all treatment groups at 24 h compared to sham values and compared to 4 h. While it has been shown that HT upregulates ERK [24], we did not see a difference compared to NT in either WT or HIF-KO and ERK signaling does not seem to be required for the protection of HT [25]. At 24 h, p-ERK/ERK is particularly strong in the HIF-KO, lending support to the idea that lack of HIF-1α or excessive ERK activation is deleterious. However, pharmacological manipulation of the ERK pathway has recently been suggested as a route to neuroprotection [26]. In sum, a controlled level of ERK1/2 signaling is essential for the outcome of brain ischemia [27].

It has been suggested that expression of RBM3 is both required for cell proliferation [28] and does not require HIF-1 [29]. This agrees with our finding that RBM3 is expressed in the cortex of HIF-KO with HT at 4 h. Expression of RBM3 also appears to be a useful early marker for HT after HI in both WT and HIF-KO cortex, yet the role of RBM3 remains to be fully understood.

It is intriguing that we see protection to the WT brain with HT but not the HIF-KO brain with HT given that we saw a similar disparity in the setting of hypoxia preconditioning. In the previous study, WT mice had reduced injury (and increased HIF-1α) when given a hypoxic insult prior to HI, but HIF-KO saw no protection [5]. Thus, the protective mechanisms of HIF-1α and its downstream regulators, including ERK, VEGF, and EPO, may be compromised when levels of HIF-1α are altered. The fact that we did not find differences due to sex may be due to the relatively small numbers of mice in some groups when divided into male and female.

HIF-1α is not the only isoform of HIF, nor is HIF induction limited to neuronal cells. While neurons, astrocytes, and microglia all respond to hypoxia via induction of the HIF-1α signaling pathway, the response is expressed differently in each cell type [30‒32]. For example, mice with a conditional neuronal deletion of HIF-2α also had worse injury after HI than WT littermates as demonstrated by volume measurements and expression of spectrin breakdown products and cleaved-caspase-3 [33].

Continued experimental work can inform the ongoing need to refine clinical practice regarding treatment of newborns with brain injury [13]. For example, therapeutic HT is the standard of care for term newborns with moderate to severe HIE, but there is ongoing debate regarding the safety and efficacy of cooling newborns with mild HIE and the optimal duration and depth of cooling [12]. Also, the need for an effective treatment for preterm infants with hypoxic or HI injury is great and is complicated by the unique and rapidly changing nature of the developing brain in both humans and rodents. Indeed, we tried many different paradigms of HT with HI in the P7 mouse and did not see protection with a variety of temperatures and durations (unpublished observations). We now know the P7 rodent brain is more like a preterm human brain rather than term (as mentioned above) and this indicates that injury to the preterm brain may, unfortunately, not be reduced by HT. A study comparing the P5 with P10 mouse undergoing the same HI insult found that inflammation began early but subsided in the younger mice while in the older mice onset of inflammation began later but continued longer [34]. Thus, attention to the developmental age of the brain is of critical importance when considering interventions [35, 36]. There is a continuing need to study the complex role of HIF-1α in brain development, injury and repair, as well as how HIF-1α is affected by treatment such as HT.

All animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee at UCSF under protocol AN202238 (D.M.F.) and carried out with standards of care in accordance with the NIH Guide for the Care and Use of Laboratory Animals.

The authors have no conflicts of interest to disclose.

This study and all authors were supported by the grant NIH R35 NS097299 and UCSF Resource Allocation Proposal grant (to D.M.F.).

R. Ann Sheldon wrote the manuscript and acquired, analyzed, and interpreted data. Fuxin Lu and Nicholas R. Stewart performed animal experiments and procedures. Xiangning Jiang provided oversight and assistance with experiments and helpful comments on the manuscript. Donna M. Ferriero conceived and provided oversight of all aspects of the study and assisted with the manuscript.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.