Introduction

Encephalopathy consequent on perinatal hypoxia–ischemia occurs in 1–3 per 1,000 term births in the UK and frequently leads to serious and tragic consequences that devastate lives and families, with huge financial burdens for society [1]. Although the recent introduction of cooling represents a significant advance, only around 40 % survive with normal neurodevelopmental function [2]. There is thus a significant unmet need for novel, safe, and effective therapies to optimize brain protection following brain injury around birth.

In this review, we discuss both preclinical and clinical research studies that have led to a better understanding of the pathophysiology and metabolic pathways following an acute hypoxic–ischemic perinatal event. Magnetic resonance spectroscopy (MRS) has been an important tool to understand the evolution of energy failure and brain intracellular pH (pHi) after hypoxia–ischemia [3]. Phosphorus-31 (31P) MRS has demonstrated an alkaline shift in brain pHi [4]; those infants with worse cerebral injury on magnetic resonance imaging (MRI) have a more alkaline brain pHi on 31P MRS [5]. It is likely that the rapid shift in brain pHi to alkaline values following hypoxia–ischemia plays an important role in cell death.

The Na+/H+ exchangers (NHE) are a family of ion membrane transport proteins involved in maintaining a normal pHi and cell volume in many mammalian cell types, by extruding protons in exchange for sodium influx into cells in an electroneutral manner. Excessive activation of NHE is likely to lead to the alkaline shifts in pHi and increased cell death via Na+ overload, which promote intracellular Ca2+ entry. Here, we also review the role of NHE in hypoxia–ischemia, the evidence of neuroprotection with NHE blockade in animal models, and the important association of alkaline brain pHi and seizures.

Acute Hypoxic–Ischemic Perinatal Brain Injury in the Term Infant

Perinatal hypoxic–ischemic brain injury in the term baby is a significant problem throughout the world. Hypoxia–ischemia may occur acutely or chronically and is most commonly associated with maternal factors (hypotension, severe hypoxia), cord factors (prolapse, occlusion), placental factors (insufficiency and abruption), and uterine factors (rupture). Neonatal postnatal events such as shock, respiratory, or cardiac arrest can also lead to hypoxic–ischemic injury. Neonatal encephalopathy (NE) is the clinical manifestation of the ensuing disordered brain function and occurs in 1–3 per 1,000 live term births in the UK and other developed world countries [1]. The encephalopathy typically presents as respiratory difficulties, depression of tone and reflexes, subnormal level of consciousness, and often seizures. Serious consequences follow moderate to severe NE; these include death in 10–15 %, cerebral palsy in 15 %, and other significant cognitive, developmental, and behavioral problems in 40 % of survivors [6]. The financial and human costs to the family and society are thus very high.

Following two decades of laboratory studies [7, 8], clinical trials [2], and recent endorsement from regulatory bodies [9], therapeutic hypothermia is now the standard clinical care for moderate to severe NE in the UK and developed world. Meta-analysis [2, 10] of three large pragmatic trials [2, 11–13] show that therapeutic hypothermia reduces death or disability at 18 months with a risk ratio of 0.81 (95 % CI 0.71–0.93) and a number needed to treat of 6–7. Importantly, data are emerging, which confirm that the favorable outcome of cooled babies at age 18 months is associated with favorable outcome at age 7–8 years [14].

However, hypothermia is only partially effective after neonatal hypoxia–ischemia, and 40 % of all treated infants still suffer severe neurodevelopmental disability or death despite treatment. Additional neuroprotective agents are needed either to augment hypothermic neuroprotection [15] or to use alone in settings where hypothermia has not been established as standard care.

Pattern of Brain Injury on Conventional MRI

Understanding the fundamental mechanisms underlying brain injury in the newborn brain at term has rapidly advanced in recent years. Using MRI, the patterns of hypoxic–ischemic brain injury that have been described in pathology studies in the past can now be demonstrated in superb detail and resolution in vivo in human neonates [16, 17]. Using MRS, the timing of the evolution of energy failure and changes in brain pHi have been fundamental in suggesting new avenues for neuroprotection.

The two main patterns of injury on MRI have been clearly described in the 1950s–1970s in a primate model: the injury pattern associated with acute total asphyxia [18] and with chronic partial asphyxia [19, 20]. These two main patterns are described below and in Fig. 1. A normal neonatal MR image is shown in Fig. 1 (top row). The corresponding 1H MR spectra acquired from the left thalamus are shown for each set of MR images. The 1H MRS values from the white matter are also shown for the watershed predominant pattern of brain injury.

Fig. 1
figure 1

Patterns of brain injury on MRI. Top row (3 Tesla): a axial T1-weighted (w) MPRAGE and b T-2 w 2D 3 mm 3 T MRI of a term infant with normal intracranial appearances on day 5 after birth. This infant presented with stage 1 (mild) neonatal encephalopathy. A normal signal intensity is seen from the posterior limb of the internal capsule on a and b. A thalamic proton (1H) magnetic resonance (MR) spectrum TE 288 ms from the same infant is shown in c. The lactate peak at 1.3 ppm is just visible. Middle row (3 Tesla): a axial T1-w MPRAGE and b axial T2-w 2D 3 mm of an infant with a predominant BGT pattern of brain injury (involvement of basal ganglia, thalami, and perirolandic cortices). A thalamic 1H MR spectrum TE 288 ms from the same infant is shown in c. A high thalamic Lac/NAA ratio of 5.57 is observed. Bottom row (1.5 Tesla): a axial T1-w; b axial T2-w 2D 3 mm of an infant with neonatal encephalopathy with a predominant watershed injury pattern. Areas of cerebral cortical infarction affecting the insular cortices, frontal, and temporal occipital lobes in anterior and posterior arterial watershed territory is seen. A thalamic 1H MR spectrum TE 288 ms from the same infant is shown in c. The Lac/NAA is normal in the thalamus. The white matter spectrum (not shown) has a raised Lac/Cr peak area ratio

Full size image

Predominant basal ganglia–thalamus pattern (BGT) is the most frequent pattern in “acute asphyxia,” affecting bilaterally the central grey nuclei (ventrolateral thalami and posterior putamina) and perirolandic cortex (Fig. 1 middle row). The hippocampus and brainstem can also be involved. This pattern of injury is most often seen following an acute sentinel event, e.g., uterine rupture, placental abruption, or prolapsed cord [21]. The basal ganglia and thalami are susceptible to acute perinatal hypoxic–ischemic injury due to their high metabolic rate and increased concentration of NMDA receptors [22]. Survivors face a range of functional impairments, which include cerebral palsy, feeding problems, speech and language problems, visual and hearing impairment, later seizures, behavioral difficulties, and cognitive impairment. Cerebral palsy affects three quarters of infants with NE with BGT lesions, and the severity of the BGT lesions is the best predictor of motor problems. A normal signal intensity of the posterior limb of the internal capsule within the neonatal period following neonatal encephalopathy is a good predictor of the ability to walk at 2 years [23].

Watershed predominant pattern of injury typically follows “prolonged partial asphyxia.” The areas involved are the vascular watershed zones (anterior–middle cerebral artery and posterior–middle cerebral artery), affecting white matter and in more severely affected infants the overlying cortex (Fig. 1, bottom row). The lesions can be uni- or bilateral, posterior and/or anterior. Conventional MRI can detect the loss of the cortical ribbon [24, 25]. These lesions may become cystic, atrophy, or develop gliosis [26]. Common associated etiological factors are hypotension, infection, and hypoglycemia [27]. Infants with a watershed pattern of injury have predominantly cognitive impairments often without functional motor deficits. Cognitive deficits include memory impairments, visual–motor or visual–perceptive dysfunction, or increased hyperactivity, sometimes in the absence of functional motor problems [28–32].

Cerebral Energy Metabolism Following Perinatal Hypoxia–Ischemia

MRS has been used to study brain energy metabolism noninvasively; over the last 30 years phosphorus-31 (31P) and proton (1H) MRS have provided unique information on cerebral energy metabolism during the evolution of brain injury following hypoxia–ischemia in the newborn infant [33], neonatal rat [34], and the newborn piglet [3, 35] (Fig 2). In human infants, shortly after intrapartum hypoxia–ischemia 31P MRS often reveals normal cerebral energetics [36]. However, in infants with adverse outcome, despite adequate oxygenation and circulation, phosphocreatine (PCr) and nucleotide triphosphate (NTP, mainly ATP) decline, and Pi increases, in the first days of life [33, 36–38]. Brain pHi has been observed to become alkaline during this phase [4, 5] (see next section). These metabolic changes were termed “secondary energy failure” (SEF) on the basis that they followed impaired intrapartum cerebral energy generation (resulting in transiently reduced PCr and NTP and increased Pi), which resolved following resuscitation [3] (Fig. 3). It was assumed that SEF was consequential to a pathological mechanism initiated by intrapartum hypoxia–ischemia or/and reperfusion/reoxygenation. 1H MRS provides complementary information to 31P MRS (in particular cerebral lactate, a marker of anaerobic metabolism and N-acetyl-aspartate (NAA), an abundant amino acid found mostly in neurons in the central nervous system (CNS) [39]). Because of the greater sensitivity of the 1H nucleus, data can be obtained from smaller regions of the brain. Cerebral lactate rises and NAA falls during transient hypoxia–ischemia; these metabolites return almost to baseline levels after successful resuscitation, only to be followed by a secondary increase in lactate and slower reduction in NAA in the hours that follow. These 1H MRS changes occur in parallel to the energy disruption (reduction in PCr/EPP and NTP/EPP) seen on 31P MRS. A recent meta-analysis of the prognostic accuracy of MR methods demonstrated that thalamic 1H MRS Lac/NAA peak area ratio acquired between days 5 and 14 after birth is a highly sensitive and specific biomarker of long-term neurodevelopmental outcome in infants with neonatal encephalopathy [40].

Fig. 2
figure 2

Representative 31P MRS spectra from a a normal baby and b a baby with severe neonatal encephalopathy. Seven main peaks can be assigned in a cerebral 31P MR spectrum: phosphomonoesters (PME), inorganic phosphate (Pi), phosphocreatine (PCr), phosphodiesters (PDE), and the three phosphate groups (α, β, and γ) in nucleotide triphosphate (NTP). The chemical shift difference between PCr and Pi forms the basis of intracellular pH measurement in the in vivo brain

Full size image
Fig. 3
figure 3

Schematic diagram illustrating the biphasic pattern of energy failure associated with a transient hypoxia–ischemic insult visualized using 31P MRS in the UCL piglet model. Nucleotide triphosphate (NTP)/exchangeable phosphate pool (EPP = Pi + PCr + NTP) is shown on the y-axis. The change in NTP/EPP during transient hypoxia–ischemia (HI), resuscitation, the latent phase (period between the recovery from acute HI and the evolution of secondary energy failure (SEF)), and SEF itself are shown. During the acute energy depletion, some cells undergo primary cell death, the magnitude of which will depend on the severity and duration of HI. Following perfusion, the initial hypoxia-induced cytotoxic edema and accumulation of excitatory amino acids typically resolve over 30–60 min with apparent recovery of cerebral oxidative metabolism (latent phase). It is thought that the neurotoxic cascade is largely inhibited during the latent phase and that this period provides a “therapeutic window” for therapies such as hypothermia and other agents. Cerebral oxidative metabolism may then secondarily deteriorate 6–15 h later (as SEF). This phase is marked by the onset of seizures, secondary cytotoxic edema, accumulation of cytokines, and mitochondrial failure

Full size image

Brain pHi Changes in Babies with Neonatal Encephalopathy

31P MRS is the only noninvasive way to measure pHi apart from positron emission tomography, which requires injection of radioactive ligands and is not feasible in babies. Using 31P MRS, brain pHi can be measured using the chemical shift difference of Pi from PCr [41], or phosphoethanolamine from PCr [42, 43] or ATP from PCr [44]. The pHi value calculated from the chemical shift difference of Pi form PCr is thought to reflect the pHi in dead or injured cells, whereas that derived from other metabolites may reflect different cell populations. A summary of the studies of brain pHi using 31P MRS performed in normal neonates and infants is shown in Table 1.

Table 1 Studies in normal neonates and infants documenting brain pHi using 31P MRS (based on [36, 38, 151–156])
Full size table

Brain pH homeostasis in the cytosol and other cellular compartments is maintained by a dynamic, finely tuned balance between proton-extruding and proton-importing processes [45]. Under physiologic conditions, the extracellular pH is ∼7.4; however, the cytosolic pH is more acid (∼7.2). Other organelles possess their own specific pH (Fig. 4). Almost all proteins depend on pH to maintain their structure and function; pH plays an important part in many metabolic functions and the charge of biological surfaces affects many cellular reactions. There is a tendency for the cytosol to accumulate acid (from metabolic reactions—ATP production in the cytoplasm by glycolysis and in mitochondria by oxidative phosphorylation) (Fig. 5). Protons are actively extruded from the cytosol by proton pumping ATPases, coupling to other substrates through exchangers; the main transporter that protects cells against acidification is the Na+/H+ exchanger (see “Brain pHi and Seizures”).

Fig. 4
figure 4

A schematic diagram showing the pH of the main cellular compartments in a typical mammalian cell (from Casey et al. [45]) The mitochondrial pH represents that in the inner mitochondrial membrane

Full size image
Fig. 5
figure 5

Ion transporters that regulate cytoplasmatic pHi. There is a tendency to acidification of the cytoplasm due to the activity of the anaerobic metabolism producing lactate from glucose and aerobic metabolism (oxidative phosphorylation in mitochondria that produce CO2). The main transporters regulating cytosolic pH are the plasma membrane Na+/H+ exchangers (NHEs) (see (1) on the diagram) and the Na+/HCO3 − cotransporters (NBCs) (2). The plasma membrane Cl−/HCO3 − or anion exchangers (AE) (3) counterbalance these mechanisms by acidifying the cell, and plasma membrane Ca2+-ATPases (4) also acidify the cytosol by exchanging cytosolic Ca2+ for extracellular H+ when intracellular Ca2+ is elevated. Importantly in the brain after hypoxia–ischemia, the monocarboxylate-H+ co-transporters (MCTs) will alkalinize the cell (5). The Na+/K+-ATPase pumps (6) establish an inward electrochemical Na+ gradient. Adapted from (from Casey et al. [45]). CA carbonic anhydrase, LDH lactate dehydrogenase, pH i intracellular or cytosolic pH, pH o extracellular or outside pH

Full size image

In 2002, we studied the chemical shift difference of Pi in whole-brain 31P MRS in relation to outcome in babies with neonatal encephalopathy [5]. In 78 babies with neonatal encephalopathy studied serially during the first year after birth (151 studies throughout the year including 56 studies of 50 infants during the first 2 weeks after birth), we demonstrated that: (1) alkaline brain pHi was associated with severely impaired outcome in babies; (2) the degree of brain alkalosis was related to the severity of brain injury on MRI and brain lactate concentration (Fig. 6); and (3) brain alkalosis persisted for several weeks in babies with a severely impaired outcome and persistence was associated with cerebral atrophy on MRI. A brain pHi of 7.15 and above measured in the first 2 weeks after birth had a sensitivity of 71 % and specificity of 92 % for predicting adverse outcome. Published studies that have measured brain pHi using 31P MRS in infants with birth asphyxia are listed in Table 2.

Fig. 6
figure 6

Association of the severity of brain injury in conventional MRI and mean brain pHi within the first 2 weeks of age in 50 infants with mild, moderate, and severe neonatal encephalopathy. Mean (SD) brain pHi at <2 weeks of age in infants with neonatal encephalopathy classified according to the brain MRI pattern (data from Robertson et al. [5]). There are significant differences between the brain pHi in the group with a normal brain MRI and the brain pHi in the group with a severely abnormal MRI

Full size image
Table 2 Studies in infants with birth asphyxia documenting brain pHi measured by 31P MRS (based on [4, 5, 33, 36, 38, 151, 154, 157])
Full size table

In vivo data suggest that the return of pHi to normal or alkaline values may be deleterious to cells that have undergone hypoxia–ischemia. This rebound alkalosis has been termed pH paradox and has been described in several cell types [46–48]. Possible mechanisms leading to pH-dependent injury include activation of phospholipases and proteases, which have an alkaline pKa, onset of the mitochondrial permeability transition pore, leading to uncoupling of oxidative phosphorylation and aggravation of ATP depletion [49], and an exacerbation of excitotoxic neuronal injury due to an increased NMDA activation at alkaline pHi [50, 51]. An understanding of one of the main transporter groups leading to brain alkalosis after hypoxia–ischemia (the NHEs) is therefore important for future neuroprotection strategies.

Na+/H+ Exchangers

The NHEs are a family of ion integrate membrane transport proteins involved in maintaining pHi and cell volume in many mammalian cell types [52, 53], by extruding protons from, and taking up sodium ions into cells in an electroneutral manner (1:1 stoichiometry) (Fig. 5).

To date at least ten NHE isoforms (NHE1 to NHE10) have been identified in mammals [54]. NHE10 was recently described as an osteoclast-specific member of NHE family, regulating osteoclasts differentiation and survival [55]. NHE1–5 are expressed on the plasma membranes in various types. NHE6–9 reside on intracellular organellar membranes of the endosomal-trans Golgi network [56, 57]. NHE isoforms have similar membrane topologies, with an N-terminal membrane domain consisting of 12 predicted transmembrane segments and a more divergent C-terminal cytoplasmic domain [45, 57]. The NHE1 isoform has been studied closely and is present at the cell surface of most cells, especially plasma membranes. It is the most abundant isoform in the CNS [58, 59]. NHE1 plays a crucial role in protecting cells from internal acidification, acting together with bicarbonate-transporting systems and restoring cell volume to steady-state levels [60].

Intracellular acidosis is the major stimulus that regulates NHE1 activity. As the H+ concentration of the cytosol rises, a rapid increase in the activity of the transporter occurs. Near-maximal velocity is achieved in approximately one pH unit (Hill coefficient >1), thereby minimizing exposure of the cytoplasm to excess acidification [61, 62].

In addition to responding to intracellular acidification, various signals can alter the NHE1 internal pH sensitivity, such as hormones, mitogens, and physical stimuli (e.g., mechanical stretch and hyperosmolarity), modulating its state of phosphorylation [45, 53, 63–67]. The exchange activity is more active at alkaline pH values due to a shift of the set point by phosphorylation of the residues. It is also regulated at the transcriptional level, supporting the control at mRNA levels and at protein production level [68, 69]. An alkaline cytoplasmic pH (pHc) is thought to provide a permissive environment for the progression of diverse cellular processes, including changes in cell shape [70], adhesion [71], migration [72, 73], chemotaxis [74, 75], and proliferation [76–78].

Effect of Hypoxia–Ischemia on NHE

Reductions in blood flow with ischemia and reduction in oxygen supply with hypoxia decrease the supply of oxygen required to maintain tissue ATP levels, especially in excitable organs such as the heart and brain that have a high demand for energy. As ATP stores are depleted, lactate, pyruvate, and protons accumulate owing to anaerobic metabolism of glycogen stores. The accompanying cytoplasmic acidification causes hyperactivation of plasma membrane NHE1 and the consequent accumulation of intracellular Na+. The Na+ overload reverses the mode of operation of Na+/Ca2+ exchange, driving excess Ca2+ into the cell. The resulting elevation in intracellular Ca2+ concentration, and subsequent activation of proteases, phospholipases, and formation of oxygen- and nitrogen-free radicals, precipitates a cascade of deleterious effects, including altered membrane excitability and contractility, generation of toxic free radicals, cellular hypertrophy, apoptosis, and necrosis—events that result in cell death [79, 80]. Activation of NHE also leads to the rapid normalization of pHi during reperfusion after hypoxia ischemia [45, 54, 81, 82].

Removal of external Na+ or reducing the external pH attenuates the postanoxia alkalization [83]. When ATP levels fall to 35 % of control in isolated CA1 neurons during chemical anoxia, NHE activity is reduced by 44 %. Activation of NHE function, Na+ pump function and the Na+ driving for NHE are all dependent upon ATP energy levels, ATP being required for phosphorylation of the protein.

NH Exchanger Blockade and Neuroprotection in Neonatal Models

Although the NHE activation is essential for the restoration of physiological pHi, hyperactivation of NHE1 in neurons, in response to the metabolic acidification associated with an ischemic–hypoxic insult [46, 84–87], disrupts the intracellular ion balance, causing intracellular Na+ and Ca2+ overload [88], which eventually leads to cell death. Experimental in vitro and in vivo studies have demonstrated neuroprotection with NHE inhibitors; these are summarized in Table 3.

Table 3 Preclinical studies—brain pH and NHE inhibitors in perinatal brain injury (based on [46, 84, 85, 88–90, 93, 101, 102, 107, 110, 111, 138, 145, 158–160])
Full size table

There are two major classes of pharmacological NHE inhibitors. One class includes amiloride and its derivatives by double substitution of the nitrogen of the 5-amino group: dimethylamiloride (DMA), N-(N-ethyl-N-isopropyl)-amiloride (EIPA), N-methyl-isobutyl-amiloride (MIA), and 5-(N,N-hexamethylene)-amiloride. They are nonselective inhibitors of the NHE system in various cell types, and EIPA is highly potent. Another class of inhibitors includes the derivatives of the benzylguanidines such as HOE 694, HOE 642 (cariporide mesilate or 4-isopropyl-3-methylsulfonylbenzoyl-guanidine methanesulfonate), eniporide, and BIIB-513, which selectively inhibited NHE1. The replacement of the pyrazine ring of amiloride by a pyridine ring or by a phenyl increased the potency and the NHE selectivity. In the last two decades, several bicyclic guanidines were prepared: zoniporide, MS-31038, SM-20220, SM-20550, SMP-300, KBR9032, BMS-284640, T-162559, TY-12533, S-3226, or SL-591227 [52]. Of these, HOE 642 is extremely potent, being 105 times more specific for NHE1 vs. NHE3 [78], and zoniporide is very potent too and 150-fold selective for NHE1 vs. the other isoforms. S1611 and S3226 are two selective NHE3 inhibitors. Harmaline is a nonamiloride NHE5 inhibitor. The inhibitory potency of NHE inhibitors towards the different NHE isoforms is described in detail in a review by Masereel et al., [52].

During in vitro cerebral ischemia, NHE inhibitors (DMA, harmaline [46], SM-20220 [88], amiloride [89], and HOE-642 [85, 87, 90]) had a protective effect, in terms of delayed pHi normalization both in neuronal cells and in astrocytes [46, 89–92], inhibition of microglia [90], and less intracellular calcium accumulation [87, 88]. This anoxia-induced alkalization was also ameliorated in NHE1−/− CA1 neurons. In parallel, cell death was reduced in wild-type neurons treated with HOE 642 or in NHE1−/− neurons in cultured mouse cortical neurons undergoing 3 h oxygen–glucose deprivation (OGD) and 21 h reoxygenation (from ∼70 % to 40–50 %) [87].

Vornov et al. [46], described that pHi decreased by 0.2 pH units, but then recovered when inhibition was removed in rat cortical neuronal cultures undergoing metabolic inhibition for 20 min. NHE1 inhibitors, dimethylamiloride or harmaline, significantly reduced the postinhibition pHi recovery, showing less brain injury in these compared to those with a fast pHi normalization.

As mentioned above, NHE1 activity also affects cellular Na+ levels. A small but significant (∼2-fold) increase in neuronal intracellular concentration of Na+ ([Na+]i) results after 2 h OGD, reaching to ∼7-fold increase during 1 h reoxygenation. HOE 642 attenuated the rise in [Na+]i following OGD. NHE1−/− neurons did not exhibit significant increase in [Na+]i. This supports that NHE1 activity is elevated upon reoxygenation [83, 86]. Three hours of OGD and 21 h reoxygenation lead to cell death in ∼70 % of the NHE1+/+ neurons. However, cell death is significantly reduced in wild-type neurons treated with HOE 642 or in NHE1−/− neurons [87].

In vivo studies mainly in rodents have found that NHE1 inhibitors reduce brain injury in hypoxia–ischemia models [93] and decrease infarct size in a focal ischemia model [87]. In NHE1 heterozygous mice, there was a significant decrease in infarct size of around 30 % compared with normal NHE+/+ mice following middle cerebral artery occlusion (MCAO) [87], similar to those NHE+/+ mice treated with HOE 642. It implies that the disruption of Na+ and Ca2+ homeostasis contributes to ischemic neuronal damage. These studies firmly demonstrate the dominant role of NHE1 among other NHE isoforms in cerebral ischemic brain damage.

Other studies show neuroprotection in ischemic models. SM-20220 (N-(aminoiminomethyl)-1-methyl-1H-indole-2-carboxamide methanesulfonate), a highly selective NHE1 inhibitor, given intravenously 1 h after MCAO significantly reduced the extent of cerebral edema and Na+ content after 2 h ischemia and 4 h reperfusion and infarct volume after 22 h reperfusion [94]. SM-20220 decreased infarct size in both transient and permanent MCAO models. The reduction of infarct size improved when the treatment is delayed for 5, 30, or 60 min after the onset of ischemia [94]. SM-20220 led to ∼50 % decrease in infarct size at 72 h reperfusion. Other groups have shown similar reductions in infarct volume and edema with other NHE blockers [95–97].

Some NHE blockers, specifically SM-20220 [98] and MIA [93], have been found to reduce the number of neutrophils in the ischemic hemisphere. SM-20220 attenuates leukocyte adhesion and migration in the mesenteric artery [99] and improves endothelial dysfunction [100]. NHE1 blockers (EIPA and HOE 642) [84, 90, 101, 102] reduced microglial activation posthypoxia ischemia and hence the microglial phagocytosis and astrocytosis in CA1 region. Cengiz et al. [102] found that HOE 642 given either prehypoxia–ischemia or starting 10 min after the insult reduced neurodegeneration in the acute stage in hippocampus, striatum, and ipsitaleral thalamus, and improved the striatum-dependent motor and spatial learning skills at 8 weeks.

NHE1 inhibitors are neuroprotective in astrocytes in a similar way to neurons. The intracellular Na+ overload is substantially mediated through activation of the ERK1/2 pathways [92]. In a more severe model of in vitro ischemia, cultured astrocytes were superfused with a solution, which mimics the ionic composition of the ischemic extracellular space (a hypoxic, acidic, ion shifted ringers (HAIR) solutions at 37 °C) [103, 104]. Exposure to HAIR caused a rapid decline in astrocyte pHi. This is consistent with the finding that NHE1 activity is inhibited by low extracellular pH (pHe) [105]. When exposed to normal buffer again, astrocytes rapidly alkalized and a significant rebound of pHi over the baseline took place [104]. Bondarenko et al. [104] have reported that HAIR and reoxygenation leads to ∼40 % cell death in astrocytes, which is significantly reduced in the presence of NHE blockers such as EIPA or HOE 694 during HAIR and reoxygenation, or during reoxygenation alone.

NHE inhibitors have also been shown to preserve neurological function after ischemic injury. In a model of ischemia and hypothermia in piglets, Castella et al. [106] observe a rapid neurological recovery in subjects receiving HOE 642 just at the onset of cooling. Several studies [84, 107] demonstrate that EIPA not only protects gerbil hippocampal neurons from ischemic injury but also reduces the magnitude of the ischemia-induced locomotor hyperactivity at both 24 h and 6 days after reperfusion. Ischemic injury to the gerbil forebrain produces an increase in locomotor activity, which is related to the degree of pyramidal neuronal damage in the CA1 region of the hippocampus [84, 108]. In another Mongolian gerbil model, the recovery time of consciousness significantly improved (lower neurological scores at 2 h postreperfusion) following a 30 min transient global cerebral ischemia in those treated with SM-20220, compared to the vehicle group [109]. This improvement in neurological deficit persisted until 24 h postreperfusion.

In summary, NHE inhibitors (or genetically modified NHE1−/−) are neuroprotective in animal models. They reduce the alkalinization of the cell (inhibition of phospholipases and kinases [107, 110] and delayed secondary energy failure [89]), and decrease the calcium and sodium intracellular overload, reducing cell edema and cell damage. They suppress microglial activation, especially in hippocampal regions (pyramidal neurons of the CA1 area in rodents) [84] and in striatum and ipsilateral thalamus in a focal ischemia model [102]. Improved functional outcome also follows NHE inhibition [102].

NHE Inhibition Following Transient Hypoxia–Ischemia in the Piglet Model of Perinatal Asphyxia

Robertson et al. [111] used the piglet model of perinatal asphyxia to study the effect of NHE1 inhibition with MIA given 10 min after the end of hypoxia–ischemia and eight hourly thereafter. MIA improved cerebral energy metabolism in the thalamic area in the 48 h after hypoxia–ischemia on MRS biomarkers and reduced cell death and microglial activation in some brain areas. The use of clinically relevant MRS biomarkers and reduced cell death in a large animal model of perinatal asphyxia is an important step towards the clinical translation of NHE inhibitors. It is important to assess whether NHE blockade would augment hypothermic neuroprotection in this devastating disease.

Brain pHi and Seizures

Relation Between Brain pHi and Seizures

Neonatal seizures are the most common manifestation of a neurological disorder in the newborn period. Neonatal encephalopathy is the most frequent cause of seizures at term. In some studies, the duration of electrographic seizures in newborn babies is associated with worse MRI appearances [112] and poor neurodevelopmental outcome [113]. Newborn babies with neonatal encephalopathy as a cause of seizures tend to have a higher seizure burden than those with a stroke [113]. Increased morbidity and mortality, brain injury, and poor neurodevelopmental outcome have been associated with a higher seizure burden [113–120]. Therefore, addressing neonatal seizures is a priority, but current drugs are largely ineffective and are not free of serious side effects on the neonatal brain [121–123].

Therapeutic hypothermia not only improves outcomes of babies with neonatal encephalopathy by reducing the rate of death and disability at 18 months of age, but it also appears to reduce seizure burden [124]. Cooling appears either to act directly to reduce seizure burden or to augment the action of conventional anticonvulsants. Xenon gas, which is currently being used in clinical phase II trials to augment hypothermic neuroprotection in babies, appears to reduce seizure burden when administered with cooling [125]. The data are consistent with those from preclinical models [124].

There is however still an unresolved controversy whether seizures worsen outcome by themselves or whether they are associated to those with a more severe degree of encephalopathy, hence the higher mortality and morbidity [126, 127]. The molecular and cellular mechanisms underlying birth-asphyxia seizures are unknown, but understanding the seizure-triggering mechanisms plays a key role in the design of novel therapeutic strategies.

A profound acidosis (blood pH 7.00 or lower) at the time of birth is an essential criterion in the diagnosis of perinatal asphyxia [128, 129]. However, postasphyxia seizures do not coincide with the maximal blood acidosis but are typically first observed during the recovery period after a delay that usually ranges in human babies from 2 to 16 h [130]. A wide range of observations have shown that changes in extra- and intracellular pH exert a strong modulatory effect on brain excitability under normal and pathophysiological conditions, whereby an alkalosis enhances excitability while an acidosis has an opposite effect [131–138]. The immature brain appears to be particularly sensitive to changes in pHi. Recent experiments have shown that changes of 0.05 pH units on neonatal hippocampal slices have a profound effect on endogenous network activity [137].

In a rat pup model of neonatal asphyxia, brain alkalosis after recovery from asphyxia played a key role in the triggering of seizures [138]. Brain pH was measured with intracortical ion sensitive microelectrodes, and EEG was recorded from parietal cortex in freely moving rat pups with simultaneous video recording [138]. There was a robust correlation between postasphyxic brain pH changes and seizure burden; that is, the more alkaline the brain pH after asphyxia, the higher the seizure burden. The alkaline brain pH overshoot after asphyxia reached values of up to 0.40 pH units above normal brain pH; these are similar finding to previous studies in rats [93], piglet [3, 111], and human newborn [4, 5]. Some pups received MIA (the NHE inhibitor) intraperitoneally 30 min before asphyxia; in those rat pups pretreated with MIA, seizure burden was reduced. These data add weight to the evidence that acidosis suppresses neuronal excitability, whereas alkalosis does the opposite [133, 139–141]. Taking this further, there is accumulating evidence that brain pHi plays a key role in modulating neuronal survival after injury [131, 142, 143].

In another study in neonatal rat pups by the same group, a rise in brain pH was accompanied by seizure activity following injection of sodium bicarbonate [144]. The authors suggested that a “graded restoration of normacapnia” should be recommended for resuscitation of neonatal encephalopathy as this may reduce seizures and improve outcome [144].

The same group explored mechanisms of postasphyxia brain alkalosis [145] in the rat pup exposed to one hour of asphyxia (simultaneous hypoxia and hypercapnia). They observed that the postasphyxia brain alkalosis was generated in the absence of a rise of blood pH. The increase in both extracellular and intracellular pH recorded in the brain implied an enhanced net loss of acid equivalents from brain tissue across the BBB. BBB disruption itself was excluded by the absence of increased sodium fluorescein extravasation into the brain and by electrophysiological characteristics of the BBB [145]. The net efflux of acid equivalents from the brain across the BBB was abolished by MIA, suppressing the postasphyxia rebound alkalosis of the brain extracellular pH without having any significant effect on the weight of the pups. Remarkably, MIA also abolished postasphyxia seizures (reduced seizure burden by 88 %) when applied preinsult in rat pups [138, 145] and ameliorated brain injury when applied before hypoxia–ischemia in neonatal mice [93]. These findings support the conclusion that activation of NHE in the BBB leads to brain alkalosis and consequent seizures.

Therapeutic hypothermia is now standard care for infants in the developed world with moderate to severe neonatal encephalopathy [146]. It is unknown if NHE inhibition combined with therapeutic hypothermia might augment hypothermic neuroprotection after perinatal asphyxia. Some in vitro studies suggest that moderate hypothermia (reducing temperature from 37 to 20 °C) itself increases NHE activity [147, 148], while others suggest that hypothermia produces a partial inhibition of NHE activity [149]. The combined effects of an NHE inhibitor with hypothermia could significantly augment hypothermic neuroprotection especially as they act on separate pathways. Importantly, studies suggest that the potency of NHE inhibition under hypothermic conditions is not influenced by a change in temperature [149] and NHE inhibition was beneficial in a perfused heart model even after hypothermic ischemia [150]. Preclinical large animal studies of combined therapy with cooling are high priority.

Summary

Perinatal hypoxic–ischemic brain injury in term babies is still a significant problem throughout the world. Therapeutic hypothermia has improved outcome, especially of those with a moderate insult. Unfortunately, around 40 % of infants treated with therapeutic hypothermia still have adverse outcomes. Experimental data suggest that the addition of another agent to cooling may enhance overall protection either additively or synergistically. Experimental models have demonstrated the central role of NHE in maintaining brain pHi and ion homeostasis, and in microglial activation in some areas of the brain. NHE inhibitors are a promising neuroprotective agent in animal models, improving brain energy metabolism in in vivo MRS, reducing cell death, seizure burden, and brain injury with improvement of functional outcome. Preclinical studies are urgently needed to assess possible augmentation of hypothermic neuroprotection with NHE inhibition.