The link between chronic pain and Alzheimer’s disease

Dysfunction of LC-NE system associated with chronic pain has been reported mostly in animals. Significant increases in the expression of tyrosine hydroxylase, dopamine beta-hydroxylase (DβH), the NE transporter, the α₂-adrenoceptor, and burst firing have been seen in the LC of rats with a chronic constriction injury, a widely used neuropathic pain model [47, 49]. This results in more NE release [50, 51], although these changes were only evident when pain became chronic (28-day pain duration) and not during the acute period (7-day pain duration) [47, 49]. Although most of the animal studies reported that the LC is activated in chronic pain, one study showed opposite results [52]. In particular, in streptozocin-treated rats, a model of diabetic neuropathy, LC firing activity and expression levels of tyrosine hydroxylase, pCREB, and the NE transporter were reduced in the LC when anxiety-like and depression-like behaviors were observed, a time point considered to be representative of chronic pain [52]. The distinct nociceptive sensitivity time-courses as well as the LC functions between the chronic constriction injury and streptozocin models indicate that specific neuroplastic mechanisms depend on pain modality [52], which adds a layer of complexity in determining how chronic pain may be affected by LC-NE neurotransmission.

Changes in LC function and plasticity will affect NE transmission in numerous brain regions, and it is very possible that changes in LC-NE function are not uniform but instead vary based on downstream targets. For example, abnormally increased LC-PFC neurotransmission has been widely reported in animals with chronic pain and leads to increased noradrenergic fiber sprouting and intrinsic excitability in the PFC, mediated in part by α₂-adrenoreceptors and HCN channels [53]. This increase in plasticity in the PFC coincides with shifting NE neurotransmission in other specific brain areas innervated by the LC [54]. Despite this increase in noradrenergic markers, microdialysis techniques showed that 28 days after chronic constriction injury, no significant difference was found in basal NE release in the LC and PFC in rats [49]. However, another study showed that in the PFC, the sensitivity of α₂-adrenoceptors is enhanced, and 6 weeks after spinal nerve ligation, the NE content of the PFC was augmented significantly, coupled with impaired attention [50]. These data suggest that pain duration is an important factor for the neuroplastic LC changes and also suggest that changes in LC-NE neurotransmission is likely more nuanced than those determined by measuring basal neurotransmission alone [47].

Overall, chronic pain-induced LC-NE system dysfunction during chronic pain is complex and likely depends on the chronic pain duration, modality, and downstream neuron function and location. However, in terms of brain regions that are affected by cognition, such as the PFC, enhanced neurotransmission of NE results in increased excitability that coincides with aberrant cognitive and emotional behaviors.

Microglia are the primary innate immune cells in the CNS, which is otherwise relatively immunoisolated [55]. As one of the constituent cells of blood-brain barrier, microglia express many kinds of receptors which recognize exogenous or endogenous insults to the CNS and initiate an immune response [56]. For example, it has been reported that microglia control the spread of neurotropic virus infection [57], which can induce encephalitis and neurodegeneration [58].

Microglial activation includes pro-inflammatory activation and anti-inflammatory activation [59, 60]. Microglial pro-inflammatory activation promote neuropsychiatric disease through release of pro-inflammatory molecules such as tumor necrosis factor-α (TNF-α), interleukin (IL)-1β, IL-6, inducible nitrous oxide synthase (iNOS), and reactive oxygen species [60]. Activated microglia due to chronic pain may play critical roles in pain occurrence and maintenance [61]. In chronic pain, neuroinflammation driven by microglia is a characteristic feature [62]. In fact, microglia are initiators of a postinjury, neuroimmune response that contributes to the transition from acute to chronic pain [63]. In the chronic pain state, pro-inflammatory microglia release cytokines and chemokines associated with inflammation, such as IL-6, IL-1β, and TNF-α [64, 65]. This pro-inflammatory state leads to changes in synaptic remodeling, brain connectivity, and network function [66].

Chronic pain-induced microglial activation and neuroinflammation have been reported in both humans and animal models. In one study using PET-MRI, patients with chronic lower back pain showed increased microglia and astrocyte activation in the thalamus and the putative somatosensory representations of the lumbar spine and leg, as evidenced by elevated signal of translocator protein, a marker of microglia and astrocytes [67]. In another study, by using [¹¹C]PBR28 PET, microglial activation signal was widely elevated in certain cortical regions of patients with fibromyalgia, a condition typified by chronic, neuropathic pain [68].

Additionally, a large number of animal studies have shown that chronic pain increases microglial pro-inflammatory activation and neuroinflammation in the brain. Specifically, microglial activation has been detected in the PFC [64, 69, 70], hippocampus [71], anterior cingulate cortex [72], amygdala, nucleus accumbens, thalamus, and sensory cortex [64, 73, 74] in different kinds of chronic pain animal models [75]. For instance, rats with spared nerve injury showed microglial activation in the PFC as evidenced by increased expression of CD68, iNOS, IL-1β, IL-6, TNF-α, and 8-OH-dG [70, 74]. Spared nerve injury mice also displayed TNF-α upregulation in bilateral hippocampi [65]. Similarly, rats with spinal nerve ligation showed microglial activation in the PFC [69], and mice with chronic constriction injury developed increased microglial inflammation in the mPFC, hippocampus, and amygdala evidenced by increased CD11b and TNF-α expression [64].

In all, chronic pain is associated with upregulation of pro-inflammatory microglia in varied brain areas, including those responsible for cognition and emotion. Microglial activation induces neuronal dysfunction, such as the LC-NE dysfunction, which may feed forward to enhance further microglia-driven neuroinflammation.

Similar to chronic pain, the function and structure of the LC are disrupted in AD, and the LC is one of the earliest brain regions affected during AD development and progression [77]. A loss of LC-noradrenergic neurons is observed in autopsy specimens of AD patients, with estimated 50% [78] to 60% [79] LC cell loss, which is more dramatic than the 41% neuron loss observed in the PFC [80]. However, there is evidence that axonal sprouting and dendritic arborization increases, perhaps as compensation for the loss of cells [81]. Concomitantly, increased activity of surviving LC neurons in AD patients is also reported [82, 83], and post-mortem autoradiography detected significantly decreased NE transporter in the human AD brain, again suggesting compensatory mechanisms adjusted for cell loss [84].

Due to the complexity in these degenerative and compensatory mechanisms, it is not surprising that there are conflicting reports of the levels of NE in various brain areas in AD patients. Some reported a decrease in NE and some demonstrated that NE levels in AD patients remain constant or even elevated in the cerebrospinal fluid [76, 78, 85]. These differences in NE neurotransmission across studies may also reflect different points in AD pathogenesis. Therefore, although neuron loss occurs in the LC, NE content does not necessarily decrease in the AD brain, as the system likely compensates by increasing LC excitability and decreasing NE reuptake, and increased NE early in the disease may further exacerbate neurodegeneration. Although noradrenergic neuron loss in the LC correlates significantly with the duration of AD [86], NE concentration decreases were found in brain regions associated with cognitive deficits, namely the midtemporal cortex and the orbital-frontal cortex [78]. Mirroring the loss of LC neurons in humans, animal models of AD also demonstrated significant aberrations in the LC-NE system. For instance, the APP/PS1 mouse has demonstrated degenerated noradrenergic neurons and fibers in the LC [87]. Further, hyperphosphorylated tau causes reduced NE neurotransmission from the LC to the forebrain and certain subcortical areas, such as the hippocampus, suggesting that tau may be one of the AD-related mediators of LC dysfunction [88]. These studies suggest that patients with AD have a progressive decrease in LC neurons, and part of the clinical symptoms of AD may be due to reduced noradrenergic neurotransmission mediated by LC degeneration, especially in mid- and late-stages of disease progression [89].

Homeostasis of the LC-NE system is important for controlling central inflammation because numerous studies have shown that NE can effectively inhibit inflammation, including microglia-related neuroinflammation [19]. On the other hand, recent studies indicate that NE could be pro-inflammatory [90]. Therefore, when the homeostasis of LC-NE system is destroyed, either excessive or insufficient NE may lead to neuroinflammation.

Microglia are well equipped to respond to NE signaling by expressing the noradrenergic α_1A, α_2A, β₁, and β₂ receptors [91, 92] and strong interactions exist between NE and microglia [87, 92]. Some evidence supports that pro-inflammatory activation of microglia may be due to activation of the LC and abnormal and excessive release of NE [90]. NE enhances Aβ-mediated IL-1β secretion through action at β-adrenoceptors in THP-1 cells [93]. NE also increases COX-2 and prostaglandin E2 production induced by LPS via β-adrenoreceptors in rat primary microglia [94]. NE plays both roles as a facilitator or a suppressor for microglial pro-inflammatory reactions via activating cAMP and modulating downstream MAPK and NF-κB signaling [95]. Therefore, chronic pain-induced LC-NE neuron hyperactivity and increased supply of NE to brain areas such as PFC may result in microglial pro-inflammatory activation and exacerbate neuroinflammation in these areas in AD [64, 70].

However, more evidence indicates that NE inhibits microglial activation and suppresses the production of pro-inflammatory factors such as IL-6 and TNF-α [91]. Studies suggested that NE dampens microglial reactivity [96], as NE negatively regulated the transcription of inflammatory genes encoding pro-inflammatory cytokines and chemokines in microglia [97, 98], likely through direct action on adrenergic receptors on these cells [92]. For instance, noradrenergic depletion with DSP4 treatment increased microglial activation and the expression of iNOS and COX2 in the hippocampus and the frontal cortex of aged APP V717 transgenic mice [87]. Co-activation of β1 and β₂ in hippocampal slice cultures reduced microglial activation from a pro-inflammatory LPS plus oxygen-glucose deprivation insult and resulted in an overall reduction in TNF-α, IL-6, and MCP-1 [99]. Similarly, β-adrenergic agonism in microglia co-cultured with cortical neurons protected these cells from death by downregulating TNF-α, IL-6, and free radical expression [100, 101]. In line with the dampening of microglial activation, catecholamines also inhibit nitric oxide production from microglia, perhaps by causing a decrease in iNOS [101].

Recently, inflammation-associated PET studies demonstrated microglial activation in the brains of AD patients. In addition, microglial activation was found occurring before cognitive decline in AD patients, suggesting that it may be an early precipitant of AD progression [102]. By using PET-MRI in AD patients, two peaks of microglial activation were detected during the trajectory of AD pathogenesis, which may represent an early protective peak and a later pro-inflammatory peak [103].

Microglial activation and neuroinflammation are found in the brains of AD animal models. For example, microglia show increased proliferation in well-characterized mouse models of AD, including APP/PS1, 5XFAD, and APP23 mice [104, 105], and increased expression of pro-inflammatory markers such as CD36, CD14, CD11c, MHC-II, and iNOS [106, 107].

Microglial activation may play a dual role in AD. At the early stage of AD, activated microglia display mainly anti-inflammatory phenotypes [108], while chronic activation of microglia contributes to neurotoxicity and induces synapse loss by triggering pro-inflammatory cascades [109, 110]. Similar to microglial activation in chronic pain, loss of microglial homeostatic functioning and subsequent transition to prolonged and pathological neuroinflammation exists in AD patients [109, 110]. Despite the complexity in the role of microglia in AD pathogenesis, studies suggest that during AD progression, microglia predominantly support pro-inflammatory processes and promote cognitive decline. In fact, reactivations of microglia and neuroinflammation are now considered characteristics of AD pathogenesis [111].

Microglia are found clustered around amyloid plaques in both humans [112] and AD mice [113] and have been shown to regulate plaque dynamics [114]. Under non-pathological conditions, microglia play an important role in regulating Aβ deposition [115], and abundant evidence suggests that properly functioning microglia are involved in the clearance of Aβ and limiting the expansion of plaques [116, 117]. With early exposure to Aβ, activated microglia may phagocytose toxic Aβ and produce survival-promoting trophic factors in the AD brain [118]. However, some studies show that microglia become activated with prolonged exposure to Aβ and will undergo a pro-inflammatory response [114, 119], resulting in the secretion of synaptotoxic/neurotoxic cytokines, chemokines, and reactive oxygen/nitrogen species [118, 120, 121]. This is highlighted by studies in post-mortem AD brains and mouse models, where prolonged Aβ deposition leads to alterations in microglia, such as P2X7 receptor upregulation and activation of the innate immune response characterized by release of pro-inflammatory cytokines, acute phase proteins, and complement components [122‐124] that cause microglia-mediated synapse and neuron loss [125].

In addition, microglia may release large amounts of fibrillar Aβ and can promote Aβ plaque formation [126]. In addition to directly mediating neurodegeneration, microglia-derived ASC specks may cross-seed Aβ in AD, causing progression of the proteinopathy and spread of pathogenesis across the brain [127]. Although the relationship between microglia and Aβ is complex, the sustained presentation of Aβ seems to promote pro-inflammatory activation of microglia that ultimately furthers the synapse loss that is so highly correlated with cognitive decline.

Neurofibrillary tangles in the brain of AD patients increase in parallel with colocalized expression of microglial pro-inflammatory activation and tau kinases [128]. Hyperphosphorylated tau, misfolded tau, and truncated tau co-occurs with microglia proliferation and increased expression of inflammatory genes such as Aif1 (encoding IBA-1), Ptgs2 (encodes COX2), IL-1β, IL-6, and Tnf-α in the LC [128]. These findings suggest that activated microglia may participate in driving tau pathology in AD.

It has been extensively reported that trans-synaptic propagation of tau occurs through anatomically connected synapses; however, microglia are implicated in spreading tau [129] by endocytic phagocytosis and exocytic release of exosomes in pathways independent of synaptic transmission [130]. Microglial uptake and exosomal release of tau may play a key role for tau spreading between cells in the brain [130]. In vitro, microglia isolated from human AD cases and rTg4510 tauopathy mice are capable releasing tau seeds [129]. These microglia took up tau in the conditioned media but cannot entirely neutralize its seeding activity. These data suggest that microglia only have limited capacity to take up and break down seed competent tau, and inefficiency in this process may play a role in the spread of tau pathology [129]. In addition, evidence indicates that reactive microglia are sufficient to drive tau pathology and are highly correlated with the spread of pathological tau in the brain [131], while depletion of microglia suppresses tau propagation [130]. In transgenic tau mouse models, when microglia were activated by deleting CX3CR1, tau pathology was increased [131, 132]. These studies suggest that microglial activation contributes to tau pathology during AD pathogenesis.

Maintaining the balance of the LC-NE system may help the prevention and treatment of AD. Animal studies suggested that modulation of microglial activation by increasing NE level would be one approach for AD alleviation [133]. NE could be neuroprotective against Aβ toxicity through redox cycling and reduction of intracellular oxidative stress [134]. The NE precursor L-threo-3,4-dihydroxyphenylserine (L-DOPS) rescued spatial memory deficits in DBH^−/− AD mice [135]. It is also reported that L-DOPS restored the balance of inflammation, facilitated microglia migration and Aβ phagocytosis, and reversed learning deficits in DSP4-induced LC lesion AD mice [87]. Thus, increasing NE concentrations may facilitate anti-inflammatory functions of microglia in AD and promote microglial migration and phagocytosis of Aβ.

Although aforementioned animal studies showed that increasing NE levels can improve AD symptoms, the efficacy of this kind of treatment still needs to be tested clinically. In a clinical trial in AD patients, atomoxetine, a selective noradrenergic reuptake blocker, did not improve cognitive function [136]. However, as the relative NE levels and downstream pathological changes are varied based on disease stage and brain location, increased NE may result in microglial activation and inflammation priming if it occurs during certain phases of AD pathogenesis [137], and increased NE is even suspected to be the etiological factor of AD [138]. Addressing this issue will help to judge the feasibility of NE therapy during different stages of AD pathogenesis. Meanwhile, if AD patients have co-morbid chronic pain, it needs to be realized that LC-NE system dysfunction is likely the result of a combination of the two disease processes. Patients with AD have a decrease in noradrenergic neuron in the LC, although some studies suggest that the remaining noradrenergic neurons can compensate for this cell loss by increasing their activity and may even increase NE neurotransmission in the AD brain during certain stages of the disease [83, 138]. A dichotomy may emerge between AD patients with co-morbid chronic pain and those without: in the AD patient without chronic pain, there may be a relative deficiency of NE in brain regions, especially when compensatory mechanisms have been exhausted, and increasing NE supply in these brain areas may be beneficial; however, in AD patients with chronic pain, NE neurotransmission may be enhanced in brain areas such as the mPFC and hippocampus, which could possibly result in microglial pro-inflammatory activation [70]. Therefore, in AD patients with chronic pain, supplementation of NE via reuptake inhibitors may paradoxically exacerbate AD pathogenesis.

Another possible way to prevent and relieve AD is to directly inhibit pathological activation of microglia. Minocycline, a semisynthetic tetracycline derivative, is widely used to inhibit microglial activation [139]. In AD animals, minocycline reduced AD symptoms by reducing neuroinflammation, CNS pathology, and preventing cell death [140, 141]. For instance, in APP/PS mice, minocycline increased the survival of new dentate granule cells and improved behavioral performance in a hippocampus-dependent learning task [142]. In Tg2576 mice, minocycline attenuated deficits in learning and memory in Aβ-infused rats [140]. Clinical trials showed that minocycline is effective to treat chronic pain such as peripheral and autonomic neuropathies in type 2 diabetic patients [143], rheumatoid arthritis [144], and affective pain evaluated by McGill Pain Questionnaire in a cohort of patients with neuropathic pain, although the pain intensities did not change [145]. These data suggest that minocycline can reduce chronic pain and microglial activation in AD patient. However, the efficacy of minocycline in AD patients remains to be revealed, although one clinical trial is evaluating minocycline’s efficacy in patients with mild cognitive impairment or AD (NCT01463384).