Early life programming of pain: focus on neuroimmune to endocrine communication

Physiological pain plays an essential and primordial function in our survival. It signals potential damage to the body and produces a wide range of sophisticated actions to prevent this damage. The perception of noxious stimuli is typically initiated by the activation of peripheral receptors (i.e. nociceptors) which signal the spinal dorsal horn and supraspinal neurons of any potentially tissue-damaging stimuli. In pathological conditions, however, failure to sense pain such as in the case of congenital insensitivity often lead to self mutilation, bone fractures, joint deformities, amputations, and even early death, underscoring the importance of the physiological, protective role of nociceptive pain [1]. Chronic pain, the unbearable pain that persists long after an injury has healed, can drive the individual to immobility, psychological distress, depression, and disruption of family relationships, and potentially loss of employment. Chronic pain is common in developed countries. Prevalence rates of 16–22 % have been reported from epidemiological surveys [2–4]. One recent study lists chronic back pain (along with depression) in the top 10 causes of morbidity in over 180 countries [5]. Chronic pain incurs a multi-billion dollar cost to most developed western countries [6, 7]. Despite mounting evidence regarding the complexity of the nature of chronic pain there remains a general resistance [8] to the understanding that the nature of chronic pain is not primarily a matter of nociception (i.e. a noxious stimulus generating neural impulse), despite ample evidence of the multidimensional nature of chronic pain [9].

Pharmaceutical treatments, still our primary treatment modality, have a 25–30 % placebo rate, making therapeutic benefits difficult to establish. Some of the drugs such as the use of opioids (i.e. morphine), which can act on opioid-receptor-expressing neurons in brain areas that control pain (i.e. PAG) are effective but have a very narrow therapeutic window and in most of the time the dose needs to be limited to avoid side effects or analgesic tolerance which occurs upon repeated administration of opioids requiring increased dosage to maintain satisfactory level of pain control. Why in the 21st century is pain still a stubborn condition? What are the limitations of the current therapeutic approaches? Are we understanding and managing pain in its full complexity?

While our understanding of the complex mechanisms involved in the development and maintenance of chronic pain has increased dramatically in recent decades, treatments in various modalities continue to be based on a reductionist approach that understanding the precise nociceptive mechanisms that cause pain, will result in more targeted and hence more effective treatments. Given the myriad of neuroimmunoendocrine links now understood to modulate chronic pain, focusing on targeted specific mechanisms may be missing the bigger picture. Furthermore, given that pain has also a cognitive, aversive component there is a need to target the cognitive/emotional aspect of pain not only the sensory discriminative component.

From Galen to the gate-control theory developed by Melzack and Wall, pain has been viewed either as a separate, independent sensation, an emotion, or a result from activation of nociceptors, spinal and supraspinal pathways. This review proposes a shift in conceptualising the problem of chronic pain away from a dichotomous standpoint to an integrative view which captures both the sensory and the emotional-cognitive in a framework that emphasises the neuroendocrine to immune communication pathways and the relevance of early premorbid events as predictors.

Specifically we focus on the early life environment and its ability to program the organism’s physiological, affective and behavioural responses in later life [10–14]. Almost invariably, randomised control trials (RCTs) commence their assessment of pain by assessing the baseline values from the time the pain started. There is an implicit assumption, that prior to the onset of pain, “patients” are physiologically naïve and functionally equivalent—though there is some acknowledgement that their psychosocial history may differ. The thesis proposed here is that a neurodevelopmental understanding is required to account for the individual variability in response to the initiating noxious insult/stimulus as well as to subsequent treatments. We argue that not taking account of the premorbid physiological milieu may be limiting efforts to develop more tailored and more effective treatment interventions. In short, chronic pain patients may be differentially predisposed to recovering (or not recovering) from a physical injury, infection or trauma because of their premorbid psychoneuroimmunological status.

Sixty-plus years of intensive research in the pain field have led to the discovery of countless drugs including aspirin and pioneer findings such as the induction of analgesia by direct stimulating of the periaqueductal grey (PAG) in rodents [43] and humans [44]. Opioids are effective in attenuating the noxious stimulus. These molecules (whether endogenous or exogenous) exert their action through activation of opioid receptors which are μ, δ (DOR/OPDR), κ (KOR/OPRK1), and opioid-like receptors (ORL1/OPRL) [45] that are expressed peripherally by leukocytes or by primary sensory neurons but also can be expressed centrally in key nuclei involved in the descending pain inhibitory pathways such as PAG and rostroventral medulla (RVM). However the use of opioids in modulating pain has a downside, they can be addictive and enhance alcohol consumption. They can also induce opioid tolerance where repeated exposure to opioids decreases drug efficacy over time. The use of slow-release drug patches, local anaesthetics, or spinal epidurals, which can act either by blocking the transmission of noxious input to the brain or by attenuating the excitability of spinal and supraspinal neurons have brought relief to some patients but not all. Surprisingly drugs that were traditionally prescribed to control depression were shown to be effective in treating chronic pain and the same discovery applied for drugs that were originally designed to treat epilepsy and were found to relieve severe chronic pain. These drugs shifted the focus onto the brain. However despite the effectiveness of these drugs, treating chronic pain is still challenging for clinicians. We argue that one of the reasons resulting from this challenge is the complexity of the interaction between multiple physiological systems that modulate pain as well as the role of early life events in programming pain responses later in life. Early life exposure to adverse events whether it is of psychological, physiological, bacterial or viral nature has been shown to exert programming effects on the hypothalamus–pituitary–adrenal axis [10, 46–48], a key system that modulates the stress response to physiological or psychological stressor. Given that pain is an aversive, stressful experience, these early adverse events are also capable of altering future pain responses through action on the endocrine system. Below we present clinical and animal evidence that demonstrate the impact of early life insult on neuroendocrine and pain responses.

The experience of stress, from an evolutionary perspective, plays a pivotal role in promoting survival of the organism. Perceiving an environmental stimulus as threatening and learning to avoid it is essential for survival. The HPA axis is the major neuroendocrine mediator of the stress response and is known to be susceptible to perinatal programming [49, 50]. The cascade of signalization in the HPA axis in response to stress begins with activation of a group of neurosecretory neurons in the paraventricular nucleus of the hypothalamus (PVN). These neurons secrete corticotropin-releasing hormone (CRH) and to a lesser extent arginine vasopressin (AVP) into the portal vessel system to reach the anterior pituitary where the synthesis of pro-opiomelanocortin (POMC) takes place and then processed to adrenocorticotropic hormone (ACTH) [51]. Subsequent increase in circulating ACTH then drives the synthesis of cortisol (in humans) and corticosterone (in rodents). Attenuation of the magnitude of the HPA stress response occurs through negative feedback of glucocorticoids on glucocorticoid (GR) and mineralocorticoid (MR) receptors expressed in neurons of the hippocampus, hypothalamus, and pituitary. This results in a negative feedback loop [52, 53] and maintains homeostasis. The HPA axis is therefore designed to provide a quick and appropriate response to stress by stimulating a relatively brief influx of glucocorticoids into the body. Moreover, this responsivity is determined, in part, by the relative affinity of glucocorticoids for MRs and GRs [54]. The HPA axis goes through a period of significant change and development during the perinatal period and exposure to adverse events during this period is known to produce long-term alterations in neuroendocrine function and behavior.

Nociceptive neural circuitries undergo significant changes during the early neonatal period, and therefore remain vulnerable to the effects of early insults. For instance, substance P, a specific marker of peptidergic C fibres, is present at birth suggesting the development of chemical properties of these fibres at an early stage. The peptidergic C fibers are a population of C fibers that release neuropeptides such as substance P and calcitonin-gene-related-peptide (CGRP) as well as the Trk A neurotrophin which binds to nerve growth factor (NGF) [31]. In the dorsal horn, substance P fibres are concentrated in Lamina I on day 1; in the next few days, the density in Lamina II is increased and by the end of the first week of development, the staining of the fibers is comparable to the one observed in adults [87]. Moreover, the non-peptidergic C fibres that express the IB4 isolectin cannot be detected in the synapse until PND 5 despite their presence in dorsal root ganglion from embryonic day 18. This delay suggests that these fibres form synapses later than peptidergic C-fibres. Finally, both neonate A-fibres and C fibres express µ-opioid receptors, although this expression becomes restricted to C fibres by PND 21 [88].

There is a functional maturation of spinal dorsal horn which is also developmentally regulated. Dorsal horn cells in deep laminae do not respond to C fibre input until PNDs 7–8. Similarly, specific C-fibre-evoked reflexes produced by mustard oil (a chemical skin irritant) do not occur until rat pups reach day 10–11 [89]. The activation of C fibres is unable to evoke spike activity in the spinal cord before the second postnatal week. Aδ fibres are completely unmyelinated until the pups reach 2–3 weeks [89]. In the mature state, Aβ fibres which convey non-noxious stimuli project into Laminae III to V [90], while nociceptive fibres terminate in the superficial laminae (Laminae I and II). However, in the immature state large myelinated fibres, which normally integrate non-painful stimuli such as touch, are found to project to Laminae I to V including Laminae II. In other words, fibres that normally convey non-noxious stimuli can integrate noxious information as well. At the supraspinal level, the inhibitory descending control system is not apparent until the second postnatal week. Interestingly, electrical stimulation of the PAG does not produce analgesia until at least PND 10 [91]. Of particular interest, a recent study showed that there is a developmental postnatal shift in the brainstem rostroventral medulla (RVM) control of spinal dorsal horn neuronal circuits from A fibre facilitation activity in young animals to inhibition of nociceptive C fibre input in the adult animal [92].

Taken together, these data demonstrate that the nociceptive system during early life presents an important “window of vulnerability” during which an early insult such as inflammation, bacterial infection or a polymodal stimulus (mechanical, chemical or thermal), may result in the disruption of the neuronal and chemical circuitry modulating pain, leading to altered behavioural and neuronal responses to noxious stimuli later in life.

Traditionally implicated in defending the organism against microbial pathogens or “non-self”, the immune system is now well appreciated to play a critical role in pain modulation [108–110]. Following injury, immune cells release proinflammatory cytokines and other inflammatory molecules such as histamine, prostaglandins, eicosanoids and cytokines that sensitize nociceptors. For instance, nociceptors are known to be interleukin (IL)-1-β sensors and IL-1-β can directly activate nociceptors to generate action potentials and induce hyperalgesia [111]. Moreover, intraplantar injection of IL-1-β increase the discharge of spinal dorsal horn neurons to thermal and mechanical stimuli [112]. The contribution to pain hypersensitivity can be mediated by inflammatory mediators released by immune cells but also by immune cells infiltration per se. The release of cytokines contributes to the recruitment of granulocytes, monocytes and mast cells at the site of inflammation. An increasing body of evidence suggests that infiltration by macrophages and T-cells in the dorsal root ganglia following nerve injury [113, 114] as well as activation of microglial cells in the spinal cord that coincides with enhanced neuropathic pain [115]. Hyperalgesia following nerve injury is substantially reduced in mice lacking T cells [116–118] and following the inhibition of microglia by minocycline [119]. Infiltration of immune cells at the peripheral level is also important in contributing to pain modulation. Following injury, mast cells are first to infiltrate the site of inflammation and they can degranulate within minutes of inflammation leading to the release of histamine, prostaglandin, bradykinin and other inflammatory markers that participate in the vasodilation and nociceptors sensitization [120]. Interestingly, mast cell degranulation requires direct contact with nociceptors via molecule of adhesion (N-cadherin) [121]. In our laboratory, we have demonstrated that neonatal exposure to the bacterial mimetic lipopolysaccharide (LPS) produced hyperalgesia in preadolescent rats that coincided with enhanced mast cell degranulation [23]. In a recent clinical study, Di Nardo et al. showed that children suffering from Irritable Bowel Syndrome (IBS) exhibit an increased number of mast cells in the proximity to nerves in the ileal and colonic mucosa [122]. More importantly, the intensity of abdominal pain reported by these children was significantly correlated to the close proximity of mast cells to nerve endings in the gut wall [122].

Other events that can contribute to the induction of the immune system are exposure to viruses, lipopolysaccharide (LPS), cell wall of gram-negative bacteria, or fungal infections during both the neonatal and adult life. When exposed to such components, immune cells including monocytes, lymphocytes, leukocytes, macrophages, T cells and mast cells not only signal other components of the immune system (i.e. complement) but also the brain [123] to control the inflammatory response and to maintain it at homeostatic levels. This is made possible by the release of a plethora of pro-inflammatory cytokines including IL-1-β [124], IL-6 [125], IL-8 [126], IL-12 [127], interferon-α (IFN-α) and -δ (IFN-δ) [128] as well as tumor necrosis factor (TNF)-α [129, 130]. Since cytokines are large molecules that are unlikely to cross the blood brain barrier (BBB) [131], it has been postulated that they can act on the brain through different pathways. For instance, via the synthesis of prostaglandin E2 by brain microvessel endothelial cells [132]. Another possibility is through vagal afferences as subdiaphragmatic lesions of the visceral vagal afferents terminating in the nucleus tractus solitaris abolished the LPS or IL-1β-induced hyperalgesia [133]. Subdiaphragmatic lesions of the vagus nerve also abolishes expression of Fos-protein in the secondary projections of this nerve such as the paraventricular nucleus of the hypothalamus (PVN), central nucleus of the amygdala (CeA), and the bed nucleus of the stria terminalis following IP injection of LPS [134]. This indicates that systemic LPS requires an intact functional vagus nerve to induce neuronal activation of brain regions involved in neuroendocrine function such as the PVN. It has also been suggested that cytokines can reach the brain via areas devoid of BBB such as the organum vasculosum lamina terminalis (OVLT) [135] or the macrophage-like cells commonly known as circumventricular organs [136]. The brain production of IL-1β following peripheral inflammatory stimuli such as in the case of LPS administration can be first restricted to choroid plexus and circumventricular organs, and then to the brain side of the BBB by slow diffusion (via volume transmission) [137]. Finally cytokines can reach the brain through a saturable transport system, which although has been demonstrated to exist in mice but not rats [138, 139], its physiological relevance remains unclear.

All of these neuroimmune pathways ultimately lead to the production of cytokines by brain microglia and induce a set of physiological responses commonly known as “sickness behavior”, which include fever, malaise, lethargy, decreased appetite and libido [123, 140]. The sickness response is an adaptive mechanism assisting in the recovery process from acute viral/bacterial infections. For instance fever, induced by the action of the pyrogenic IL-1 on anterior hypothalamus, elevates the body’s temperature to assist the immune system in fighting infections by blocking the growth of microorganisms [141]. Fatigue helps conserve energy so much needed in order to fight infection. However, in some circumstances, this process can go seriously wrong such as in the case of uncontrolled, exaggerated and prolonged inflammatory response in the CNS. This maladaptive response can be exemplified in clinical settings of post-infectious chronic fatigue syndrome [142], irritable bowel syndrome [143], or post-viral depression [144]. Importantly, pain facilitation or hyperalgesia is also considered to be an integral part of sickness behavior [145, 146]. Proinflammatory cytokines released by immune cells are known to induce hyperalgesia when administered at the peripheral or central level. Among the many types of pro-inflammatory cytokines, IL-1β is the most characterized potent mechanical and thermal hyperalgesic agent when injected into peripheral tissue or the CNS [112, 133, 147, 148]. For instance, intracerebroventricular (ICV) injection of the recombinant human IL-1β (rhIL-1β) in rats significantly decreased paw lick latency to thermal noxious stimulus [148]. This thermal hyperalgesia was abolished by ICV administration of the IL-1β antagonist IL-1ra [148]. In our laboratory, we have demonstrated increased formalin-induced behavioural responses in LPS-treated adult rats that paralleled enhanced circulating IL-1β 1-h following formalin injection [23]. Additionally, adult rats treated with LPS at PND 3 and 5 displayed increased IL-1β in the hippocampus 1 h following formalin injection, an effect that coincided the increased formalin-induced nociception observed in adult LPS-treated rats. The neurocircuitry underlying central action of cytokines is not only restricted to the brain but can also involve the spinal cord by increasing the excitability of spinal dorsal horn neurons [149, 150]. As discussed previously, activation of microglia at the spinal level plays a critical role in producing this cytokine-induced hyperalgesia [151].

Neonatal or adult exposure to the bacterial mimetic has been reported in both clinical and animal studies to produce increased pain sensitivity later in life. Boisse L et al. were the first to report that rats administered with LPS at PND 14 exhibited a significant decrease in withdrawal latency to thermal and mechanical stimuli in adulthood [152]. These adult rats displayed allodynia (a stimulus that is normally not painful is perceived as painful) and hyperalgesia (exaggerated responses to stimuli) in both tests of thermal (i.e. hot-plate test) and mechanical nociception (i.e. Von Frey test). Of particular interest, no significant differences were observed in nociceptive responses when a second LPS injection was administered in adulthood. This implies that a single LPS administration is enough to produce long-term alterations in nociceptive responses. Moreover, the thermal and mechanical hyperalgesia in neonatally LPS-treated rats was associated with upregulation of COX-2 protein level in the spinal cord [152], suggesting a potential role of prostaglandins in the hyperexcitability of spinal dorsal horn neurons in LPS-treated rats. We have replicated these findings such that rats exposed to LPS during PND 3 and 5 displayed increased formalin-induced nociceptive responses (i.e. flinching and licking) following formalin injection at PND 22 and PND 80–97 but not PND 7 or PND 13 [10, 23, 24]. This LPS-induced hyperalgesia was accompanied by peripheral and central neuroimmune and neuroendocrine changes as well as altered spinal and supraspinal neuronal responses [10, 23, 24]. More recently Hunter D and colleagues reconfirmed these findings showing that at 4 h and 8 h post LPS administration, PND 21 rats (PND0 as birth) displayed enhanced formalin-induced nociception during the late phase [153]. Adult LPS administration in rats is also reported to significantly decrease the tail-flick latency, which starts within few minutes of LPS injection and lasts for 60 min [154]. The LPS-induced hyperalgesia can be blocked following administration of IL1-ra, suggesting a critical role of IL-1 in inducing the LPS-induced hyperalgesia [155]. These animal studies are consistent with clinical findings showing that humans subjected to an IV injection of LPS exhibited enhanced flare, hyperalgesia, and allodynia in response to capsaicin administration [156] as well as lower pressure pain thresholds that correlated with enhanced circulating levels of IL-6 and TNF-α, an effect more pronounced in women [157]. This is consistent with a recent study, which demonstrated that IV injection of LPS produced a significant decrease in rectal and pressure pain thresholds [158]. This LPS-induced visceral hyperalgesia was accompanied by a significant increase in circulating pro-inflammatory cytokines as indicated by a peak of TNFα and IL6 1 and 2 h (respectively) following LPS administration [159].

Taken together, these animal and clinical findings suggest that far from the traditional view that posits that pain can solely be attributed to activation of neurons, the immune system via peripheral and central release of pro-inflammatory cytokines contribute significantly to pain modulation. Integrating the neuroimmune interface in pain modulation is thus a necessary step in order to find efficient treatments to treat chronic pain.

During the late 80s, a series of elegant studies mainly carried out by Sapolsky and Besedovsky group have demonstrated that components of the immune system, namely cytokines and in particular IL-1β can directly activate the HPA axis leading to the release of glucocorticoids [160, 161]. In a clinical study on post-mortem tissue, immunohistochemistry staining using antiserum directed against human IL-1β revealed that the human hypothalamus is densely innervated by IL-1β fibers [161]. The densest innervations were found in the periventricular region that controls the anterior pituitary. IL-1β-immunoreactive fibers were also found in the arcuate nuclei and parvocellular part of the paraventricular nucleus [161]. In rats, intravenous injection of IL-1β induced cfos expression in corticotropin-releasing hormone (CRH) containing neurons in the paraventricular nucleus of the hypothalamus (PVN) [162]. Systemic or intracerebroventricular injection of IL-1β also induced the release of circulating corticosterone and adrenocorticotropic hormone (ACTH) [163, 164]. Moreover, immunoneutralization of IL-1β or CRH blocked the stimulatory effect of IL-1β on ACTH [164]. This response is primarily mediated by the release of CRH from the PVN [164]. Additionally, IL-1β has been reported to act directly on the anterior pituitary cells to release ACTH [165].

Adrenocorticoids such as cortisol in humans and corticosterone in rats are known for their immunosuppressive function [166]. However their exact contribution to pain modulation is still not clear and literature regarding the role of glucocorticoids in modulating pain is at best conflicting. It is not surprising that glucocorticoids play an important role in pain modulation since pain is an aversive stressful event and hence capable of activating the HPA axis. Whilst several clinical studies have reported hypocortisolemia in patients with fibromyalgia and chronic low back pain [167–169], others have found that corticosterone reduces the inflammation associated with pain [170, 171]. In our laboratory, we demonstrated that neonatal LPS exposure produced a developmentally regulated activation of HPA axis activity following a noxious stimulus (i.e. formalin injection) as indicated by increased plasma levels of corticosterone in PND 13 and 22 but not PND 7 rats [10]. Preadolescent rats neonatally treated with LPS also exhibited decreased GR mRNA transcripts in the hypothalamus compared to rats that had saline during the neonatal period [10]. This is in agreement with previous studies indicating that noxious stimuli including formalin injection are capable of activating the HPA axis [172–174]. These findings indicate that the relationships between peripheral HPA axis activity and pain modulation is complex. Further studies investigating the exact contribution of the HPA axis in pain modulation are needed.