Intermittent hypoxia and effects on early learning/memory: exploring the hippocampal cellular effects of pediatric OSA

Abstract

This review provides an update on the neurocognitive phenotype of pediatric obstructive sleep apnea (OSA). Pediatric OSA is associated with neurocognitive deficits involving memory, learning, and executive functioning. Adenotonsillectomy (AT) is presently accepted as the first line surgical treatment for pediatric OSA, but the executive function deficits do not resolve post-surgery, and the timeline for recovery remains unknown. This finding suggests that pediatric OSA potentially causes irreversible damage to multiple areas of the brain. The focus of this review is the hippocampus, one of the 2 major sites of postnatal neurogenesis, where new neurons are formed and integrated into existing circuitry and the mammalian center of learning/memory functions. Here, we review the clinical phenotype of pediatric OSA, and then discuss existing studies of OSA on different cell types in the hippocampus during critical periods of development. This will set the stage for future study using preclinical models to understand the pathogenesis of persistent neurocognitive dysfunction in pediatric OSA.

Introduction

Obstructive sleep apnea (OSA) is characterized by repeated episodes of partial or complete airway obstruction, circadian dysrhythmias, and sleep fragmentation. Several types of neurocognitive deficits have been described based on the child’s age and disease severity ^1,2. The pathogenesis of these changes, as well as why the neurocognitive changes in the pediatric population remain persistent despite treatment is a fundamental gap in our understanding of pediatric OSA ^3,4. We briefly describe the neurocognitive phenotype followed the effect of intermittent hypoxia on individual cell types in the hippocampus.

Methods

We searched PubMed and Cochrane databases for English language studies with the following keywords: Pediatric OSA, intermittent hypoxia, hippocampus, neural progenitors, ANPs, neural stem cells (NSCs), immature and mature neurons published from 2010-2020 along with relevant references within those articles. Emphasis was given to randomized clinical trials, and original controlled research studies along with articles/journals that are frequently studied by clinicians and scientists.

Overview of OSA

OSA is characterized by cyclical hypoxia/normoxia that induces reactive oxygen species (ROS) and oxidative stress ⁵. Intermittent hypoxemia causes arousal from REM sleep with hypoxic hypercapnia being the most potent stimulus⁶. Sleep becomes fragmented, leading to daytime somnolence. Furthermore, the blood brain barrier (BBB), gray matter volume, and cerebral blood flow (CBF) are altered. The BBB becomes more permeable, gray matter volumes are decreased, and CBF fluctuates leading to a number of pathogenic changes ⁷ (Figure 1). Systemic inflammation⁸ and sympathetic excitation ⁹ are additional hallmark features.

Figure 1:

Demonstrates the multifactorial effects of arousal and somnolence on memory and learning. There are multiple pathological mechanisms that are affected, leading to multipathway effects on several organ systems, including the CNS. These effects lead to a variety of clinical presentations. The classical adult presentation of OSA is with daytime somnolence, whereas in pediatric OSA a number of children present with problems in the school environment. Reproduced with permission from: Beebe DW, Gozal D. J Sleep Res 2002;11(1):1-16.

While direct data in pediatric OSA is lacking, changes in the BBB during periods of chronic inflammation have been implicated in sleep and neurodegenerative disorders in the adult population¹⁰. There are several contrasts between pediatric and adult OSA with regards to the CNS features of the disease including symptomatology and areas of brain involved (Table 1).

Table 1:

Brain areas, postulated mechanisms, and reversibility of neurocognitive dysfunction.

	Adult	Pediatric
Areas of Brain involved	• Right Middle Frontal Lobe Gyrus11    • Posterior cingulate gyrus    • Left inferior frontal gyrus12    • Bilateral hippocampi13	• Bilateral hippocampi, Right frontal cortex14
Pathogenesis of neurocognitive dysfunction	• Small vessel disease15    • CBF changes    • Systemic inflammation16	• Oxidative Stress    • Sleep Spindle Changes 17
Reversibility with therapy	• Improvement in most cognitive domains after CPAP	• Improvement in behavior after surgery, but persistence in executive function deficits3,18

Hypercarbia

Arterial PCO₂ rises during the apneic/hypopneic episodes in adult OSA¹⁹. Unfortunately, CO₂ levels are often not measured or are not reported in many studies on pediatric OSA²⁰. From sparse data available, higher baseline PCO₂ levels in pediatric OSA correlated with persistent OSA following adenotonsillectomy ²¹. But even with very severe pediatric OSA, the elevation in CO₂ is small and remains there for short periods of time in the context of total sleep time²¹. Therefore, we focus on chronic intermittent hypoxia rather than the hypercarbia, which in the context of current data appears to attenuate or has no effect on the nature of CNS injury.

Clinical Neurocognitive Phenotype of Pediatric OSA

Numerous studies in the pediatric OSA population have focused specifically on the CNS. Most of these studies are on school-aged children (5 years or older) with SDB/mild OSA^22–24. Although changes secondary to OSA have been described in several areas of the brain, the hippocampal changes are the focus of this paper as it is the mammalian center of memory, learning, and early emotional behavior.²⁵ Any changes early in the process of rapid neuronal growth have significant effects later in life²⁶, partially explaining why even mild OSA in children can manifest with learning or cognitive deficits. Also, significant clinical deficits take time to come to the attention of parents or teachers as learning and memory integration is an ongoing process, ²⁷. What mediates ongoing learning and memory deficits post-AT remains unclear²⁸, and a preclinical model of the disease is urgently required. Given that early learning and memory are hippocampally mediated in the young²⁹, the remainder of this article will focus on pre-clinical studies in the hippocampus, specifically the effects of hypoxia on the various cell types.

Adult disease affects several key brain areas through both local mechanisms and effect on vasculature. The effect of these factors on brain areas in the pediatric population remain unknown, however there are changes in sleep spindles. Neurocognitive morbidity is seen in both populations. For example, with dementia, as seen in the adult population, there is improvement in cognitive metrics after CPAP therapy. In children however, there are not notable increases in learning memory function after intervention.

Hippocampus

The hippocampus is one of two neurogenic niches for postnatal mammalian neurogenesis ³⁰, a dynamic throughout the lifespan ³¹ and has numerous functions including the generation of new memories, learning, and pattern separation which are a function of newborn neurons ³². The neural stem cells (NSCs) reside in the sub granular zone (SGZ) of the dentate gyrus and give rise to amplifying neural progenitors (ANPs), which rapidly divide and eventually give rise to neuroblasts. Neuroblasts demonstrate high rates of apoptosis ³³ and those that survive differentiate into immature neurons and subsequently mature neurons, which integrate into the hippocampal circuitry ³⁴ (Figure 2).

Figure 2:

Various cell types and the effect of hypoxia on these cell types in the mammalian hippocampus. Hypoxia has different effects at different parts of the neurogenic cascade, and promotes apoptosis beyond the NSC stage. The morphology of each cell type under intermittent hypoxia has not been studied.Therefore, while varying levels of hypoxia may effect different areas of the cascade differently, the effects of intermittent hypoxia, with consequent ROS formation has not been elucidated. There could be differences in number, morphology, and function, all of which need to interrogated to properly understand how the mammalian hippocampus responds to intermittent hypoxia.

They influence both regional physiology and the functional connectivity of the hippocampus with more distant brain regions, such as the prefrontal cortex, amygdala, and other structures within the limbic system.³⁵ Pattern separation, which is a critical function of the hippocampus, facilitates temporal event discrimination, spatial processing, and short-term memory storage ³⁶.

The role of Reactive oxygen species

Reactive Oxygen Species (ROS) are generated during oxidative metabolism in mitochondria and consist of, amongst others, hydrogen peroxide, radical and non-radical oxygen species. Excess ROS formation, which occurs in OSA³⁷, leads to oxidative stress damage(Figure 3).

Figure 3:

The interplay between ROS, HIF-α, Superoxide dismutase (SOD), haemoxygenase 2(HO2), and pathophysiological changes including sympathetic nervous system activation. HIF-α has downstream effects on several other pathways as well. The effects of ROS in the context of the pediatric OSA, specifically with regards to the intermittent hypoxia stimulus, remain unknown. There is a convergence on the end-point of both SNS/adrenal medulla activation, which leads to a higher level of circulating catecholamines in these patients. This has been one of the postulated mechanisms of hypertension in both pediatric and adult OSA. AM: Adrenal Medulla, SNS: sympathetic nervous system, Nox2 (NAPDH oxidase 2), PKG: Protein Kinase G, PKC: Protein Kinase C. With permission¹¹³

ROS have a dual role in that they are vital to cell repair and longevity ³⁸, via both effects on transcription and synaptic activity. ROS effect transcription of several key gene mediators leading to several local and systemic effects. Acute intermittent hypoxia also does not lead to neuroinflammation ³⁹. However, continuous exposure is detrimental to CA-1 neurons in the hippocampus, which are selectively vulnerable to hypoxia in adult rodent models of OSA⁴⁰. CA-1 neurons show delayed injury pattern with impaired cellular metabolism⁴¹, from mitochondrial dysfunction⁴² as well as endoplasmic reticulum (ER) stress^43,44.

Preclinical Rodent Models of Adult OSA

There have been two model types described for adult OSA. The first is the intermittent hypoxia model. This model demonstrated that after intermittent cyclical hypoxia for 14 days, adult rats demonstrated deficits in the Morris Water Maze Task (MWM)⁴⁰. This test interrogates spatial memory which is a hippocampal function, and postmortem studies demonstrated CA-1 deficits. Follow up studies have revealed abnormalities in dentate gyrus (DG) function as well⁴⁵. This exact model was replicated in younger rats⁴⁶, utilizing a moderate-severe clinical phenotype, which are not the children in whom neurobehavioral deficits are typically seen. The second model is the tracheal balloon occlusion model, which utilizes a surgically implanted tracheal balloon with intermittent inflation⁴⁷. This model results in hypoxia and hypercarbia as seen in OSA. The latter model has not been neurobehaviorally validated. To date, there is no preclinical model recapitulating the neurocognitive phenotype of pediatric OSA. This has at least partially been due to a lack of validated learning and memory paradigms in neonatal mice. For example, MWM is not technically feasible due to hypothermia risk, and object recognition is not validated for use in this age group⁴⁸. Accordingly, to enable further study of pediatric OSA, an age-appropriate animal model is necessary.

Oxygen tension in the rodent neonatal hippocampus

Hippocampal O₂ levels are far lower than the alveolar oxygen levels of 21% (room air/normoxia) ⁴⁹ however remain exquisitely vulnerable to hypoxic injury ⁵⁰. In the hippocampus normoxia is 8%, mild hypoxia is 5-8%, moderate hypoxia is oxygen at 1-5%, and less than 1% is severe hypoxia. Pimonidazole HCl labelling of hypoxic cells demonstrated that much of the SGZ lies within hypoxic zones which is believed to be less than 20 mm Hg⁵¹. Relative hypoxia in vitro (<5%) drives proliferation of neural progenitor and precursor cells^51,52 (Figure 4).

Figure 4:

The effects of various levels of hypoxia on progenitor cell behavior. Severe continuous hypoxia seems to facilitate astrocytic transformation, whereas moderate hypoxia promotes precocious differentiation.Since self-renewal is important in the maintenance of the stem cell pool, both forms of hypoxia cause depletion of the progenitor pool leading to a net loss of available NSCs. The long term effect of progenitor pool depletion on lifelong learning and memory is not well understood. Xie Y, Lowery W. Methods 2018 with permission.

However, once these cells migrate away from hypoxic zones usually due to differentiation they seem to lose the hypoxic “protections” and are more susceptible to apoptosis ^53–55. All the studies were performed using continuous hypoxic conditions; the effects of intermittent hypoxia on early stem cell growth remains unknown. These findings demonstrate that oxygen balance within the dentate gyrus is tightly regulated and a small change in either direction has a profound effect on NSC behavior ⁵⁶. This balance is at least partially mediated by uncoupling protein 2(UP2), which effected mitochondrial-ER connections in the rodent cortex and hippocampus⁵⁷, as well as by nitric oxide(NO) for perivascular cells⁵⁸.

HIF-α is central to the cellular response to hypoxia⁵⁹. HIF-α protects against hypoxic insult and allows for injury attenuation⁶⁰. HIF-α inactivation causes impaired neurogenesis and learning deficits⁶¹ even in the absence of hypoxia. Under hypoxic conditions, HIF-α promotes NSC proliferation⁶². This is through HIF-α effect on numerous mechanisms/canonical pathways including p53, Notch, Wnt/beta-catenin, and Oct4 ^63–67.

Hypoxic Effect on Neural Progenitors

Santilli et al. ⁶⁸demonstrated that severe hypoxia encouraged early ANP production from NSCs as compared to normoxic conditions. The same study showed that continuous in-vitro mild hypoxia drives self-renewal of NSCs. Similarly, deFillipis ⁶⁹ demonstrated in vitro that mild hypoxia encouraged ANP production from NSCs whereas severe hypoxia facilitated apoptosis and quiescence. Chronic continuous hypoxia facilitates the survival and proliferation of progenitor cells [49]. In Notch knockout mice, the survival effect is not present, demonstrating that Notch1, at least in part, may mediate progenitor survival [85]. Furthermore, although hypoxic preconditioning increases differentiation of NSCs⁷⁰ , the presence of co-existing hypertension had no added effect on the differentiation ⁷¹, suggesting that change in CBF/small vessel disease contributes minimally to the behavior of NSCs. Even perivascular placement, with greater access to blood supply did not seem to confer any survival advantage⁶⁵.

ANPs have a precipitous drop in population after an acute hypoxic ischemic insult ⁷². However, self-renewal begins after a brief dip in the progenitor pool ⁷³. Chronic severe hypoxia, however, appears to produce the opposite effect, decreasing the pool of ANPs, which in turn decreases the number of neuroblasts and immature/mature neurons as a downstream effect ⁷⁴.

Intermittent hypoxia leads to an increase in progenitors; however, with a lower number of newborn neurons compared to normoxia in adult murine OSA models ⁴⁵. Intermittent hypobaric hypoxia also increases the number of progenitors and newborn neurons in the adult rodent hippocampus ⁷⁵, again suggesting that changes in oxygen tension and baricity (pressure) within a small range can cause large shifts in the progenitor pool. Given the sensitivity of the progenitor pool to minor O2 tension differences, work is required on the effect of IH on the progenitor pool in the young.

Hypoxic Effect on Neuroblasts

Neuroblasts are exquisitely vulnerable to hypoxic injury⁷⁶. It has been demonstrated that immature but committed neuroblasts die after an acute hypoxic insult, followed by proliferation of ANPs that eventually replace the lost neuroblast pool ⁷⁶ Because progenitors divide rapidly, a fast turnover of the neuroblasts can rescue the damage from the hypoxic injury ⁷⁷. Therefore, despite the sensitivity of certain neuroblast populations to hypoxic injury, there are multiple mechanisms by which this population is able to repopulate⁷⁸.

Hypoxic Effect on Immature Neurons

The mechanisms of injury and death of immature neurons is age-dependent and changes from apoptosis in young mammals to necrosis in older mammals ⁷⁹. Both GABA and glycine have been noted to be critical mediators ⁸⁰, as well as Notch⁸¹. Neuronal death can be attenuated with iron chelators and free radical scavengers ⁸², suggesting that ROSs are involved ⁸³. The window of maximal damage is in postnatal days 2-5 in mice (corresponding to birth to 6 months age window in humans) ^84,85. Mild hypothermia has been demonstrated to be somewhat protective after intermittent hypoxia at this stage ⁸⁶.

Hypoxic Effect on Mature Neurons

Chatzi et al ⁵³ used adult mice to test for the effects of hypoxia on newborn neurons. They demonstrated that these cells vacillate between continuous low-oxygen-tension environments, which promote early precursor proliferation, and continuous higher-oxygen-tension environments, which may be deleterious for newborn neurons. Taken together, these data suggest that oxygen tension is a critical component of newborn neuron fate and that the balance between hypoxia and normoxia is carefully maintained for newborn neurons ^68,81 Given that integration into the hippocampal circuit is a competitive process, and not all newborn neurons are guaranteed survival benefit⁸⁷, oxygen tension being a key mediator of survival. This further suggests that any disruption in oxygen tension may have effects not only on the neuronal differentiation process, but also their ability to contribute to hippocampal circuitry and function.

CA-3 neurons are more resistant to hypoxia vs CA-1 neurons, postulated to be due to Ca⁺² mediation, ⁴⁰ although the mechanisms are multifactorial⁴¹. Multiple mechanisms of hippocampal neuronal alterations have been suggested including c-AMP-protein kinase A signaling and CREB-mediated gene transcription [80]. Furthermore, glutamate mediates CA-1 neuronal apoptosis to hypoxia in guinea pig apnea models ⁸⁸. Magnetic resonance spectroscopy (MRS) has also demonstrated that N-acetyl-aspartate (NAA, neuronal integrity biomarker) and choline (membrane turnover) were decreased in rodents exposed to intermittent hypoxia from P2-P12 [83]. This effect was validated in humans, where change in choline levels in the midbrain were correlated with an increased in excitotoxic biomarkers including glutamate⁸⁹. Cofilin, an actin-binding protein which disassembles actin filaments, mediates dendritic spine loss and decreased hippocampal neuronal plasticity in CA-1 neurons ⁹⁰. Elevated cofilin activity from sleep deprivation is caused by cAMP-degrading phosphodiesterase-4A5 (PDE4A5), which hampers cAMP-PKA-LIMK signaling. Inhibition of cofilin activity prevented spine loss and increased plasticity. Overall, there is a clear decline in the number of mature neurons after hypoxic insult, depending on degree and duration of hypoxemia ⁹¹. While adult OSA models have demonstrated disorganization of CA-1 neuron architecture after IH exposure⁸⁵, the pediatric neuronal changes remain unknown.

Hypoxic Effect on Glial Cells

Glial cells (astrocytes, oligodendrocytes, and microglia) have scavenging and neuroprotective functions. Astrocytes have been found to be relatively more resistant to hypoxia than are microglia, through multiple mechanisms ⁹².

Astrocytes

Intermittent hypoxia in rats produces hippocampal astrocytosis with a high rate of neuronal cell death; however, chronic hypoxia increases both neuronal survival and expression of S100β (secreted by astrocytes or by injured neurons) ⁹³ as compared to acute hypoxia. NADPH oxidase(Nox), a mammalian enzyme involved in both astrocytic and microglial phagocytic burst ⁹⁴, has been demonstrated to be putative to several processes including hypoxia/reoxygenation injury, carbonylation, and pro-inflammatory states⁹⁵. NADPH has been demonstrated to be involved in the pathogenesis of the neurocognitive deficit seen in adult OSA ⁹⁶. A postulated mechanism is due to upregulation of CHOP (CCAAT enhancer binding homologous protein), which mediates HIF-α, Nox 2, oxidative and apoptosis markers⁹⁷. Hypoxia also stimulates astrocyte expression of aquaporin 4 (AQ4), which leads to local edema ⁹⁸. Aquaporin 1 is also increased in areas of high content of astrocytes, suggesting that cytotoxic edema is central to pathogenesis in OSA ⁹⁹.

At least a part of astrocytes’ initial tolerance to hypoxia is conferred by upregulation of glucose transport and uptake in both astrocytes and mature neurons ¹⁰⁰, postulated to be due to osmotic and electrochemical gradients¹⁰¹ and altered transport of other ions such as Na+ and neurotransmitters including glutamate across the cell junctions¹⁰². However, altering RAGE (receptor for advanced glycosylation end products) and nuclear factor kappa b (NF-kB), which are involved in converting astrocytes into a pro-inflammatory phenotype, reduced the neurocognitive deficit in adult male mice ¹⁰³. Taken together, these suggest that astrocytic function is central to continued neuronal survival and health in OSA.

Oligodendrocytes

Oligodendrocytes myelinate axons and are found only in the CNS. After intermittent hypoxic exposure in neonatal rats, pathological findings include decreased oligodendrocyte markers ¹⁰⁴. Several myelin proteins (MBP, PLP, MAG, and CNPase) were also down-regulated after exposure to intermittent hypoxia, suggesting arrest of oligodendrocyte maturation ¹⁰⁵. Oligodendrocytes have also been demonstrated to be a key player in memory consolidation in mammals¹⁰⁶. Therefore, any arrest or injury would not only affect motor function, but memory as well.

Microglia

Microglia mediate central nervous system ROS through NADPH oxidase, mitochondria, and excitatory neurotransmitters ¹⁰⁷. Given the neurocognitive damage from OSA is at least partially mediated through ROS, there is a yet to be elucidated role for microglia in the pathogenesis of OSA induced neurocognitive dysfunction. Since microglia are also involved in synaptic pruning, perturbations in microglial activity have been postulated in several psychiatric disorders¹⁰⁸. Suppression of microglia activation after sleep deprivation has been demonstrated to lead to improvement in adult neurogenesis and spatial memory¹⁰⁹.The implication is that microglial activation is a key player in sleep deprivation based neurocognitive dysfunction.

Furthermore, TLR4, MAPK, transcription factors, and epigenetic factors all have been suggested to contribute to microglial activation ¹¹⁰. In animal models, the injection of LPS in stressed rats, but not unstressed rats, activate the pro-inflammatory microglial phenotype, leading directly to cell injury ¹¹¹. This finding suggests that the interplay of oxidative stress and other environmental factors including ROS may be central to the pathogenesis of neurocognitive change, especially in the young ¹¹². In summary, glial cells play critical roles in the post-hypoxic neuronal injury models; however, much remains to be elucidated about their function, especially in the pediatric population.

Conclusion:

Intermittent hypoxia in the young may have long lasting neurocognitive effects. These changes, while affecting multiple areas of the brain, are more pronounced in the hippocampus, which is the mammalian center for early learning and memory. These changes can be attributed to several constitutive changes in hippocampal cell types and function. Given the connections between the hippocampus and other areas of the brain, the changes can be more widespread. While the clinical reversibility of these changes is unknown, there is an urgent need for a preclinical model to further study this common disorder of childhood.

Abbreviations:

ANP

(amplified nuclear progenitors)

AT

(adenotonsillectomy)

BBB

(blood brain barrier)

CBF

(cerebral blood flow)

CIHH

(chronic intermittent hypoxia and hypercarbia)

CNS

(central nervous system)

OSA

(obstructive sleep Apnea)

ROS

(reactive oxygen species)

SES

(socioeconomic status)

REM

(rapid eye movement)

NREM

(non-rapid eye movement)

SGZ

(sub granular zone)

NSC

(neural stem cells)