The development of pain circuits and unique effects of neonatal injury

Abstract

Pain is a necessary sensation that prevents further tissue damage, but can be debilitating and detrimental in daily life under chronic conditions. Neuronal activity strongly regulates the maturation of the somatosensory system, and aberrant sensory input caused by injury or inflammation during critical periods of early postnatal development can have prolonged, detrimental effects on pain processing. This review will outline the maturation of neuronal circuits responsible for the transmission of nociceptive signals and the generation of pain sensation—involving peripheral sensory neurons, the spinal cord dorsal horn, and brain—in addition to the influences of the neuroimmune system on somatosensation. This summary will also highlight the unique effects of neonatal tissue injury on the maturation of these systems and subsequent consequences for adult somatosensation. Ultimately, this review emphasizes the need to account for age as an independent variable in basic and clinical pain research, and importantly, to consider the distinct qualities of the pediatric population when designing novel strategies for pain management.

1). Introduction

Pain is a natural response to protect the body from tissue damage; however, it is not always advantageous—many times hindering activities in daily life and causing serious emotional disturbances. Recurrent non-cancer pain currently affects nearly half of adults in the United States, and unfortunately this incidence has increased by at least 25% in the past 15 years (Nahin et al. 2019). In the United States alone, pain-related healthcare costs and loss of work productivity added up to an astounding $600 billion per year based on recent estimates (Gaskin and Richard 2012).

Pain is a complex sensory manifestation involving both neural circuits and non-neural mechanisms in the periphery, spinal cord, and brain. In short, sensory neurons detect noxious stimuli across the body and these nociceptive signals are conducted via the spinal cord to the brain where pain is perceived (Peirs and Seal 2016; Todd 2010). Neuronal signals that ultimately generate pain are processed and modulated throughout the entire journey through the nervous system via complex networks of neurons, glia and immune cells (Chen et al. 2018; Zouikr and Karshikoff 2017).

While sensory networks in the visual and auditory systems clearly require modality-specific input for their functional maturation, nociceptive circuits do not appear to require noxious sensory experience in order to develop normally. Rather the development of nociceptive circuits occurs via a cross-modality mechanism which requires spontaneous tactile input during a critical period of early life (Granmo et al. 2008; Waldenstrom et al. 2003). It is logical to assume that such a mechanism is evolutionarily advantageous, since tissue damage is harmful and often irreversible. This could also explain why intrinsically driven spontaneous activity is prevalent in developing nociceptive circuits. Spontaneous activity is a hallmark feature of developing sensory circuits, including nociceptive networks, which may make them particularly susceptible to permanent reorganization in response to aberrant patterns of afferent input. Activity-dependent strengthening and pruning is evoked and shaped by external stimuli—which allows our nervous system to adapt to our environment (Zouikr and Karshikoff 2017). Unsurprisingly, a dramatic amount of nervous system development occurs after birth, and the nervous system can be shaped in maladaptive ways by exposure to harmful stimuli, such as those accompanying tissue or nerve injury.

Unfortunately, injuries during the neonatal period are common. Birth, for example, is a traumatic experience and nerve damage can often occur during delivery (Barr et al. 2011; Malessy and Pondaag 2011). Neonates who are delivered preterm, or with complications, are admitted to neonatal intensive care units where they are routinely subjected to medically-necessary, yet invasive, procedures (Simons et al. 2003). It is well established that neonatal injury causes myriad sensory and emotional disturbances in humans, which will be discussed in depth in this review. A large and invaluable body of work in animal models has also expanded our understanding of somatosensory circuit development and the effects of neonatal injury.

Birth at full-term in humans has been considered roughly equivalent to the start of the second postnatal week in rodents (Clancy et al. 2001; Pinto et al. 2015), and somatosensory pathway maturation in rodents gradually occurs throughout the early postnatal period (Chang et al. 2016; Fitzgerald 2005). The sensory and behavioral disturbances caused by neonatal injury in humans are qualitatively similar in rodents (Hohmeister et al. 2010; LaPrairie and Murphy 2007; LaPrairie and Murphy 2010; Moriarty et al. 2019; Oberlander et al. 2000; Walker et al. 2018; Walker et al. 2015; Walker et al. 2009). Importantly, many of the changes associated with neonatal injury are preventable with the administration of local anesthetics at the time of injury, again implicating their dependence on aberrant neuronal activity. Blocking nerve activity can effectively prevent the acute elevation in spinal glutamate release (Li et al. 2009a), long-term alterations in descending inhibition of the spinal cord (Walker et al. 2015), and exacerbated pain after adult reinjury (Moriarty et al. 2018), observed after neonatal hindpaw injury. However, aberrant nerve activity is certainly not the sole driver of all repercussions of neonatal injury, as skin hyperinnervation is not prevented by nerve block (De Lima et al. 1999).

This review will give a broad overview of the key aspects of somatosensory circuit maturation, while highlighting the short- and long-term effects of injury sustained during critical periods of early life on the function of neuronal and immune signaling in the developing CNS.

2). Nociceptive circuits in the periphery and spinal cord dorsal horn

Somatosensation is initiated via specific peripheral end organs that detect particular stimulus modalities. These end organs or free nerve endings transform mechanical, thermal, or chemical stimuli into electrical signals in pseudo-unipolar dorsal root ganglion (or trigeminal ganglion) neurons. Research has classically identified four main classes of sensory neurons based on their electrical threshold and speed of signal conduction—although many more subtypes have now been identified based on their complex sensory functions and gene expression profiles (Ray et al. 2018; Teichert et al. 2012; Usoskin et al. 2015). Generally speaking, highly myelinated Aα and Aβ fibers are low-threshold and serve to transmit proprioception and touch, while thinly myelinated Aδ and unmyelinated C fibers are primarily high-threshold and have been predominantly associated with transmitting noxious stimuli (Abraira and Ginty 2013; Lawson 2002). These different classes of sensory neurons not only transmit distinct sensory modalities, but also undergo unique developmental trajectories.

a). Skin and sensory neurons

Skin innervation occurs embryonically. GAP-43, a general marker of proliferating nerve axons and their distal growth cones, is first detectable in the skin of rodent limbs around embryonic day (E) 12 and finally culminates in the phalanges at E19 (Reynolds et al. 1991). A and C fibers innervate the skin around E13–14 (Jackman and Fitzgerald 2000). Nerves terminating in muscle mature slightly later, beginning the innervation process at E15 and establishing clusters of nerve terminals in the muscle mass from E17 to E21 (Reynolds et al. 1991).

Cutaneous receptive fields are larger and less focused in early life (Andrews and Fitzgerald 1994; Holmberg and Schouenborg 1996). Receptive field areas of the skin are large at birth and gradually shrink over the course of development—by at least 75% in the rat (Fitzgerald 1985). These enlarged receptive fields contribute to the disorganized and unrefined nature of the somatosensory system in early life—and likely contribute to infants being hypersensitive to tactile stimulation (Fitzgerald 2005).

Injury during early life has exacerbated effects in the skin when compared to similar adult injuries. Injury-induced skin hyperinnervation is particularly pronounced if the injury occurs during the neonatal period in rodents (Reynolds and Fitzgerald 1995). Importantly, this neonatal injury-evoked skin hyperinnervation is not prevented by simultaneous nerve block (De Lima et al. 1999)—suggesting an alternate underlying mechanism involving neurotrophic factors and immune signaling (Beggs et al. 2012a).

b). Spinal dorsal horn (SCDH)

Peripheral sensory neurons innervate the SCDH, and somatosensory signals are processed and integrated in this region before being transferred to the brain. Projection neurons in lamina I of the SCDH, the dorsal-most portion of the spinal cord gray matter, are principally responsible for transmitting these sensory signals to the brain. This ascending nociceptive transmission is strongly modulated by an incredibly complex network of locally-arborizing, excitatory and inhibitory interneurons within the dorsal horn (Todd 2010).

i). Primary afferent innervation of the developing SCDH

A fibers innervate the SCDH around E15–17 and initially synapse in a widespread pattern in laminae I–IV (Figure 1, top), before gradually withdrawing to laminae III–IV during the first three weeks of life as shown in Figure 1, bottom (Fitzgerald et al. 1994; Nakatsuka et al. 2000; Park et al. 1999). Importantly, this process is activity-dependent and withdrawal is blocked with spinal NMDA receptor antagonism (Beggs et al. 2002). However, likely due to their mismatched dorsal horn innervation, the convergence of A and C fibers onto dorsal horn cells does not appear to occur until the second postnatal week (Fitzgerald 1985).

Fig. 1. Innervation patterns of distinct sensory neuron classes during postnatal development.

Illustration of the stages of primary afferent innervation to the SCDH. A fibers innervate the gray matter first. The A fibers initially overshoot their final synaptic destinations and innervate laminae I-IV (top), and then gradually withdraw their axons to laminae III-IV during the first three postnatal weeks in rats—depicted on the bottom at P25 (Fitzgerald et al. 1994; Nakatsuka et al. 2000; Park et al. 1999). C fiber innervation begins after A fibers, but C fibers synapse in their appropriate adult destinations in lamina I and II from the beginning as shown on the bottom at P25 (Benn et al. 2001).

C fibers begin to innervate lamina I and II of the SCDH around E18–20, which is later than the initial contacts into the skin (Jackman and Fitzgerald 2000). Unlike A fibers, C fiber synapses initially form in their respective adult destinations in lamina I-II (Figure 1, bottom), and C fibers responses can be detected from birth in the superficial dorsal horn of rodents (Fitzgerald 1985). However, some C fiber synapses form much later, with the SCDH connections of some non-peptidergic afferents not fully maturing until postnatal day (P)5–P7 (Benn et al. 2001). In addition, patch clamp recordings reveal that capsaicin application increases the frequency of miniature excitatory postsynaptic currents (mEPSCs) in SCDH neurons from P0 (Baccei et al. 2003). Although there is a considerable increase in the capsaicin response between P5 and 10, these results support the notion that C fiber synapses are present in the SCDH at birth (Baccei et al. 2003).

The timing differences in A and C fiber innervation into the dorsal horn could have many implications for nociception at a young age. For example, the function of particular neuronal populations in the SCDH could differ in early life if cells receive input from different primary afferent neurons transmitting distinct sensory modalities. Nonetheless, sensory neuron synapses are not the only factor contributing to differences in the functional organization of the SCDH during early life—as several properties of dorsal horn networks undergo significant postnatal maturation.

ii). Age-related changes in the firing of SCDH neurons

Neuronal activity is a key determinant of plasticity—and hyperexcitable or intrinsically driven cells in the SCDH may provide a means for nociceptive circuits to mature properly in the absence of injury. In rat SCDH neurons, initial action potential firing caused by pinching of the skin is followed by prolonged periods of low-frequency firing (termed afterdischarge), which both decrease sharply after P8 (Fitzgerald 1985).

SCDH cells also demonstrate a progressive increase in background activity after repetitive high-threshold stimulation at a moderate frequency (≥0.5 Hz). This phenomenon is known as “wind-up” and is thought to be a C fiber- and NMDA receptor-dependent process that can contribute to hyperalgesia via dorsal horn neuronal sensitization (Dickenson and Sullivan 1987; Mendell and Wall 1965; Xu et al. 1992). Wind-up in spinal reflexes and dorsal horn cells decreases significantly after the first postnatal week in rats (Fitzgerald 1985; Fitzgerald and Gibson 1984). These changes could be linked to the higher concentration of NMDA receptors in the neonatal SCDH (Gonzalez et al. 1993), or the increased NMDA-mediated Ca²⁺ influx, observed during early life (Hori and Kanda 1994).

This evoked hyperexcitability is coupled with intrinsic spontaneous activity in the developing SCDH. During early life, the frequency of spontaneous excitatory postsynaptic currents (sEPSCs) can be significantly reduced by the voltage-gated Na⁺ channel antagonist tetrodotoxin (TTX) (Baccei et al. 2003), while TTX does not appear to significantly lower sEPSC frequency in the adult dorsal horn (Bao et al. 1998; Yang et al. 1998). This observation points to an elevated level of spontaneous action potential discharge in the immature SCDH compared to adulthood. Furthermore, “pacemaker” neurons, which demonstrate spontaneous, intrinsic burst-firing activity and may play a role in synchronizing neuronal network activity (Chevalier et al. 2016), exist in lamina I of the newborn SCDH, but their prevalence decreases significantly throughout the first three weeks of life in the rat (Li and Baccei 2011b).

iii). Inhibitory tone develops slowly in the SCDH

Inhibitory receptive fields in the SCDH are diffuse and less refined in early life. Excitatory and inhibitory receptive fields are significantly less “aligned” in neonates compared with adults—indicating that the critical convergence and balance between excitatory and inhibitory inputs in the SCDH develops postnatally (Bremner and Fitzgerald 2008). The prevalence of GABAergic neurons in the spinal cord increases steadily during the first week after birth in the rat, and then decreases to reach adult levels (Schaffner et al. 1993). Inhibitory synaptic signaling is weaker at early ages, as the frequency of both spontaneous and miniature inhibitory postsynaptic currents mediated by GABA or glycine increase drastically in the SCDH during the first two weeks of life (Baccei and Fitzgerald 2004).

The strength of local inhibitory synapses increases with age, and the primary inhibitory neurotransmitter, GABA, can evoke either depolarization or hyperpolarization of dorsal horn neurons during the neonatal period (Baccei and Fitzgerald 2004; Ingram et al. 2008). Cl⁻ homeostasis within SCDH neurons is altered early in life due to an imbalance in the expression of the KCC2 and NKCC1 K⁺/Cl⁻ cotransporters (Hubner et al. 2001; Rivera et al. 1999). This imbalance causes a high intracellular Cl⁻ concentration, leading to a more positive Cl⁻ reversal potential and causing depolarizing GABA responses in immature SCDH neurons. Despite the more positive reversal potential of GABA_A receptors during early life, the depolarization does not appear to reach action potential threshold (Baccei and Fitzgerald 2004), thus predicting that GABA_A receptor activation reduces the likelihood of action potential firing in neonatal SCDH neurons as seen in the adult.

However, it is important to note that although the ongoing developmental changes in spinal GABAergic transmission might predict weak inhibition during early life, it appears that GABA_A receptor activation can effectively dampen neuronal activity in the dorsal horn of both neonatal and preadolescent rats. Spinal administration of the GABA_A receptor antagonist gabazine enhanced the firing of dorsal horn neurons in response to cutaneous stimulation, and enlarged the receptive fields in the skin, to a similar degree at P3 and P21—indicating that fast ionic GABAergic inhibition is largely intact in the SCDH during early life (Bremner et al. 2006). Notably, excitatory synaptic transmission is also strengthening during the early postnatal period. For example, the frequency of spontaneous and capsaicin-evoked miniature excitatory postsynaptic potentials increase markedly throughout the first week of life (Baccei et al. 2003). The concurrent increases in GABAergic and glutamatergic signaling could potentially explain why the balance between excitation and inhibition is virtually preserved in these nascent dorsal horn circuits.

iv). Neonatal injury “primes” the SCDH

Neonatal injury has both acute and persistent effects on the SCDH that are unique to a critical period in somatosensory development. A few days after neonatal hindpaw incision or inflammation, mEPSC frequency increases in SCDH neurons, which suggests an increase in the spontaneous release of glutamate within the dorsal horn (Li and Baccei 2009; Li et al. 2009a). Notably, this increased excitatory signaling after incision may come from high-threshold primary afferents that are sensitive to capsaicin (Li and Baccei 2011a), which likely represent nociceptors (Todd 2010). The enhanced glutamatergic transmission after neonatal incision occurs onto both excitatory and inhibitory interneurons, and this effect is dependent on nerve growth factor (NGF) signaling through trkA receptors (Li and Baccei 2011a). Importantly, injury or inflammation after the first week of life in rodents do not affect these same measures of spontaneous excitation, uncovering a distinct critical period when somatosensory circuits in the SCDH are highly susceptible to tissue damage (Li and Baccei 2009; Li et al. 2009a).

This enhanced excitation in the SCDH originating from primary afferents in the aftermath of neonatal injury appears to persist into adulthood and is coupled with a reduction in inhibitory transmission (Li et al. 2015). Decreased intrinsic excitability of GABAergic interneurons could partially explain this loss of inhibition (Li and Baccei 2014). This altered balance of synaptic excitation vs. inhibition is observed in lamina I projection neurons, which could have direct effects on pain perception since these cells transmit nociceptive signals to the brain.

Another persistent change which occurs after neonatal injury is an enhanced propensity for synaptic plasticity in nociceptive SCDH circuits. Long-term potentiation (LTP), a form of activity-dependent synaptic strengthening, occurs at SCDH synapses after noxious sensory input, including that resulting from peripheral nerve injury (Sandkühler and Liu 1998). Furthermore, LTP protocols involving high frequency electrical stimulation of skin and muscle can amplify peripheral transmission of a nociceptive input and cause increased pain sensitivity (Schilder et al. 2016). In adulthood after prior neonatal injury, the synapses from sensory neurons onto lamina I projection neurons in the SCDH show greater potentiation after high frequency stimulation (Li and Baccei 2016). This amplified potentiation in SCDH nociceptive circuits could increase the incidence or severity of pain in situations where sensory neurons are highly active (e.g. after subsequent injury).

In the hippocampus, spike-timing-dependent plasticity (STDP; a form of classical Hebbian learning) can induce synapse strengthening when a presynaptic input reaches a postsynaptic neuron ~20 ms before it fires, but when the postsynaptic neuron fires before the presynaptic input arrives, the synapse either weakens or remains unchanged (Levy and Steward 1983). Sensory synapses onto projection neurons also require these strict temporal requirements for potentiation under normal conditions, but after neonatal injury the timing requirements become very relaxed—making these synapses more susceptible to strengthening. In adulthood after neonatal injury, synapses from primary afferents to lamina I projection neurons become potentiated with pairing intervals greater than 20 ms, and even if the pairing order is reversed such that the postsynaptic cell (i.e. projection neuron) is stimulated before the presynaptic cell (i.e. primary afferent neuron). These results indicate that neonatal injury persistently increases the susceptibility for activity-dependent strengthening at this critical nociceptive synapse (Li and Baccei 2016).

Neonatal injury alters a variety of ion channels and neurotransmitter receptors in the SCDH, which may underlie many of the observations described above and emergent behavioral effects described later in this review. Calcium-permeable AMPA receptors (CA-AMPARs) allow for synaptic strengthening and play a role in spinal nociceptive circuit plasticity after inflammatory injury in adulthood (Hartmann et al. 2004). Neonatal hindpaw incision or inflammation in the first two weeks of life acutely increases the relative expression of CA-AMPARs in unidentified rat SCDH neurons. Peripheral inflammation at P17 also induces an increase in CA-AMPARs in the SCDH (Li and Baccei 2009), but incision at this age does not (Li et al. 2009a). Furthermore, neonatal incision elevates CA-AMPAR expression in adult projection neurons and unmasks a novel role for these receptors in spike timing-dependent LTP (Li and Baccei 2016).

Ion channels play a paramount role in neurons—having significant effects on the cell resting potential, membrane resistance, and action potential firing. Under physiological conditions, inward-rectifying K⁺ (K_ir) currents lower the membrane resistance and intrinsic excitability of SCDH neurons (Coetzee et al. 1999; Li and Baccei 2014; Li et al. 2013). Neonatal hindpaw incision persistently increases K_ir function in GABAergic, but not glutamatergic, interneurons in lamina II—this reduction in GABAergic cell excitability strongly suggests reduced inhibition of adult nociceptive SCDH networks after neonatal incision (Li and Baccei 2014). Neonatal injury also persistently alters inhibitory G protein-coupled receptor (GiPCR) signaling in the SCDH—potentially via a principal downstream effector of many GiPCRs—the G protein-coupled inward rectifying K⁺ (GIRK) channel. Projection neurons, but not GABAergic interneurons, show increased functional expression of GIRK channels in adulthood after prior neonatal incision (Brewer and Baccei 2018)—which could increase responses to GiPCR ligands such as opioids. These changes affecting projection neurons could have direct effects on nociceptive signals transmitted to the brain, but there are also myriad developmental changes and distinct responses to neonatal injury that occur in supraspinal neuronal circuits.

3). Nociception transmission in the developing brain and pain perception

a). Lower brain centers

The hindbrain and midbrain not only transmit nociceptive signals from the spinal cord to the brain, but also serve to either inhibit or facilitate these nociceptive signals directly in the spinal cord. Descending modulation from lower brain centers onto the spinal cord involves many neurotransmitters and neuromodulators such as GABA, opioids, serotonin, and norepinephrine (Todd 2010).

i). Maturation of descending modulatory pathways to the spinal cord

The periaqueductal gray (PAG) indirectly produces descending inhibition onto the spinal cord via opioid release and modulation of the rostroventral medulla (RVM). However, electrical stimulation of the PAG does not produce analgesia until P21 in the rat (van Praag and Frenk 1991). Applying opioid agonists directly to the adult PAG reduces spinal cord excitability—potentially due to decreased firing of μ opioid receptor-expressing GABAergic neurons, which in turn leads to disinhibition of neurons in the RVM and stronger descending inhibition of spinal cord circuits (Fields 2004; Morgan et al. 1992). However, applying opioid agonists to the PAG at P21 in the rat has the reverse effect—increasing the excitability of spinal cord circuits and lowering mechanical thresholds. Furthermore, opioid agonist application in the PAG at P10 in rats has little to no effect on spinal excitability or mechanical withdrawal thresholds, which could be due to immature opioid signaling since the expression of pro-opioidmelanocortin in this region is lower during early life. In support of these results, opioid antagonists administered into the PAG are pro-nociceptive in adult rats, yet analgesic at P21, and have little to no effect at P10 (Kwok et al. 2014).

Descending modulation from the RVM plays a critical role in nociception (Antal et al. 1996). In adulthood, the RVM facilitates nociception at low stimulus intensities but suppresses nociceptive transmission at high stimulation intensities (Figure 2, middle), which likely allows for discrimination between innocuous and noxious somatosensory input (Hathway et al. 2009; Walker et al. 2015). However, during the first three weeks of life in the rat, the RVM is predominantly pro-nociceptive (Hathway et al. 2009). The developmental switch from facilitation in preadolescent rats to bimodal inhibition and facilitation in adults is dependent on μ-opioid receptor signaling in the RVM, as μ-opioid agonists accelerate this transition, while antagonists prevent this normal developmental shift (Hathway et al. 2012).

Fig. 2. RVM modulation of nociception develops postnatally and is shaped by neonatal injury.

The RVM area of the brainstem is a source of descending modulation onto nociceptive circuits in the spinal cord. The RVM is primarily pro-nociceptive during early development (left), both inhibits and facilitates nociception in adulthood (middle), but only inhibits nociception in adulthood after neonatal injury as shown on the right (Hathway et al. 2009; Walker et al. 2009).

RVM modulation is also selective for specific sensory modalities at the spinal cord level during different periods in development. In adult animals, RVM-mediated inhibition appears to be directed to spinal circuits that receive input from C fibers (Lu et al. 2004). However, the RVM appears to primarily target A fiber input during the first three weeks of life in the rat (Koch and Fitzgerald 2014). The slow postnatal maturation of C fiber innervation and the predominant A fiber colonization in the immature SCDH could explain the preferential A fiber targeting, and the exclusive pro-nociceptive nature of the RVM, during early life (Koch and Fitzgerald 2014).

ii). Neonatal injury

Neonatal hindpaw inflammation increases opioid tone/release in the rat PAG, and decreases opioid receptor expression in this area (LaPrairie and Murphy 2009). Inhibiting the PAG via direct opioid application initiates a complex cascade of neuronal signaling through the RVM that ultimately results in increased opioid release into the spinal cord (Morgan et al. 1992). In support of the notion that neonatal inflammation-induced hyperactive opioid signaling in the PAG affects modulation of the RVM and subsequent descending inhibition onto the spinal cord, opioid antagonist administration in the PAG reduces the widespread hypoalgesia that occurs in adulthood after neonatal injury (LaPrairie and Murphy 2009).

Hindpaw incision during early life increases descending inhibition from the adult RVM onto the spinal cord. As previously mentioned, the mature RVM normally facilitates nociception at low stimulus intensities but inhibits these reflexes at high stimulation intensities (Figure 2, middle). However, after neonatal incision, the RVM is exclusively inhibitory in adulthood (Walker et al. 2009; Zhang et al. 2010). Collectively, this evidence suggests that the descending modulation of spinal nociceptive signaling is immature early in life, but importantly, also favors greater nociceptive transmission.

b). Higher brain centers

Nociception is transformed into pain in higher brain centers. A complex cortical network termed the “pain matrix” transforms these nociceptive signals into a cognitive perception of pain. The pain matrix includes the frontal, parietal, somatosensory, insular, and cingulate cortices; these cortical areas detect the intensity and unpleasantness of noxious stimuli, direct attention towards the origin of the stimuli, and react to the stimuli (Legrain et al. 2011).

i). The development of innocuous sensation in the brain

EEG analysis suggests that touch evokes non-specific, widely dispersed neuronal bursting activity in preterm infants—but more localized and sensation-appropriate activation in full-term infants (Fabrizi et al. 2011). Similarly, electrocorticographic (ECoG) recordings in rats reveals bursting activity in the primary somatosensory cortex (S1) at P8 and 11, but not at older ages (P14–30), and overall field potential strength decreases with age in S1 (Chang et al. 2016). Importantly, ECoG analysis also suggests that S1 is not able to efficiently discriminate between innocuous tactile and noxious input before P14 (Chang et al. 2016). This inefficient discrimination could potentially contribute to exaggerated somatosensory responses in infants and/or partially explain the unique effects of neonatal injury on pain behavior—as tactile input in early life directs the maturation of nociceptive circuits (Fitzgerald 2005).

ii). The development of pain perception in the brain under normal conditions

Noxious stimuli-induced hemodynamic activity in the “pain matrix”, which is thought to transform nociception into conscious pain perception, has some similarities in infants and adults. Noxious mechanical stimuli evoke a similar hemodynamic response in the somatosensory, anterior cingulate (ACC), and insula cortices in adults and children (Goksan et al. 2018). However, a painful stimulus also increases blood flow to the amygdala and orbitofrontal cortex only in adults, and these brain regions are thought to play a role in sensory stimuli discrimination and identification (Goksan et al. 2015). The lack of activation in these areas in infants could reflect an immature ability to delineate specific somatosensory input (Chang et al. 2016). There are additional brain areas that respond to noxious input only in infants, such as the auditory cortex, caudate, and hippocampus (Goksan et al. 2015). The hippocampus plays a well-established role in memory formation and storage (Bird and Burgess 2008), and the caudate is known as a feedback processor—employing stimulus-response learning as a means to use past experiences to influence future actions (Grahn et al. 2008). These regions could be playing a role in the encoding and retention of memories of painful input in the infant brain. Additionally, in adults, the parietal lobe, precuneous cortex, and pallidum are only activated on the side contralateral to the peripheral noxious input, while infants show bilateral activation of these areas (Goksan et al. 2015). This evidence again supports the notion of somatosensory input eliciting widespread, nonspecific brain activity in infants.

iii). Effects of neonatal injury on nociceptive circuits in the brain

As a result of preterm birth or complications, infants are often admitted to the neonatal intensive care unit, where they can receive an average of 14 invasive procedures per day—often causing tissue damage and pain (Simons et al. 2003). For example, heel lance causes increased blood flow to the contralateral somatosensory cortex as early as 25 weeks postmenstrual age (Slater et al. 2006). This response increases in strength as the infant ages (Slater et al. 2006), potentially indicating sensitization to the lance or normal activity-dependent strengthening of connections within the somatosensory system. The latency of these cortical responses to heel lance also decreases with age (Slater et al. 2006), likely due to continuing myelination that occurs in the nervous system throughout early postnatal development.

ECoG recordings indicate a depression in total energy in S1 shortly after incision in older rats (P14), but this effect does not occur at P8 (Chang et al. 2016). There is also a specific rebound in gamma-type energy activation after the initial depression that occurs at P14—but this is not present until P21 and increases at P30 (Chang et al. 2016). Gamma activity is associated with pain in adult rats (Li et al. 2010), suggesting altered pain processing or even blunted pain responses in younger animals.

Collectively, the above evidence suggests that neuronal circuits involved in somatosensation undergo significant maturation during early postnatal development and are highly susceptible to injury during critical periods of life. Nonetheless, it should be noted that these neuronal pathways do not function in isolation, but are rather profoundly modulated by interactions with the immune system.

4). Neuroimmune interactions in nociception and pain

The immune system is very important in the regulation of the nervous system and clearly shapes pain processing, as demonstrated by the effectiveness of NSAIDs in alleviating inflammatory pain. Immunostimulants (such as bacteria entering the body through a cut) activate peripheral immune cells such as macrophages, which then produce cytokines and prostaglandins that directly activate primary afferent neurons in the DRG (Cunha et al. 2000; Watkins et al. 1994). Cytokines at the brain-immune interface also stimulate neuroimmune cells like microglia, which ultimately promote hyperalgesia and allodynia (Hasegawa-Ishii et al. 2016; Yoon et al. 2012).

a). Development of immune signaling

i). Systemic immune system

Adaptive immune system development occurs embryonically and postnatally, as the ability of umbilical cord blood T cells to recognize pathogens is lower in preterm infants compared to those born full-term (Scheible et al. 2015). Even the innate immune system is immature early in life—immune system stimulants such as intra-abdominal LPS elicit less anti- and pro-inflammatory cytokine release in peritoneal immune cells in neonatal mice compared to adults (Gentile et al. 2014). Similarly, immune cells retrieved from neonatal cord blood also demonstrate less cytokine release after Toll-like receptor (TLR) agonist application than peripheral blood taken from healthy adults (Kollmann et al. 2009). The lack of efficacy in the neonatal innate immune system could also contribute to the greater mortality observed in early life after infection (Gentile et al. 2014; Kollmann et al. 2009; Maródi 2006).

ii). Neuroimmune system

Similar to the nervous system, the neuroimmune system develops based on external influences. Infections strengthen and prime specific neuroimmune pathways to prepare the body for future invading pathogens. The neuroimmune system undergoes a drastic switch around P21 in the rodent—when microglia in the brain become more responsive and discriminatory through TLR receptor action (Scheffel et al. 2012). Additionally, forebrain microglia exist in primarily an amoeboid (active) state during early life and do not transition to a ramified (resting) state until P10—due to a downregulation of the transcription factor Runx1 (Nayak et al. 2014). Furthermore, as seen with the systemic immune system, intraspinal injections of LPS produce less spinal microglial activation in P10 rats than adults (Moss et al. 2007). The immune system is less efficient in discriminating between, and therefore responding to, pathogens during the neonatal period (Scheffel et al. 2012).

In summary, during early life the immune system remains in a primed and ready state to induce inflammation, but a lack of ability to detect and recognize pathogens may lead to an absence of a pro-inflammatory response, which has critical implications for the effects of injury or inflammation on the somatosensory system.

b). Neonatal injury

Tissue damage can strongly modulate the neonatal immune system and thereby cause prolonged alterations in pain processing. Peripheral nerve injury provokes less microglial activation in the spinal dorsal horn during early life (Moss et al. 2007), and in fact, nerve injury actually appears to cause spinal neuroimmune suppression of neuropathic pain in young male rodents (McKelvey et al. 2015; Vega-Avelaira et al. 2012). In contrast, adult nerve injury evokes a rapid pro-inflammatory response in the dorsal horn (McKelvey et al. 2015; Vega-Avelaira et al. 2012). After nerve injury in young rodents, the activation of macrophages, microglia, and astrocytes in the DRG and SCDH is delayed until adolescence (Vega-Avelaira et al. 2012). Interestingly, nerve injury in young rats produces an anti-inflammatory response with the release of IL-4 and IL-10 cytokines in the spinal cord, which is not seen in adults. This anti-inflammatory response can also be driven by high intensity stimulation of the sciatic nerve exclusively in young animals. Blocking IL-10 signaling in the spinal cord after neonatal nerve injury unmasks neuropathic pain (McKelvey et al. 2015). Although the immature neuroimmune system appears to enter a latent, anti-inflammatory stage after nerve injury, it is important to note that rat SCDH neurons become sensitized to the pro-inflammatory cytokine TNFα after spared nerve injury at P6 (Li et al. 2009b). Taken together, these results suggest that nerve injury during early life causes an anti-inflammatory response at the spinal cord level that actively suppresses pain, but may also prime neuronal responses in the SCDH to pro-inflammatory cytokines. Subsequently, once the injured animals reach adolescence, the cytokine profile in the spinal cord becomes pro-inflammatory and drives neuropathic pain, as seen in the adult following peripheral nerve damage.

Mounting evidence suggests that neonatal tissue damage can “prime” the developing neuroimmune system. Hindpaw surgical incision in early life causes an exacerbated immune response after adult re-injury compared with controls—with higher levels of microglial activation, proliferation, and reactivity present in the adult SCDH after prior neonatal injury (Beggs et al. 2012b; Moriarty et al. 2019). Intriguingly, minocycline administered before adult re-incision completely prevents the exacerbated inflammatory spinal microglial response to re-injury in males, but has no effect on adult incision alone (Beggs et al. 2012b). Minocycline administered before neonatal incision also reduces this exacerbated microglial response to re-injury in males, but not females (Moriarty et al. 2019). Sex differences in neuroimmune activation after tissue damage has been attributed to adaptive immune system activation in females (i.e. T and B cells), while spinal microglial activation appears to promote hyperalgesia in males (Sorge et al. 2015). It is important to note that the female mechanism for immune system activation is not fully understood and has only been investigated after adult injury (Sorge et al. 2015); it is thus important to explore these sex differences after neonatal injury more thoroughly.

5). Pain behavior

a). Mechanical sensation

Early in life, mechanical sensitivity is high—and thresholds for withdrawal from tactile input increase with age (McKelvey et al. 2015; Vega-Avelaira et al. 2012). The flexion withdrawal reflex is longer, exaggerated, and more widespread in human neonates (Andrews and Fitzgerald 1994; Cornelissen et al. 2013; Holmberg and Schouenborg 1996). Additionally, although the flexion withdrawal reflex only occurs after a noxious stimulus in adults, it can be provoked in infants by mere touch. The reflex involves limbs on both sides of the body in infants, but adults only react using the ipsilateral side of the body affected by the stimulus. Finally, repeated stimuli causes a sensitization in infants, while adults habituate over time (Cornelissen et al. 2013).

b). Thermal sensation

In contrast to mechanical sensation, thresholds for noxious heat actually decrease with age in rodents (McKelvey et al. 2015; Vega-Avelaira et al. 2012). This effect also occurs in humans, although sensitivity to cold and innocuous warmth decreases with age (Huang et al. 2010). Collectively, these results suggest that not all sensory modalities are shaped by postnatal development in the same manner.

c). Affective dimensions of pain

The affective or emotional aspect of pain can have a huge effect on a child’s wellbeing or quality of life (Weiss et al. 2013). Noxious stimuli evoke cortical activation in humans as early as 24 weeks after conception (Bartocci et al. 2006; Slater et al. 2006). Pain-motivated aversion can be produced by pairing innocuous and noxious stimuli starting at around P10 in rats (Moriceau and Sullivan 2006; Sullivan et al. 2000). Sensory and affective pain circuits in the brain can be activated in tandem beginning at P12 in the rat (Sperry et al. 2017).

d). Neonatal injury

After minor tissue damage in the neonatal period, humans show a general hypoalgesia at 9–14 years old, particularly near the site of injury (Hermann et al. 2006; Schmelzle-Lubiecki et al. 2007). This is accompanied by increased sensitization to repetitive or prolonged noxious stimulation (Hermann et al. 2006; Hohmeister et al. 2009), a higher frequency of pain episodes (Hohmeister et al. 2009), and elevated pain catastrophizing or anxiety (Hohmeister et al. 2009). Rodent models of neonatal hindpaw incision also show a widespread baseline hypoalgesia, as well as an exacerbated response to re-injury, in adulthood (Beggs et al. 2012b; Walker et al. 2015; Walker et al. 2009). Neonatal inflammation via carrageenan causes a general hypoalgesia that emerges during adolescence in rats—with the first signs of decreased mechanical sensitivity apparent 31 days after injury (LaPrairie and Murphy 2007). In contrast, the exacerbation of hyperalgesia after re-injury appears much more rapidly, and can be observed immediately following the resolution of the initial injury (Ren et al. 2004).

e). Pharmacological interventions and neonatal injury

Opioids are currently the first line treatment for those with painful health conditions, and the usage of these drugs for severe pain conditions has more than doubled in the past ten years (Nahin et al. 2019). A recent study found that morphine causes severe respiratory depression in infants, and even suggested little to no analgesic efficacy of the drug in mitigating pain or brain activity after heel lance (Hartley et al. 2018).

Opioids could also potentially have long-term effects on nociceptive processing. One study found that neonatal inflammation reduces the effectiveness of morphine in adulthood by increasing the ED50 (dose required for 50% of the maximal effect), and this effect is reversed if morphine is given before the neonatal inflammatory insult. Preemptive morphine treatment before neonatal inflammation also reduces the hypoalgesia, and exacerbated response to re-inflammation, that typically follow neonatal injury (LaPrairie et al. 2008). However, this does not seem to be the case for neonatal incision, as another study found no effect of preemptive morphine (i.e. before the neonatal incision) on hyperalgesia following adult re-incision (Moriarty et al. 2018). Interestingly, neonatal morphine alone has no effect on the ability of morphine to decrease inflammatory mechanical hyperalgesia in adulthood (LaPrairie et al. 2008). However, this finding was challenged in another study that found neonatal morphine alone lowered the effectiveness of morphine in mitigating a reflex response and reducing spontaneous pain in the conditioned place preference assay after adult incision—suggesting reduced efficacy of the drug (Moriarty et al. 2018).

Using local anesthetics at the site of injury seems to prevent the somatosensory “memories” of tissue damage, at least in the case of neonatal incision. Pretreatment with levobupivacaine, a local anesthetic, prevents re-incision hyperalgesia in adulthood after prior neonatal injury (Moriarty et al. 2018; Walker et al. 2009). As mentioned previously, anti-inflammatory regimens can also mitigate neonatal hyperalgesia following adult re-incision, regardless if the treatment occurs before the neonatal incision or prior to adult re-injury (Beggs et al. 2012b; Moriarty et al. 2019). There is no doubt that the potential to develop therapies that are safer and more effective than opioids for pediatric injury and pain management exists, and further research is needed to uncover and refine these treatments for clinical use.

6). Future directions

A major focus of recent research has been identifying spinal circuits that mediate itch and pain signaling. For example, interneuron populations expressing the transcription factor Bhlhb5 or neuropeptide Y have been implicated in suppressing itch (Bourane et al. 2015; Kardon et al. 2014). The interneurons expressing Bhlhb5 include a subpopulation that expresses the opioid peptide dynorphin, and dynorphin-expressing interneurons have also been implicated in the suppression of mechanical pain (Duan et al. 2014). Additionally, inhibitory interneurons expressing the calcium-binding protein parvalbumin have been associated with reducing neuropathic pain after nerve injury, likely by inhibiting excitatory interneurons expressing the kinase PKCγ, which are thought to drive the neuropathy-induced mechanical allodynia (Petitjean et al. 2015).

This research into the function of neuronal subpopulations will likely be indispensable in pain research, but none of these studies have taken into account the dramatic changes in spinal dorsal horn circuitry occurring throughout postnatal development. As highlighted in this review, peripheral afferent input to the spinal cord, neuronal excitability, inhibitory circuits, and neuroimmune actions are drastically different during early life when compared with adulthood. Changes in primary afferent input could alter the types of sensory modalities these subpopulations regulate in the early postnatal period. High neuronal excitability coupled with poorly tuned inhibitory networks in the dorsal horn could influence the magnitude of nociceptive transmission from the spinal cord to the brain. The anti-inflammatory phenotype of the neuroimmune system before adolescence could mask the existence of aberrant nociceptive circuits which are fully revealed when the immune system matures and becomes pro-inflammatory.

Basic science research is beginning to emphasize the affective dimensions of pain to complement the longstanding efforts to understand the discriminative aspects of pain sensitivity. This is difficult in simple animals like rodents, but nonetheless seems feasible and imperative to the field. Recently, researchers have identified specific behaviors that occur after injury that are thought to resemble coping or aversive behaviors. Importantly, ablating a population of neurons that directly project to the brain eliminates these coping behaviors and conditioned aversion—supporting the notion that these behaviors are brain-mediated and represent affective processing (Huang et al. 2019). All of these factors highlight the need to address the critical gap in affective pain research during the neonatal period—not only for the sake of better understanding the developmental neurobiology of nociceptive pathways, but also for applying this knowledge directly to improve the treatment of pain in the pediatric population.

It appears that neonatal injury causes a prolonged, constitutive surge in endogenous opioid signaling in the CNS—perhaps as a means to compensate for ongoing pain (LaPrairie and Murphy 2009). This may be related to the recently discovered phenomenon of “latent pain sensitization” where increases in endogenous opioid signaling persistently mask pain after injury, and opioid antagonist administration long after recovery from the injury are able to reinstate hyperalgesia and allodynia (Corder et al. 2013; Walwyn et al. 2016). Hyperactive endogenous opioid signaling could also potentially affect the efficacy of externally administered opioids as analgesics, the motivation to self-administer pain treatments after injury, and the propensity for substance abuse later in life. Therefore, more research is warranted to determine the efficacy of opioids during the neonatal period and identify the long-term effects of these drugs on the maturation of developing nociceptive circuits.

It is also critical to find alternate, non-opioid based approaches to manage pediatric pain, as these drugs can cause serious side effects such as respiratory depression, sedation, and gastrointestinal disturbances. Furthermore, the clinical utility of opioids is limited by tolerance and the risk of addiction (Morgan and Christie 2011), which highlight the need for alternative therapies. In the race to find these therapies, it is imperative to recognize the distinct functional organization of immature nociceptive pathways and their unique response to injury. A consideration of age as an independent variable in biomedical research will undoubtedly facilitate efforts to design novel interventional strategies that are best suited to alleviate pain in the pediatric population.