Impact of inflammation on developing respiratory control networks: rhythm generation, chemoreception and plasticity

Abstract

The respiratory control network in the central nervous system undergoes critical developmental events early in life to ensure adequate breathing at birth. There are at least three “critical windows” in development of respiratory control networks: 1) in utero, 2) newborn (postnatal day 0–4 in rodents), and 3) neonatal (P10–13 in rodents, 2–4 months in humans). During these critical windows, developmental processes required for normal maturation of the respiratory control network occur, thereby increasing vulnerability of the network to insults, such as inflammation. Early life inflammation (induced by LPS, chronic intermittent hypoxia, sustained hypoxia, or neonatal maternal separation) acutely impairs respiratory rhythm generation, chemoreception and increases neonatal risk of mortality. These early life impairments are also greater in young males, suggesting sex-specific impairments in respiratory control. Further, neonatal inflammation has a lasting impact on respiratory control by impairing adult respiratory plasticity. This review focuses on how inflammation alters respiratory rhythm generation, chemoreception and plasticity during each of the three critical windows. We also highlight the need for additional mechanistic studies and increased investigation into how glia (such as microglia and astrocytes) play a role in impaired respiratory control after inflammation. Understanding how inflammation during critical windows of development disrupt respiratory control networks is essential for developing better treatments for vulnerable neonates and preventing adult ventilatory control disorders.

1. Introduction

At birth, breathing must rapidly begin to maintain blood gas homeostasis. For this to occur, the neural circuitry controlling breathing must begin developing in utero, yet continues to mature over the first few weeks in rodents and months in humans (reviewed in Greer, 2012; Feldman et al., 2013; Del Negro et al., 2018). These maturational changes are associated with “normal” development of respiratory control networks and represent “critical windows” in respiratory control, where key structural or functional developmental events occur (Carroll, 2003). During embryonic and early life, essential developmental processes occur, such as cell genesis, migration, synaptogenesis, circuit formation, myelination and circuit refinement (reviewed in Rice and Barone, 2000). Since many of these processes continue to develop postnatally, insults early in life have the ability to derail important developmental processes, resulting in chronic impairments in adult respiratory control (Rice and Barone, 2000). In this review, we will highlight three critical developmental windows for respiratory control during perinatal development (Fig. 1): (1) genesis and initial network connections during the “in utero critical window” (reviewed in Greer, 2012), (2) increased inhibitory signaling just after birth (rodent postnatal day 0–4, P0–4) (Sterni et al., 1999; Liu and Wong-Riley, 2010a) during the “neonatal critical window”, (3) decreases in excitatory signaling and increases in inhibitory signaling shift the respiratory control network transiently towards inhibition at P10–13 in rodents during the “newborn critical window” (reviewed in Wong-Riley et al., 2013; Bavis and MacFarlane, 2017). This review will discuss fundamental aspects of developing respiratory circuitry as it pertains to these critical developmental windows, and recent work unraveling the mechanisms behind how one particular insult, inflammation, undermines respiratory control. This is not an exhaustive review on all critical windows in respiratory control, but instead focuses on three of the most well-studied critical windows.

Fig. 1. Critical windows in respiratory control development in humans (A) and rodents (B) in the central nervous system.

At least three critical windows of respiratory control network development exist: gestation, newborn, and neonatal. In humans (A), the neuromuscular junction is formed between the 5th and 7th gestational weeks (Wells et al., 1948). Fetal breathing movements (FBM) begin at gestational week 10 and continue until birth (Trudinger and Knight, 1980). Two disorders associated with respiratory control occur during these critical windows: apnea of prematurity (Kitterman, 1996) and Sudden Infant Death Syndrome (SIDS) (Goldberg et al., 2018). Peak incidence of SIDS corresponds with a decrease in serotonin signaling (Cummings and Hodges, 2019). In the rat (B), neurogenesis of motoneurons occurs from embryonic day (E)10–15 (Greer and Funk, 2005; Babiuk et al., 2003) neurogenesis and migration of the preBötzinger Complex (preBötC) from E10–17 (Pagliardini et al., 2003), and formation of the neuromuscular junction at E15–16 (Babiuk et al., 2003). At E17, respiratory rhythm generation begins and drives fetal breathing movements (Greer, 2012). The circuitry undergoes refinement until postnatal day (P)14 (Wong-Riley et al., 2019). During the newborn critical window, the dominant inspiratory oscillator switches from the parafacial respiratory group (pFRG) to the preBötC, network inhibition increases and the hypoxic ventilatory response (HVR) is blunted (Liu et al., 2006), contributing to increased apneas. From P10–13, an increase in network inhibition and decreased excitation highlights a period vulnerability in respiratory control (Wong-Riley et al., 2019, Wong-Riley et al., 2013). Lasting consequences of neonatal inflammation occurs in both humans (A), where it increases the incidence of adult obstructive sleep apnea (McNamara and Sullivan, 2000), increased age-related cognitive decline (Yagami et al., 2016), and increased risk of developing neuropsychiatric disorder (Rantakallio et al., 1997, Hornig et al., 1999). While in rats (B), neonatal inflammation abolishes adult respiratory motor plasticity (Hocker et al., 2019), hippocampal-dependent learning (Cooper et al., 2018) and blunts neuroimmune responses (Mouihate et al., 2010, Spencer et al., 2011).

Inflammation is a common early life insult (Camacho-Gonzalez et al., 2013) and is associated with breathing deficits, such as apneas (Stojanovska et al., 2018). Preterm infants are particularly vulnerable to inflammation, where greater than 65% of preterm infants have at least one infection during their hospital stay (Stoll et al., 2002, 2004). Preterm infants with inflammation have further increased apnea frequency and are at higher risk of mortality (reviewed in Di Fiore et al., 2013). Half of preterm infants experience recurrent apneas, which contribute to early life inflammation and exacerbate respiratory deficits, such as irregular breathing patterns and blunted chemosensitivity (Poets et al., 2015; Poets and Southall, 1994). Further, neonatal infections account for approximately 1/3 of all newborn deaths in the first week of life (Black et al., 2010; Chan et al., 2015), highlighting the devastating consequences of neonatal inflammation. Neonatal inflammation can arise from diverse stimuli including: group B streptococcus (Santos-Sierra et al., 2006), respiratory syncytial virus (Kurt-Jones et al., 2000), chorioamnionitis (Yan et al., 2019), sepsis (Redondo et al., 2014), meningitis (Tellería-Orriols et al., 2014), pneumonia (Tellería-Orriols et al., 2014), apnea of prematurity (Di Fiore et al., 2013), intermittent hypoxia (Johnson et al., 2018), maternal separation (Gulemetova and Kinkead, 2011), lung injury (Litvin et al., 2018) and even maternal obesity (Du et al., 2012). Additionally, many neonatal respiratory diseases and disorders are commonly associated with disruption of respiratory control networks, such as apnea of prematurity (Martin and Wilson, 2012; Di Fiore et al., 2013), respiratory failure preceding SIDS (Mitchell and Stewart, 1997; Paterson et al., 2006; Moon et al., 2007; Kinney and Thach, 2009), and childhood sleep disordered breathing (Kapsimalis and Kryger, 2002). These diseases and disorders are more prevalent in males, suggesting sex-specific vulnerability. Yet despite the prevalence of neonatal inflammation, our understanding of its impact (acute or chronic) on respiratory control remains incomplete.

With improved medical interventions, more infants (particularly those born preterm) are surviving neonatal inflammation (Blencowe et al., 2012; Glass et al., 2015). These infants have increased vulnerability to both short- and long-term neurodevelopmental disabilities (Dammann et al., 2002; Stoll et al., 2002; Ferreira et al., 2014; Patro et al., 2015; Zanghi and Jevtovic-Todorovic, 2017). The long-term effects of neonatal inflammation include: blunted immune function (Mouihate et al., 2010; Spencer et al., 2011), increased stress reactivity (Butkevich and Mikhailenko, 2016), impaired plasticity (such as hippocampal plasticity, reviewed in Cooper et al., 2018), increased incidence of neuropsychiatric disorders (Rantakallio et al., 1997; Hornig et al., 1999), increased incidence of adult obstructive sleep apnea (OSA) (McNamara and Sullivan, 2000), and increased age-related cognitive decline (Yagami et al., 2016). Yet, the lasting impact of neonatal inflammation on adult respiratory control is only beginning to be investigated (Genest et al., 2004; Hocker et al., 2019).

This review will highlight studies investigating how inflammation acutely in neonates and adults disrupts the respiratory control network, and induces chronic deficits in respiratory control. Given the importance of the respiratory system for homeostasis, understanding the mechanisms underlying the negative impact of inflammation on respiratory control is imperative to develop better therapeutic interventions for all age groups.

1.1. The respiratory control network: rhythm generation, chemoreception and plasticity

Three key aspects of respiratory control are respiratory rhythm generation, chemoreception, and respiratory plasticity (reviewed in Feldman et al., 2003). To maintain blood gases throughout life, consistent and rhythmic breathing must be maintained and is controlled by rhythm generating centers, such as the preBötzinger Complex, in the medulla (Smith et al., 1991; Cui et al., 2016). The respiratory control network must be robust, but also adaptable to respond to changing external and internal conditions (reviewed in Fuller and Mitchell, 2017). This adaptability arises from both short-term modulation associated with chemoreception and long-term plasticity within respiratory control networks. We are beginning to appreciate that insults, such as inflammation, disrupt each of these three elements and limit the ability to maintain adequate breathing, highlighting the need for more studies to understand the complex interactions between respiratory control circuitry and inflammation. Here, research investigating how inflammation disrupts respiratory rhythm generation, chemoreception and respiratory plasticity acutely during development and causes chronic impairments into adulthood will be summarized.

1.1.1. Respiratory rhythm generation

The nuclei in the brainstem responsible for respiratory rhythm generation continues to be a heavily studied subject (reviewed in Dick et al., 2015; Feldman and Kam, 2015; Del Negro et al., 2018; Ramirez and Baertsch, 2018). Work to-date indicates the preBötzinger Complex (preBötC) is the primary inspiratory rhythm generator (Smith et al., 1991; Pace et al., 2007; Cui et al., 2016 and reviewed in Feldman et al., 2003; Gourine and Funk, 2017; Del Negro et al., 2018). Signals generated within the preBötC are transmitted to relevant respiratory-related muscles, such as the diaphragm, external intercostals, and the upper airway muscles to control air flow into the lungs (reviewed in Feldman and Kam, 2015; Del Negro et al., 2018). The preBötC is composed primarily of excitatory, rhythmogenic, propriobulbar neurons and pattern-forming interneurons to initiate periodic inspiratory drive and coordinate all phases of the breathing cycle (reviewed in Feldman and Kam, 2015). While expiration is generally a passive phase of the respiratory cycle, during times of increased metabolic demand (such as exercise), active expiration is driven by the parafacial respiratory group (pFRG) (Huckstepp et al., 2018). Recently, a new rhythm generating region, the post-inspiratory complex (PiCo), has been identified as contributing to post-inspiratory activity, though less is known about this region (Anderson et al., 2016). Development of each of these portions of the respiratory control network is discussed in detail below.

1.1.2. Chemoreception

Respiratory control networks maintain blood gas oxygen and carbon dioxide levels within homeostatic ranges to sustain life (reviewed in Guyenet and Bayliss, 2015; Prabhakar and Peng, 2017). Chemosensitivity has been reviewed in detail elsewhere for oxygen chemoreception (Barnett et al., 2017; Gourine and Funk, 2017; Prabhakar and Peng, 2017) and carbon dioxide chemoreception (Hodges and Richerson, 2010; Teran et al., 2014; Funk et al., 2015; Guyenet and Bayliss, 2015; Forsberg and Herlenius, 2018). In brief, decreases in arterial partial pressure of O₂ activate peripheral chemoreceptors in the carotid body (Gonzalez et al., 1994). The carotid body projects to medullary respiratory nuclei, including the nucleus tractus solitarius (NTS), dorsal motor nucleus of the vagus (DMV) and the medullary raphe to increase respiratory drive (reviewed in Pilowsky, 2014). The hypoxic ventilatory response consists of two phases: phase I - the early phase of the hypoxic ventilatory response is characterized by a steep increase in ventilation (increased frequency and tidal volume), mainly mediated by the carotid body (reviewed in Pamenter and Powell, 2016). While phase II - the late phase of the hypoxic ventilatory response is characterized by ventilatory depression, mainly mediated by central respiratory nuclei, such as the retrotrapezoid nucleus (RTN) and NTS (reviewed in Gourine and Funk, 2017; Prabhakar and Peng, 2017). In addition to O₂ chemosensitivity, the partial pressure of CO₂ is a key factor influencing breathing (reviewed in Guyenet and Bayliss, 2015). The RTN is indirectly activated by high arterial partial pressure of CO₂ via intrinsic proton receptors on RTN neurons and P_2Y receptors on astrocytes in the RTN (reviewed in Gourine and Funk, 2017). During hypercapnia, activation of neurons in the RTN, the NTS, DMV and the medullary raphe stimulate the preBötC to increase breathing rate (reviewed in Guyenet and Mulkey, 2010; Moreira et al., 2015). Collectively, these regions work to sustain adequate breathing throughout life, and disruptions to chemoreception limit the ability to maintain homeostasis.

1.1.3. Respiratory plasticity

Respiratory plasticity is defined as a change in future behavior or activity based on prior experience (Mitchell and Johnson, 2003). While the physiological role of plasticity is still under investigation, it likely compensates and stabilizes breathing during pathology (Fuller and Mitchell, 2017). Respiratory plasticity has been observed as lasting increases in carotid sinus nerve activity, amplitude (motor plasticity) or frequency (frequency plasticity) in respiratory-related nerves (reviewed in Feldman et al., 2003; Baker-Herman and Mitchell, 2008) and medullary respiratory centers (Pamenter and Powell, 2016).

One commonly studied model of adult respiratory motor plasticity is long-term facilitation (LTF), evoked by acute intermittent hypoxia (AIH, 3 × 5 minutes of hypoxia, 5–10.5% O₂). At least two working models describe the cellular mechanisms of LTF (reviewed in Perim and Mitchell, 2019). The “Q-pathway” is elicited by moderate AIH (mAIH, 3 × 5 minutes of hypoxia, PaO₂ = 35–45 mmHg) (Baker-Herman et al., 2004). Moderate AIH activates medullary raphe nuclei, inducing release of serotonin in the vicinity of the phrenic motoneurons, which activates G_q protein-coupled serotonin (5-HT₂) receptors (Bach and Mitchell, 1996; Ling et al., 2001; Baker-Herman and Mitchell, 2002; Macfarlane et al., 2009). This activation of serotonin receptors results in a cascade requiring spinal ERK-MAP kinase activity (Hoffman et al., 2012), new BDNF protein synthesis (Baker-Herman and Mitchell, 2002; Baker-Herman et al., 2004), activation of TrKB receptors (Dale et al., 2017), and increased activity of protein kinase C-θ (Devinney et al., 2015). Alternatively, the “S-pathway” is evoked by severe acute intermittent hypoxia (sAIH, 3 × 5 minutes of hypoxia, PaO₂ = 25–35 mmHg) (Nichols et al., 2012; Seven et al., 2018). Severe AIH activates G_S protein-coupled metabotropic receptors, such as adenosine 2A (A_2A) and serotonin (5-HT₇) receptors (Golder et al., 2008; Hoffman and Mitchell, 2011). The mechanisms underlying S-pathway plasticity include increased cyclic adenosine monophosphate (cAMP) signaling (Fields and Mitchell, 2017), activation of exchange protein activated by cAMP (EPAC) (Fields, Springborn and Mitchell, 2015), Akt and mammalian target of rapamycin (mTOR) (Dougherty et al., 2015). In contrast to the Q-pathway, the mechanisms underlying S-pathway-evoked plasticity require new synthesis of TrkB, not BDNF (Golder et al., 2008; Seven et al., 2018). These two pathways, however, do not occur independently; co-activation of Q- and S-pathway-evoked plasticity results in cross-talk inhibition, where the S-pathway inhibits the Q-pathway-evoked plasticity through PKA (Fields and Mitchell, 2017; Perim et al., 2018, 2019), limiting or even abolishing expression of respiratory motor plasticity (reviewed in Devinney et al., 2016; Perim and Mitchell, 2019). AIH is being exploited therapeutically as a promising treatment to enhance motor recovery following spinal cord injury (Christiansen et al., 2018). However, enhanced motor recovery is not observed in all spinal cord injury patients (Christiansen et al., 2018), indicating impaired respiratory motor plasticity may limit its use as a treatment in these patients. This inhibition of plasticity may be a result of cross-talk inhibition or inhibition by other signaling mechanisms, such as inflammation. We propose that enhancing our understanding of how inflammation influences plasticity has the potential to improve therapeutic outcomes for spinal cord injury patients.

While much of the research to date on respiratory plasticity has focused on adults, respiratory plasticity also occurs in neonates (Blitz and Ramirez, 2002; Bocchiaro and Feldman, 2004; Lovett-Barr et al., 2006). In neonatal (P0–4) brainstem spinal cord preparations, plasticity is induced by episodic serotonin (5-HT, 5μM), resulting in lasting increases in C4/C5 burst amplitude (Lovett-Barr et al., 2006). Neonatal plasticity can also be evoked by episodic α-methyl-5-HT (Bocchiaro and Feldman, 2004), episodic anoxia in rhythmic slice preparations to induce augmented hypoglossal burst amplitude (Blitz and Ramirez, 2002) or frequency plasticity evoked by intermittent application of substance P (Johnson et al., 2019). However, episodic anoxia is the only stimulus tested to induce a lasting increase in hypoglossal burst frequency, supporting the hypothesis that mechanisms to frequency plasticity are distinct from amplitude motor plasticity (Baker-Herman and Mitchell, 2008; Fields et al., 2019). In vitro plasticity can be evoked in neonatal rodents as young as P0, but in vivo ventilatory LTF (lasting increase in minute ventilation) following acute intermittent hypoxia (10 episodes, 1 minute each of hypoxia, 5% O_2, 90 minutes) is not observed until P10 (Julien et al., 2008). Ventilatory LTF emerging at P10 indicates the mechanism(s) involved in neonatal plasticity are not fully functional at birth and develop postnatally.

1.2. Critical windows in respiratory control

At least three critical developmental windows exist early in life in rodents: in utero, newborns (P0–4) and neonates (P10–13), where vital developmental processes occur and insults, such as inflammation, have the potential to disrupt normal function of the respiratory control network (Fig. 1) (Olsson et al., 2003; Rourke et al., 2016; Johnson et al., 2018). These critical windows are discussed in detail below.

1.2.1. In utero critical window: Genesis, connectivity and fetal breathing movements in respiratory control networks

Essential components of the respiratory control network begin developing in utero. Neurogenesis of phrenic motoneurons peaks at E11, whereby they migrate to their final location in the ventral cervical spinal cord at E15 (Ren and Greer, 2003) and immediately begin innervating the diaphragm (main inspiratory muscle) via the neuromuscular junction (reviewed in Mantilla and Sieck, 2008). The preBötC begins developing in utero with peak neurogenesis occurring at E12.5 – E13.5 in rats, whereby these neurons migrate from the ventricular zone to their final location in the ventrolateral medulla (Pagliardini et al., 2003). Once preBötC neurons reach the ventrolateral medulla, they quickly establish key connections with the ventral respiratory group at E17, who innervate phrenic motoneurons (Pagliardini et al., 2003). Thus, critical components of the respiratory circuitry are intact at E17, when fetal breathing movements commence (reviewed in Greer, 2012; Greer and Martin-Caraballo, 2017). Fetal breathing movements facilitate circuit maturation by augmenting initially weak signals from the preBötC (Kobayashi et al., 2001), increasing membrane potential of phrenic motoneurons (Ren and Greer, 2003), and facilitating lung proliferation and differentiation (Hooper and Harding, 1990; Bissonnette et al., 1991; Liu et al., 1992; Kitterman, 1996; Liu and Post, 2000; Hooper and Wallace, 2006). Precise timing of each of these events is critical for the respiratory control network to maintain blood gases immediately at birth without assistance.

1.2.2. Newborn critical window: Birth and the transition from placental gas exchange to air breathing

In the first few days of life for newborn rodents and weeks for human preterm infants, breathing is unstable and chemosensitivity is blunted, contributing to increased apneas compared to age matched term infants (Sterni et al., 1999; Fairchild et al., 2016). Similar disruptions in breathing are seen in preterm mice (G17-G19), who have disrupted breathing patterns (low frequency and large tidal volume) that are strong predictors of mortality (Ramirez et al., 2019). Further, the combination of maternal inflammation and prematurity increase mortality compared to either individual insult (Ramirez et al., 2019), demonstrating inflammation negatively impacts preterm infants. Rodents have proven to be a useful model to study early life respiratory dysfunctions, such as apnea of prematurity, since the newborn rodent (P0–4) respiratory control network is comparable to the respiratory system at the third trimester in humans (Di Fiore et al., 2013). The early newborn period in rodents represents a critical window due to shifts in oscillator dominance and expression of neurotransmitter receptors, resulting in overall decreased excitation in the respiratory control network (Fig. 1). In utero and in newborn rodents (P0–4), inspiratory rhythm generation is initially maintained by both the pFRG and the preBötC, with the pFRG being the dominate oscillator (Onimaru and Dutschmann, 2012). During the first week of life, the preBötC shifts to being the dominate oscillator and the pFRG plays a larger role in active expiration and, possibly chemoreception (Barnes et al., 2007; Huckstepp et al., 2015). Many questions remain about the developmental switch in oscillator dominance, such as the age when this switch occurs, the mechanisms underlying the switch, and how early life insults, such as inflammation, impact this developmental process. Further, we know even less about development of post-inspiratory complex (PiCo) rhythm generating networks or its role during this, or other, critical windows. At the cellular level, the expression of glutamate and NMDA receptors are moderately decreased in P3–4 rats in regions important for respiratory control (preBötC, NTS, and Amb, XII, and DMV) compared to the first week of life, resulting in an overall decrease in excitability (Liu and Wong-Riley, 2002, 2005). Additionally, expression of GABA_A, GABA_B, and glycine receptors transiently plateaus in P3–4 rats, before continuing to increase throughout neonatal development, further contributing to overall decreased excitability (Liu et al., 2006). Receptors for other neuromodulators, such as thyrotropin-releasing hormone, serotonin (5-HT_1A, 5-HT_1B, and 5-HT_2A) are also low during the first few days of life in rodents (P0–4) before peaking at P5 to P11 (Liu and Wong-Riley, 2008, 2010a, 2010b). Collectively, these fluctuations in receptor expression decrease overall excitability of respiratory control regions, potentially contributing to the increased apnea frequency and increased variability in breathing frequency (Wong-Riley et al., 2013).

In addition to the developmental processes occurring in rhythm generating nuclei, changes in chemoreception occur early in life, contributing to increased apnea frequency in neonates (Prabhakar et al., 2007). The insensitivity of the carotid body to O₂ in newborns contributes to blunted hypoxia responses common in newborns (Sterni et al., 1999; Bairam et al., 2001; Prabhakar et al., 2007; MacFarlane et al., 2013) and limits the ability of newborns to compensate and stabilize breathing during a hypoxic challenge (Julien et al., 2008).

Thus, newborn developmental processes in respiratory rhythm generating nuclei and chemoreceptive regions contribute to increased frequency of apneas in preterm infants and newborns, potentially leaving the newborn respiratory control network more vulnerable to insults.

1.2.3. Neonatal critical window: Neurochemical shifts in neonatal respiratory control development

A third critical window in respiratory control occurs during the second week of life in rodents (P10–13), corresponding to the 2^nd and 4^th month in human infants (Ballanyi, 2004), and is characterized by transient shifts in glutamate receptor subunits altering receptor-ligand binding affinity (reviewed in Wong-Riley et al., 2013, 2019). A shift in NMDA receptor subunits from NR1 to NR2A within the preBötC, NTS, nucleus ambiguous (Amb), hypoglossal (XII) and DMV at P10–13 transiently decreases NMDA receptor affinity for glutamate (Liu and Wong-Riley, 2010b; Gao et al., 2011; Liu and Wong-Riley, 2012; Liu and Wong-Riley, 2013). Concurrently, abrupt increases in expression of glycine receptor subunit α1 and a change from α2 to α3 subunits in the preBötC, NTS, and XII increases receptor affinity for inhibitory neurotransmitters, thus increasing inhibitory signaling (Liu and Wong-Riley, 2004; Liu and Wong-Riley, 2013). Further, hyperpolarizing of neurons in respiratory control regions from P12–13 is caused by a developmental shift in expression of the Na-K-Cl co-transporter 1 (NKCC1) to the K-Cl co-transporter 2 (KCC2) (Liu and Wong-Riley, 2012). This shift in NKCC1-KCC2 increases intracellular chloride and increases spontaneous inhibitory postsynaptic potentials resulting in decreased membrane potential of hypoglossal neurons (Gao et al., 2011) and the pre-BötC (Ren, 2006). Thus, decreased excitation and increased inhibition occurs at P10–12 in key respiratory control regions (Fig. 1). Such shifts in neurotransmitter receptor expression are a cellular mechanism likely contributing to behavioral changes, such as blunted acute hypoxic responses at P12–13 (Liu et al., 2006). Insults, such as inflammation, have the potential to disrupt these cellular changes and alter ventilatory control.

Another neuromodulator undergoing changes at P10–13 is serotonin, which undergoes a transient decrease in signaling during this critical window of development. The importance of decreases in serotonin signaling can be recognized by its hypothesized role in Sudden Infant Death Syndrome (SIDS), where infants have reduced serotonin receptor expression in several brainstem nuclei, including the RTN and NTS, and display signs of hypoxia (Panigrahy et al., 2000; Ozawa and Okado, 2002; Paterson et al., 2006). In humans, the peak incidence of SIDS occurs between the 2^nd and 4^th months after birth (Hakeem et al., 2015). Transient decreases in serotonin receptor expression in rodents disrupts excitatory signaling in respiratory control regions at P10–13 and is positively correlated with decreased stability in breathing (Liu and Wong-Riley, 2010a), suggesting decreased serotonin signaling increases SIDS vulnerability. While the cause(s) of SIDS remain unknown, the triple-risk model is the leading hypothesis for the underlying etiology of SIDS (Kinney and Thach, 2009). In this model, the combination of intrinsic risk factors (such as abnormalities in serotonin signaling in the brainstem), a critical window of development, and extrinsic risk factors (such as infection, intermittent hypoxia, neonatal stress, sustain hypoxia or nicotine exposure) contribute to SIDS pathology (reviewed in Highet, 2008; Kinney et al., 2009; Goldberg et al., 2018). Further, others have hypothesized respiratory dysfunction associated with SIDS (Goldberg et al., 2018; Cummings and Hodges, 2019), highlighting the need to study changes in respiratory regions at P10–13. Other hypotheses include impaired microglia (immune cells of the brain) and disrupted neuroinflammatory signaling in SIDS etiology (Blackwell and Weir, 1999; Goldwater, 2004; Highet, 2008; Blood-Siegfried, 2009; MacFarlane et al., 2016), although a direct link between inflammation, microglia dysfunction and SIDS has yet to be established.

1.3. Neuroinflammatory signaling

Before discussing the effects of inflammation on developing respiratory circuits, it is important to review some of the mechanisms by which peripheral inflammatory signals influence the central nervous system (CNS) and the important role of inflammatory signals in neurodevelopment. While the CNS was once thought to be immune privileged (Medawar, 1948), we now appreciate the tremendous impact of peripheral immune signals on brain function (reviewed in Bilbo and Stevens, 2017). Antigen presenting cells (macrophages, dendritic cells and B cells) in the spleen and lymph nodes recognize foreign cells or substances, initiating the peripheral immune response (reviewed in Janeway and Medzhitov, 2002; Chaplin, 2010). Activated antigen presenting cells then activate immune cells, including helper T cells, B cells, macrophages, leukocytes and others, inducing cell proliferation and production of pro-inflammatory cytokines to propagate inflammatory signals throughout the body (reviewed in Chaplin, 2010). Peripheral immune system activation induces CNS inflammation through several distinct mechanisms (reviewed in D’Mello and Swain, 2016). Peripheral pro-inflammatory cytokines signal to the CNS by: active transport across the blood brain barrier (Banks, 2006), activating endothelial cells to release cytokines into the CNS (Balan et al., 2011), or activating vagus nerve afferents projecting to the NTS (Balan et al., 2012) (summarized in Fig. 2). Ultimately, stimulation of the peripheral immune system, activates glia within the CNS (Fig. 2). Glia, predominantly microglia and astrocytes, are the main mediators of inflammatory signals within the CNS (Bilbo and Schwarz, 2009). Microglia, the resident macrophages of the CNS (Graeber et al., 2011), mediate CNS inflammation by releasing cytokines to increase microglia motility and phagocytosis of synapses, and remove cellular debris and apoptotic cells (reviewed in Bilbo and Schwarz, 2012; Bilbo et al., 2012). These activated microglia, in turn, activate astrocytes to release cytokines and growth factors to either augment or reduce CNS inflammation (reviewed in Colombo and Farina, 2016). Ultimately, activation of microglia and astrocytes modulates neuronal activity to change behavior (Pascual et al., 2012), including respiratory behavior (reviewed in Funk et al., 2015 and Hocker et al. 2017). Changes in neuroinflammatory status are communicated back to the periphery through the efferent vagus nerve and the hypothalamic-pituitary-adrenal axis to ultimately resolve acute inflammation (reviewed in Chrousos, 1995; Bonaz et al., 2016) (Fig. 2). Dysregulation in the communication between the peripheral immune system and central nervous system results in chronic inflammation, which is associated with the etiology of many diseases and disorders (reviewed in Fogel et al., 2017). A common model to study the impact of peripheral inflammation on CNS function is lipopolysaccharide (LPS), a component of the cell wall of gram-negative bacteria activating toll-like receptor 4 (TLR4) to induce systemic inflammation. While LPS itself does not cross the blood-brain barrier, it acts as a peripheral TLR4 receptor agonist and activates this peripheral to central communication, making LPS a powerful model to study the detrimental impact of systemic inflammation on the respiratory control network (reviewed in Hocker et al., 2017).

Fig. 2. Bidirectional communication between the peripheral immune system and the central nervous system.

Lipopolysaccharide (LPS), a component of gram-negative bacteria, is a common experimental model activating the peripheral immune system and resulting in central nervous system inflammation. Activation of peripheral immune organs increases peripheral cytokine production. Peripheral inflammation is communicated to the central nervous system in at least three ways: (1) active transport of peripheral cytokines across the blood brain barrier, (2) peripheral cytokines activate vagal afferents, and (3) peripheral cytokines activate endothelial cells of the blood brain barrier to induce cytokine release in the central nervous system. Peripheral immune signals directly or indirectly activate microglia and astrocytes in the central nervous system. To propagate and resolve central nervous system inflammation, microglia and astrocytes increase cytokines, chemokines, cytotoxins, phagocytosis, and release neuromodulators, resulting in neuroinflammation spreading and/or the protection of neural circuits until neuroinflammation is resolved. Microglia and astrocytes signal to the hypothalamus and dorsal motor nucleus of the vagus to send feedback signals to the peripheral immune system, participating in the resolution of systemic inflammation. (Adapted from Schwarz and Bilbo, 2011).

While the inflammatory response is important during disease, injury or pathogen invasion, endogenous inflammatory signaling is required during development of the nervous system (reviewed in Bilbo and Schwarz, 2012). Microglia are derived from the yolk sac and invade neural tissue during gestational week 5.5 in humans (Lenz and Nelson, 2018) and at E8.5 in rodents (Lenz and Nelson, 2018). Microglia are guided to the central nervous system by immune signals, such as cytokines (e.g. IL-6, IL1-β, TNF, and IFN_Y proteins), chemokines, MHC cell surface proteins, and complement proteins (reviewed in Deverman and Patterson, 2009; Shigemoto-Mogami et al., 2014). Upon arrival in the CNS, microglia rapidly migrate and proliferate until their numbers peak between gestational weeks 20–26 in humans (Menassa and Gomez-Nicola, 2018) and the end of the second postnatal week in rodents (Alliot et al., 1999; Kierdorf et al., 2013). During development, microglia support synaptic patterning (Paolicelli et al., 2011; Hong et al., 2016; Sellgren et al., 2019) and myelination (Shigemoto-Mogami et al., 2014; Hagemeyer et al., 2017), promote neurogenesis (Cunningham et al., 2013; Ueno et al., 2013), and regulate apoptosis (Hornik et al., 2016). Between E12-E15, endogenous cytokines are critical for astrocyte differentiation from neural progenitor cells (Deverman and Patterson, 2009). Further during perinatal development, astrocytes participate in formation and maintenance of the blood brain barrier (Goldstein, 1988), support CNS angiogenesis (Risau, 1997), and facilitate axon guidance (Bixby and Harris, 1991; Dickson, 2002; Inatani, 2005). Astrocytes also participate in complement-dependent synaptogenesis (Kucukdereli et al., 2011) and neural circuit refinement through the expression of chemokines and IFN_Y proteins (Bialas and Stevens, 2012). Thus, endogenous inflammatory signaling is essential for normal CNS development of glia and disruptions in immune signaling during embryonic and early life have the potential to derail normal neurodevelopment. Only a few studies have focused on how glia (such as microglia and astrocyte) during critical windows of development contribute to respiratory control vulnerability (Kahn et al., 2002; MacFarlane et al., 2016; Tenorio-Lopes et al., 2017). Further, microglia and astrocytes are essential components of the neuroinflammatory response and understanding how microglia and astrocytes respond to inflammation in the developing brainstem is fundamental to unraveling the impact of inflammation on neonatal respiratory control.

2. Acute effects of inflammation in neonates

Diverse stimuli (such as recurrent apnea, neonatal stress, infection, and sustained hypoxia) induce perinatal inflammation and disrupt respiratory control during the first week of life (Genest et al., 2004, 2007; Herlenius, 2011; Bavis et al., 2019). In the following sections, we will describe the effects of inflammation on respiratory rhythm generation, and the effects of moderate and low levels of inflammation on chemoreception and respiratory motor plasticity.

2.1. In utero inflammation disrupts neonatal respiratory control

2.1.1. Rhythm generation

Maternal sleep disordered breathing (SDB) during pregnancy induces fetal neuroinflammation (Pamidi et al., 2014). While direct evidence linking maternal SDB and fetal hypoxia has not been established, growing evidence supports the hypothesis that maternal SDB causes adverse perinatal outcomes, such as preterm delivery, low birth weight, increased incidence of neonatal intensive care admission (Chen et al., 2012), intrauterine growth restriction and a low Apgar score (Louis et al., 2012) and increased mortality (Louis et al., 2014 and reviewed in Ding et al., 2014). It remains possible that the effects of maternal SDB on the infant are indirectly caused by maternal stress, increased hypothalamic-pituitary axis activity and increased fetal exposure to stress hormones (Johnson et al., 2018). To our knowledge, no studies have examined stress hormone levels in pregnant women with SDB. However, gestational stress induces lasting epigenetic changes in the offspring (Ślusarczyk et al., 2015) and alters the morphology and immune responses of adult microglia (Fournier et al., 2013), suggesting gestational stress causes lasting impairments in adult neuroimmune responses.

We are beginning to appreciate how inflammatory insults in utero sex-dependently impair neonatal respiratory rhythm (Gozal et al., 2003; Johnson et al., 2018) and alter responses to subsequent inflammatory challenges (Makinson et al., 2017). SDB during pregnancy is particularly common in obese mothers and is associated with newborn respiratory dysfunction (Bourjeily et al., 2013; Pamidi et al., 2014). Recent work using rodent models has shed new light on the mechanisms underlying newborn respiratory dysfunction by modeling SDB in rodents with gestational intermittent hypoxia (dams from G9–21, 90–120 seconds 10.5% O₂, 90–120 seconds 21% O₂, 8 hours/day) (Gozal et al., 2003;Johnson et al., 2018). Such work has revealed repetitive episodes of gestational intermittent hypoxia induces fetal inflammation (Gozal et al., 2003; Johnson et al., 2018) and disrupts respiratory circuitry in neonates (P0–3) in vitro (Johnson et al., 2018). In isolated respiratory circuitry (brainstem-spinal cord preparations, P0–3), gestational intermittent hypoxia decreased respiratory rhythm regularity, but did not otherwise impair fictive respiratory activity in males or females (Johnson et al., 2018). Yet despite no changes in activity, basal COX-2 gene expression was elevated in male spinal cords (P2.5–3.5), but not females, suggesting increased immune signaling in males after gestational intermittent hypoxia (Johnson et al., 2018). Following a subsequent immune challenge in neonates (P0–3, 0.1 mg/kg LPS) exposed to gestational intermittent hypoxia, acute hypoxia slowed respiratory rhythm in vitro in both sexes (Johnson et al., 2018). Yet, females (P2.5– 3.5) had blunted neuroimmune responses to a subsequent neonatal immune challenge following gestational intermittent hypoxia (Johnson et al., 2018). Thus, gestational intermittent hypoxia has sex-specific effects on respiratory activity and “re-programs” neuroinflammatory signaling responses in females, thereby blunting inflammatory responses to subsequent immune challenges. It is possible microglia contribute to the sex-specific differences observed following gestational intermittent hypoxia as there are considerable sex-dependent differences in microglia development and immune function during perinatal development (Crain et al., 2009, 2013; Crain and Watters, 2015; Hanamsagar and Bilbo, 2016; Hanamsagar et al., 2017). Although, it is unclear whether the blunting of neuroimmune responses in females is beneficial or detrimental for respiratory control, gestational intermittent hypoxia impairs respiratory activity and additional work is needed to determine whether changes in glia are involved.

Perinatal nicotine exposure also disrupts respiratory control development, impairs neonatal breathing (Tiesler and Heinrich, 2014) and is hypothesized to be a contributing risk factor for SIDS (Kinney and Thach, 2009). Additionally, infants born to mothers who smoke are prone to childhood obstructive apnea (Kahn et al., 1994), neonatal respiratory instability, increased frequency of neonatal apneas and lower overall breathing rate (Katz et al., 2012; Ferng and Fregosi, 2015). Perinatal nicotine exposure (by using a subcutaneous implant in pregnant dams at G5, 6 mg/kg/day of nicotine) disrupts inhibitory and excitatory signaling in neonatal respiratory neurons (Huang et al., 2004; Pilarski et al., 2011; Jaiswal et al., 2013; Cholanian et al., 2017; Wollman et al., 2018). This insult depresses overall medullary output (Huang et al., 2004) and involves augmented affinity of GABA receptors in respiratory rhythm generating nuclei (Luo et al., 2004; Luo et al., 2007). Interestingly, hypoglossal motoneurons increase excitability to compensate for the drop in medullary drive (Cholanian et al., 2017). At a cellular level, perinatal nicotine impairs dendritic arborization on hypoglossal motoneurons and desensitizes nicotinic acetylcholine receptors on postsynaptic hypoglossal motoneurons (Powell et al., 2016). Although both inflammation and perinatal nicotine independently increase apneas by disrupting the respiratory control network, it is unclear if perinatal nicotine increases neonatal inflammation and contributes to the pathology after perinatal nicotine or what role glia might play.

2.1.2. Chemoreception

In addition to disrupting respiratory rhythm, gestational intermittent hypoxia also disrupts neonatal chemoreception (Gozal et al., 2003). Gestational intermittent hypoxia decreases neonatal (P5) peak hypoxic ventilatory responses and increases hypoxic ventilatory depression compared to older animals (> P10) (Gozal et al., 2003). Since gestational intermittent hypoxia disrupts neonatal hypoxic responses and hypoxia is associated with SIDS etiology (Goldberg et al., 2018), maternal SDB may be a contributing risk factor for SIDS and warrants further investigation. Further, novel perspectives on SIDS implicate CO₂ chemoreception deficits as contributing to SIDS etiology (Goldberg et al., 2018; Cummings and Hodges, 2019), yet it is unclear if CO₂ chemoreception is disrupted following gestational intermittent hypoxia.

2.1.3. Respiratory motor plasticity

It is unclear whether gestational intermittent hypoxia impairs respiratory plasticity occurring in the carotid sinus nerve, medullary respiratory rhythm generating nuclei, or respiratory motoneurons. Gestational inflammation induced by maternal obesity impairs plasticity in other CNS regions (Bilbo and Tsang, 2010; Bolton and Bilbo, 2014; Edlow et al., 2019), demonstrating the potential for gestational inflammation to impair neonatal respiratory plasticity. Future studies are needed to better understand how gestational insults may alter plasticity.

2.2. Inflammation during the newborn critical window (P0–4 rodents) disrupts respiratory control

2.2.1. Rhythm generation

Our understanding of the effects of inflammation on rhythm generation stem largely from our understanding of the effects of sepsis, life-threatening organ dysfunction caused by a dysregulated inflammatory response to infection (Shankar-Hari et al., 2016). Severe inflammation gives rise to sepsis and is the major cause of morbidity and mortality in neonatal populations (Fleischmann-Struzek et al., 2018). Sepsis also increases the incidence of mixed apneas (an obstructive apnea followed by central apnea) in preterm infants (Fanaroff et al., 1998; Martin and Fanaroff, 1998; Olsson et al., 2003; Hofstetter et al., 2008 and reviewed in Herlenius, 2011). One proposed mechanism for these respiratory impairments was investigated using sepsis-level LPS (100 mg/kg, i.p.). LPS increased peripheral IL-1β and activated IL-1 receptors on vascular endothelial cells of the blood brain barrier, stimulating the release of PGE₂ in the NTS and the RVLM to disrupt central respiratory rhythm generating nuclei (Hofstetter and Herlenius, 2005; Hofstetter et al., 2007, 2008; Siljehav et al., 2014; Siljehav et al., 2015). Increased central PGE₂ activates prostaglandin E receptor type 3 on respiratory rhythm generating neurons in the rostral ventrolateral medulla (RVLM), depressing their activity and contributing to increased apneas (Siljehav et al., 2014). As such, increased PGE₂ in human cerebral spinal fluid are used as early biomarkers of infection in preterm and term infants (Siljehav et al., 2015).

While sepsis causes life threatening respiratory control disruption, lower levels of inflammation also impair newborn respiratory control networks. Even non-sepsis levels of neonatal inflammation (LPS, 100 μg/kg i.p.) in mice lowers breathing frequency during normoxia (Olsson et al., 2003; Hofstetter and Herlenius, 2005). We recently demonstrated neonatal respiratory control circuitry is sensitive to even low levels of systemic inflammation induced by LPS (1–10 mg/kg, i.p.) as early as 1 hour with peak impairment at 3 hours, coinciding with the peak inflammatory response (Morrison et al., 2019). Interestingly, respiratory control regions have different sensitivities to inflammation. Low to moderate systemic inflammation (1–5 mg/kg LPS) disrupts both amplitude and frequency in brainstem spinal cord preparations, while only more severe inflammation (10 mg/kg, LPS) impaired frequency in rhythmic slice preparations. Thus, low to moderate inflammation does not disrupt respiratory rhythm generating circuitry directly, but it impairs activity by influencing other modulatory medullary regions to indirectly alter respiratory control networks, such as the preBötC. The mechanism(s) underlying these impairments are unknown; however, we hypothesize a contribution of immune signaling from microglia and astrocytes. Nonetheless, since moderate and low-level inflammation caused the greatest impairments, it indicates even low-level infections have the potential to disrupt developing respiratory control circuitry and induce deficits in breathing.

2.2.2. Chemoreception

Neonatal inflammation in P0–4 rodents also impairs O₂ chemoreception by blunting the carotid body chemosensitivity (Olsson et al., 2003; Gauda et al., 2013), inducing spontaneous intermittent hypoxia events, and causing significant O₂ desaturation (Master et al., 2015, 2016). Intermittent hypoxia in human infants is associated with other non-respiratory pathologies, such as retinopathy of prematurity (Martin et al., 2011), abnormal neurodevelopment (Martin et al., 2011) and even death (Poets et al., 2015). Impairments in O₂ chemoreception are a result of disruptions to carotid body structure by decreasing the volume of type II cells (glia-like cells in carotid body) and blunted carotid sinus nerve activity in response to hypoxia in vitro (Master et al., 2015). Such changes result in increased spontaneous apneas in vivo (Master et al., 2016). Further, neonatal inflammation increases inflammatory cytokine expression (including IL-1β, IL-6, TNFα) acutely (48 hrs) and chronically (2 weeks) in the carotid body, resulting in chronic chemosensitivity impairment (Master et al., 2016). Additionally, neonatal inflammation increases dopamine levels in the carotid body (Master et al., 2016), an inhibitory modulator of the carotid body activity (Nurse, 2010). Responses to changes in CO₂ have yet to be assessed, demonstrating further studies are needed to assess the impact of neonatal inflammation on CO₂ chemosensitivity. While we are beginning to appreciate the peripheral changes, less is known regarding how inflammation during this critical window impacts central respiratory nuclei involved in chemoreception, such as the RTN, medullary raphe, locus coeruleus, and neurokinin 1 receptor expressing neurons in the ventral respiratory column (reviewed in Richerson et al., 2001; Nattie and Li, 2010; Teran et al., 2014).

In addition to neonatal inflammation, neonatal repetitive intermittent hypoxia (10% O₂, 9 episodes/h, 8 h/day, 10 days) impacts carotid body function (reviewed in Prabhakar and Semenza, 2016). Neonatal repetitive intermittent hypoxia does not alter baseline sensory activity from the carotid body (Pawar et al., 2008), but augments the response to hypoxia in the carotid body (Peng et al., 2004; Pawar et al., 2008) and induces hyperplasia of glomus cells (Pawar et al., 2009). Impairments of the carotid body by neonatal repetitive intermittent hypoxia endure into adulthood (Pawar et al., 2009) and involve epigenetic up regulation of reactive oxygen species (Prabhakar and Peng, 2017; Makarenko et al., 2019).

2.2.3. Respiratory plasticity

While we are beginning to understand the effects of neonatal inflammation on neonatal respiratory rhythm generation and oxygen chemosensitivity, many questions remain about the impact of inflammation on neonatal respiratory plasticity. Plasticity exists in neonates in the forms of ventilatory long-term facilitation evoked by acute intermittent hypoxia (Julien et al., 2008), frequency plasticity evoked by episodic anoxia in in vitro brainstem spinal cord preparations (Blitz and Ramirez, 2002), frequency plasticity evoked by intermittent application of substance P (Johnson et al., 2019), motor plasticity evoked by episodic serotonin (5-HT) in vitro in brainstem spinal cord preparations (Lovett-Barr et al., 2006), and motor plasticity evoked by episodic α-methyl-5-HT in in vitro rhythmic slices (Bocchiaro and Feldman, 2004). Although the impact of inflammation on neonatal respiratory motor plasticity has yet to be studied, more is known about how inflammation abolishes adult respiratory motor plasticity (reviewed in Huxtable et al., 2011; Hocker et al., 2017) and is discussed in detail below. One possibility is neonatal responses mirror that of adults, whereby neonatal inflammation abolishes neonatal motor plasticity. Loss of neonatal plasticity would limit adaptability and learning of the neonatal respiratory control network. However, it is also possible inflammatory signaling pathways in neonates are distinct from those in the adult and have different effects on neonatal respiratory plasticity. Further, inflammatory mechanisms in neonates and adults may be similar, but mechanisms to plasticity between the age groups might be distinct, resulting in different responses to inflammation. Thus, this area warrants additional investigation.

2.3. Inflammation during the neonatal critical window (P10–13 rodents) disrupts respiratory control

2.3.1. Rhythm generation

Inflammation induced by neonatal maternal separation disrupts neonatal rhythm generation at P10–13 in rodents (Gulemetova and Kinkead, 2011). Neonatal maternal separation, defined as 3 hours of maternal separation in neonatal rodents from P3–12, increases variability in neonatal respiratory frequency and increases apneas at P12, particularly in males (Gulemetova and Kinkead, 2011). At the cellular level, neonatal maternal separation increases the frequency and amplitude of spontaneous excitatory postsynaptic currents in the male DMV, but not in females (Baldy et al., 2017). Thus, neonatal maternal separation disrupts neonatal respiratory rhythm generation sex-dependently (reviewed in Rousseau et al., 2017). Many neonatal respiratory control disorders are more prevalent in males, yet we are only beginning to uncover the sex-differences in respiratory rhythm generation caused by neonatal insults. A continued focus on sex-dependent effects is vital to improve neonatal survival rates after inflammation and develop relevant sex-specific treatments.

While neonatal inflammation increases mortality in both human infants (Patrinos and Martin, 2017) and neonatal rodents (P10) (Rourke et al., 2016), the direct impact of neonatal inflammation on respiratory rhythm generation remains unclear. Little is known about the impact of inflammation on central respiratory rhythm generating nuclei in P10– 13 rodents. Known developmental changes, such as decreases in serotonin, decreases in serotonin receptors, and shifts in subunits of NMDA receptors and GABA receptors decrease excitatory signaling and increase inhibitory signaling in central respiratory control nuclei (Liu and Wong-Riley, 2013). These changes likely contribute to heightened mortality in response to inflammation in rodents at P10 (Rourke et al., 2016). In order to increase survival of neonates with inflammation, further research is needed to establish how systemic inflammation during this critical window disrupts respiratory rhythm generation.

2.3.2. Chemoreception

Low levels of inflammation (0.1 mg/kg LPS, i.p.) in P10 rodents disrupt O₂ chemosensitivity by attenuating phase I of the hypoxic ventilatory response, while the hypercapnic ventilatory response was unaffected (Rourke et al., 2016). Further, neonatal inflammation at P10 evokes an exaggerated increase in TNFα mRNA expression in the brainstem (Rourke et al., 2016), leading to the hypothesis that TNFα may be involved in impairments of the hypoxic ventilatory response following systemic inflammation. Further, some SIDS infants have abnormally high levels of peripheral TNFα, indicating heightened expression of inflammatory signaling is likely another risk factor for SIDS (Ferrante et al., 2008; Moscovis et al., 2015).

Inflammation induced by other inflammatory insults also impairs chemoreception. Sustained hypoxia as model of neonatal inflammation has provided unique insights into one mechanism by which systemic inflammation impairs chemosensitivity. Sustained hypoxia induces neuroinflammation and impairs both O₂ and CO₂ chemosensitivity (Mayer et al., 2013). This impairment is mediated, at least in part, by activating microglia and decreasing serotonin signaling in respiratory regions, such as the NTS and DMV (Mayer et al., 2014; MacFarlane et al., 2016). Given the parallels between critical windows in rodents and humans (Ballanyi, 2004), changes in rodent chemosensitivity might suggest new insights into the pathology of SIDS and are worthy of additional investigation.

While supplemental oxygen is a common treatment for infants with respiratory control impairments (Owen et al., 2019), both chronic sustained hyperoxia (60% O₂; P0–4) (Mu et al., 2018) and chronic intermittent hyperoxia (60% O₂; 4.5 minutes; 5 times per hour; P0–4) blunt the HVR in P15 rats (Logan et al., 2016), and reduce the size and number of unmyelinated nerve fibers in the carotid sinus nerve (Erickson et al., 1998). Chronic intermittent hyperoxia impairs the HCVR acutely after four or six days; however, it normalizes by P13–14 (Bavis et al., 2017). Further, chronic exposure to hyperoxia (60% O₂) during the first two weeks of life in rodents (P0–14) induces a lasting reduction in the adult HVR (Bavis et al., 2003), but the adult HCVR is not affected (Ling et al., 1996), suggesting neonatal hyperoxia selectively impairs oxygen chemoreception (reviewed in Bavis and Mitchell, 2008). Less is known, however, about the impact of chronic hyperoxia on central respiratory nuclei.

2.3.3. Respiratory motor plasticity

While neonatal inflammation increases apneas and disrupts chemoreception, the effects of neonatal inflammation on neonatal respiratory plasticity in P10–13 rodents remains largely unknown. Although the impact of inflammation induced by LPS or maternal neonatal separation on respiratory plasticity has not been investigated, chronic intermittent hypoxia, a model for childhood obstructive sleep apnea, demonstrates the detrimental impact of early life inflammatory insults on neonatal respiratory plasticity. Neonatal chronic intermittent hypoxia (10 cycles of 5% O₂ for 100 seconds/hour, 24 hours, 10 consecutive days) abolishes ventilatory LTF at P10 and increases the frequency of apneas at P10–12 with greater frequency in males (Julien et al. 2008), suggesting neonatal LTF is also sensitive to exogenous stressors. Since chronic intermittent hypoxia induces inflammation (Cai et al., 2011; Lam et al., 2012; Farré et al., 2018), it is possible ventilatory LTF is abolished by inflammation induced by chronic intermittent hypoxia, though this has yet to be directly tested. Thus, chronic intermittent hypoxia in neonates disrupts development of at least one form of respiratory plasticity, although, it is unclear if the effects of chronic intermittent hypoxia are due to brainstem inflammation or how other models of inflammation influence this or other forms of plasticity.

3. Acute, systemic inflammation impairs adult respiratory control

3.1. Rhythm generation

It is unknown whether inflammation in adults disrupts respiratory rhythm generation. Sepsis in adults induces respiratory failure and death (reviewed in Kim and Hong, 2016), suggesting central respiratory impairments may contribute to respiratory failure in sepsis-related deaths. Currently, it is unclear whether inflammation precedes central respiratory failure or is a consequence of it, limiting therapeutic developments for respiratory failure in adults with sepsis (Menezes et al., 2005; Calfee et al., 2015). Based on observations in the neonatal rat (Ballanyi et al., 1997; Solomon, 2002), we hypothesize sepsis-type inflammation in the adults would disrupt respiratory rhythm generation. There is much still to be learned here, such as: which mechanism is responsible for transmitting information about systemic inflammation to central respiratory nuclei, and how do microglia and astrocytes in respiratory control regions contribute to respiratory failure during sepsis. Future research is needed to address these unknowns to improve therapeutic interventions and survival rates for adults with sepsis-related respiratory dysfunction.

3.2. Chemoreception

In adults, moderate inflammation acutely disrupts chemoreception, limiting the ability to maintain adequate breathing (Pawar et al., 2008). Moderate systemic inflammation (3 mg/kg, LPS, i.p.) blunts short-term hypoxic phrenic responses in cats (Fernández et al., 2008), piglets (McDeigan et al., 2003), and adult rats (Vinit et al., 2011; Huxtable et al. 2013), although the extent of impairments in respiratory control regions are not fully understood. Repetitive episodes of intermittent hypoxia in adults (15 s of hypoxia, 5 min of normoxia, 9 episodes/h, 8 h/day) temporarily increases baseline sensory activity in the carotid body (Peng et al. 2003), yet does not alter carotid body structure (Peng et al. 2003). To date, no studies have investigated the impact of adult, systemic inflammation on hypercapnic ventilatory responses. Further, the contribution of microglia and astrocytes in respiratory control regions to inflammation-induced impairments in chemoreception have not been investigated.

3.3. Respiratory motor plasticity

While we know little concerning how neonatal inflammation disrupts neonatal respiratory plasticity, systemic inflammation abolishes adult respiratory motor plasticity (Vinit et al., 2011). Low levels of acute systemic inflammation (0.1 mg/kg LPS, i.p.) in adults disrupts Q-pathway-induced respiratory motor plasticity (Huxtable et al., 2011, 2013; Hocker and Huxtable, 2018). In addition to bacterial induced systemic inflammation, viral-induced systemic inflammation (polyinosinic:polycytidylic acid, 750 μg/kg, i.p.) also abolishes adult Q-pathway evoked respiratory motor plasticity (Hocker and Huxtable, 2019). Systemic administration of a general anti-inflammatory drug, ketoprofen, restores Q-pathway-evoked respiratory motor plasticity, demonstrating impairments in respiratory motor plasticity are mediated, at least in part, by peripheral inflammatory signaling (Huxtable et al., 2013; Hocker and Huxtable, 2019). Spinal IL-1 receptor antagonism also restores Q-pathway-evoked respiratory motor plasticity after acute LPS, demonstrating a role for CNS inflammation in inhibition of Q-pathway-evoked respiratory motor plasticity following systemic inflammation (Hocker and Huxtable, 2018). However, spinal recombinant rat IL-1β in healthy animals does not abolish Q-pathway-evoked respiratory motor plasticity, indicating IL-1 receptor activation alone is insufficient to inhibit Q-pathway-evoked respiratory motor plasticity and other inflammatory signaling mechanisms are likely involved (Hocker and Huxtable, 2018). Unlike Q-pathway-evoked respiratory motor plasticity, adult, systemic inflammation has no effect on S-pathway-evoked respiratory motor plasticity, demonstrating the S-pathway is inflammation resistant (Agosto-Marlin and Mitchell, 2017; Hocker and Huxtable, 2019). Thus, this pathway may represent a “back-up” pathway when Q-pathway-evoked respiratory motor plasticity is compromised (Agosto-Marlin and Mitchell, 2017; Hocker and Huxtable, 2019).

The therapeutic potential of inducing respiratory motor plasticity as a treatment for obstructive sleep apnea is promising (reviewed in Mateika and Komnenov, 2017), but likely limited by the presence of systemic inflammation in adults suffering from obstructive sleep apnea (Gerst et al., 2011). Chronic intermittent hypoxia protocols in rodents vary in severity and length (15 seconds – 5 minutes, 5–10% O₂ episode followed by normoxia for 8 hours a day, 3–2,400 episodes/day, 1–14 days), whereby pathology increases with increasing hypoxia episode number, duration, and severity (reviewed in Navarrete-Opazo and Mitchell, 2014). Low doses of intermittent hypoxia (5 min hypoxic episodes, 5 min intervals, 12 h/d, 7 days) enhance respiratory motor plasticity (Ling et al., 2001) and ventilatory long term facilitation (McGuire et al., 2003). In contrast, more severe doses of intermittent hypoxia induce pathology, likely due to increasing inflammatory signaling and/or decreasing expression of trophic factors known to promote motor plasticity (Baker-Herman et al., 2004; Dale-Nagle et al., 2011; Dale et al., 2012). Yet, just 8-hs of intermittent hypoxia (2 min, 10.5% O₂, 2 min normoxia) increases microglial IL-1β gene expression in the spinal cord and abolishes Q-pathway-evoked respiratory motor plasticity (Huxtable et al., 2015). Similar to LPS-induced inflammation, ketoprofen restores Q-pathway-evoked respiratory motor plasticity following 8-hs of intermittent hypoxia (Huxtable et al., 2015), demonstrating intermittent hypoxia impairs Q-pathway-evoked respiratory motor plasticity by inducing systemic inflammation. Additionally, inhibition of spinal p38 MAP kinase, a downstream enzyme mediating the initiation of and response to inflammation, restores Q-pathway-evoked respiratory motor plasticity following intermittent hypoxia, demonstrating the role of spinal inflammatory signaling in inhibition of Q-pathway-evoked respiratory motor plasticity by inflammation (Huxtable et al., 2015). While this research establishes inflammatory signaling in the spinal cord is essential to abolishment on Q-pathway-evoked respiratory motor plasticity, it is unclear if increased p38 MAP kinase activity is occurring within respiratory motoneurons, microglia, astrocytes or all cell types in the spinal cord. The exact molecular mechanisms underlying inflammatory impairment of respiratory motor plasticity pathways remains under investigation, but we hypothesize protein phosphatases are a key link between inflammation and inhibition of Q-pathway-evoked respiratory motor plasticity (Huxtable et al., 2018). Respiratory motor plasticity is hypothesized to be a novel therapeutic treatment for OSA, yet not all patients respond with increased motor output, possibly due to ongoing inhibition of Q-pathway-evoked respiratory motor plasticity by low-level inflammatory signaling in the spinal cord (Mateika and Komnenov, 2017). Additional research is needed to reveal more detailed mechanisms by which inflammation impairs adult respiratory motor plasticity to increase efficacy of therapeutic interventions for OSA.

4. Chronic impairments in adult respiratory control after neonatal inflammation

4.1. Rhythm generation

At present, there is no evidence neonatal inflammation (P4, LPS 1 mg/kg, i.p.) disrupts rhythm generation as no overt breathing deficits or deficits in phrenic nerve activity are observed in adult males or females following neonatal inflammation (Hocker et al., 2019). These results are similar to those in other regions in the central nervous system, such as in the hippocampus, where neonatal inflammation also does not cause any baseline disruptions in neural function (Bilbo et al., 2005). However, neonatal inflammation in combination with a subsequent adult inflammatory challenge blunts adult microglia and astrocyte immune responses in the hippocampus (Bilbo et al., 2005). The consequences of a subsequent adult inflammatory challenge on respiratory rhythm generation have yet to be determined. It is unclear if the respiratory control network is similarly “primed” by neonatal inflammation, resulting in enhanced adult dysfunction following a sub-threshold dose of inflammation. Since humans rarely experience only one inflammatory insult during their lives, understanding if neonatal inflammation primes central respiratory activity to a subsequent adult inflammatory challenge will likely provide novel insights into vulnerability to develop adult ventilatory control disorders.

4.2. Chemoreception

While neonatal inflammation does not impair adult respiratory rhythm generation, neonatal inflammation has stimulus-specific effects on adult chemoreception. Neonatal inflammation induced by neonatal maternal separation causes lasting increases in adult oxygen chemosensitivity (Genest et al., 2004; Kinkead et al., 2005). Neonatal maternal separation augments adult phrenic responses to hypoxia (Genest et al., 2004) and augments carotid sinus nerve activity in adult, anesthetized rats (Kinkead et al., 2005), suggesting neonatal maternal separation alters oxygen sensing and programs hyper-responsive adult hypoxic ventilatory responses. However, this only occurs in awake male rats (Soliz et al., 2016), where males also have blunted hypercapnic responses following neonatal maternal separation (Dumont and Kinkead, 2010). On the contrary, hypercapnic ventilatory responses are augmented in females, following neonatal maternal separation (Dumont and Kinkead, 2011). In contrast, neonatal inflammation induced by LPS, (P4, 1 mg/kg, i.p.) does not impair hypoxic or hypercapnic ventilatory responses in anesthetized, mechanically ventilated adult male and female rats (Hocker et al., 2019). Additional research is needed to understand the mechanisms underlying the stimulus-specific effects of inflammation to better understand the implications of early life inflammation on adult chemosensory responses.

4.3. Respiratory motor plasticity

Our recent work was the first to demonstrate one bout of neonatal inflammation (P4, 1 mg/kg LPS, i.p.) abolishes adult respiratory motor plasticity (Hocker et al., 2019). Surprisingly, both Q-pathway- and S-pathway-evoked respiratory motor plasticity are abolished in adults following neonatal inflammation (Hocker et al., 2019), which is distinct from the effects of acute, adult inflammation on plasticity (Huxtable et al., 2011, 2013). In adults after neonatal inflammation, Q-pathway-evoked plasticity is restored with an anti-inflammatory treatment, suggesting a lasting inflammatory signal undermines Q-pathway-evoked respiratory motor plasticity. In contrast, S-pathway-evoked respiratory motor plasticity is not restored by an adult anti-inflammatory treatment after neonatal inflammation. Since microglia are critical in the inflammatory response (reviewed in Shabab et al., 2017) and are undergoing important developmental changes at the time of the inflammatory stimulus (reviewed in Nayak et al., 2014), we hypothesize microglia play a role in impairment of Q-pathway-evoked plasticity. In contrast, S-pathway-evoked respiratory motor plasticity is not restored by an adult anti-inflammatory treatment after neonatal inflammation. However, S-pathway-evoked respiratory motor plasticity is evoked by spinal activation of adenosine 2A receptors in adults after neonatal inflammation, suggesting the mechanism of impairment after neonatal inflammation is inflammation-independent and is upstream from the phrenic motoneurons, where this form of plasticity is believed to occur (Johnson and Mitchell, 2002 reviewed in Fuller et al., 2001; Devinney et al., 2013; Gonzalez-Rothi et al., 2015). Since a major source of adenosine during hypoxia is astrocytes (Gourine et al., 2010), we further hypothesize that astrocytes play a role in impairment of S-pathway-evoked plasticity. Thus, inflammation early in life has lasting detrimental effects on respiratory motor plasticity in adulthood, though we are still investigating what other respiratory control deficits exist following neonatal inflammation. Such chronic impairments in respiratory motor plasticity limit lasting adaptability of the respiratory system to ever-changing internal and external environments. Eliciting respiratory motor plasticity is now being used to treat human spinal cord injury (Trumbower et al., 2012; Hayes et al., 2014; Gonzalez-Rothi et al., 2015) and sleep apnea patients (Mateika et al., 2015; Mateika and Komnenov, 2017). The success of these translational applications is potentially limited by acute or chronic inflammation impairing motor plasticity following neonatal inflammation, demonstrating the need to better understand the interactions between inflammation and plasticity. Future research into the mechanisms underlying how neonatal inflammation results in inhibition of adult respiratory motor plasticity will provide novel treatments for adults with ventilatory control disorders.

5. Conclusion

In conclusion, inflammation during critical windows of perinatal development impact respiratory rhythm generation, chemoreception and plasticity (summarized in Fig. 3). Further, inflammation likely contributes to increased risk of morbidity and mortality, and heightens the risk of respiratory control disorders. Since our appreciation of the interactions between the peripheral immune system and central nervous system are growing, we need to further investigate the mechanisms by which inflammation is impairing respiratory circuitry. A critical understudied area is the role of neuroimmune cells, such as microglia and astrocytes, in respiratory control deficits. Additionally, as many neonatal respiratory control disorders, such as apnea of prematurity and SIDS, are more prevalent in males, understanding sex differences in the response of respiratory control networks to neonatal inflammation is an important next step to developing preventative therapeutics for vulnerable infants. Lastly, inflammation during these early critical windows of development result in lasting impairments to adult respiratory control networks. More research is needed to understand the lasting deficits in adult respiratory control circuitry after early life stressors to develop more effective therapeutic interventions for adult ventilatory control disorders, such as sleep apnea.

Fig. 3. Inflammation disrupts respiratory rhythm generation, chemoreception, and plasticity acutely during critical windows and causes lasting impairments in adult respiratory control.

Neonatal inflammation disrupts respiratory rhythm generation acutely in neonates, but does not result in lasting impairments in adult rhythm generation. Neonatal inflammation acutely attenuates the hypoxic ventilatory responses (HVR) in neonates. In contrast, inflammation in adults acutely increases carotid sinus nerve activity. The chronic consequences of neonatal inflammation include augmented HVR in males, blunted hypercapnic ventilatory responses (HCVR) in males, and augmented HCVRs in females. The effects of gestational inflammation or newborn inflammation on respiratory plasticity are unknown. Neonatal inflammation (P10–13) impairs ventilatory long term facilitation (LTF) in males and females. In adults, acute systemic inflammation acutely impairs Q-pathway-evoked respiratory motor plasticity, but chronic intermittent hypoxia increases respiratory motor plasticity. Neonatal inflammation impairs both the Q- and S-pathway-evoked respiratory motor plasticity in males and females. (LPS = lipopolysaccharide, NMS = neonatal maternal separation, CIH = chronic intermittent hypoxia, SH = sustained hypoxia.)