The Locus Coeruleus (LC) is the main noradrenergic nucleus in the brain. It is involved in several neuropsychological functions and in the regulation of the sleep/wake cycle. Apart from these functional effects, noradrenaline (NA) released by LC terminals exerts a variety of effects. Among them, several studies in the last decades have shown a significant modulating role on different aspects involved in neuroinflammation, with a net anti-inflammatory effect of LC-NA. Neuroinflammation is considered to play a critical role in the pathogenesis of different neurological disorders, including Alzheimer’s disease (AD) and Parkinson’s disease (PD). In these two neurodegenerative diseases (NDDs), significant degeneration of LC, which starts years before the clinical onset of the diseases, has been well documented. Such LC degeneration might concur to the pathogenesis of these NDDs through, among others, a potentiation of neuroinflammation.
In this review, we will describe and discuss the role of LC in neuroinflammation, indeed with a special emphasis on its role in NDDs. In the first paragraphs, we will provide a brief overview of neuroinflammation and describe the neuroanatomical features of LC. In the following ones, we will focus on the available experimental data supporting a role of LC in neuroinflammation and, eventually, on the available evidence on the specific involvement of LC degeneration in neuroinflammatory phenomena occurring in AD and PD.
The term “neuroinflammation” defines the inflammatory steps taking place selectively within the central nervous system (CNS). Neuroinflammation is based on the interaction between different cell types, namely, the microglia, astrocytes, neurons, endothelium, and pericytes. All of them concur to the defense of the CNS from external noxae. Moreover, blood-derived circulating cells concur to neuroinflammation after getting into the CNS via the blood–brain barrier (BBB), which can be impaired primarily by the same noxae determining the neuroinflammation, or secondarily by mediators secreted by glial cells.
Despite representing by definition a mechanism of defense against external agents, neuroinflammation can concur to damaging the CNS through several mechanisms (see also [1
]). For instance, there is a large amount of evidence for a significant role of neuroinflammation in the pathogenesis of degenerative disorders of the CNS, such as AD and PD, as will be mentioned in more in detail in paragraph 5.
Stepping back to neuroinflammation components, microglia play a key role in its onset and maintenance. Microglia, similarly to macrophages, are of mesodermal origin, as opposed to astrocytes, which are of neuroectodermal origin. Upon activation by different types of molecules, microglia can get a pro-inflammatory (M1) or an anti-inflammatory phenotype (M2); these can occur at different stages and time points after a same pathological event [2
], and one of those can be prevalent upon the other one.
Microglia are activated by different types of ligands, mainly the pattern recognition receptors (PRRs) (which bind fragments of molecules related to pathogens), including Toll-like receptors (TLRs), as well as by other receptors such as CD-14,-33, -34, RAGE, or TREM-2 [3
]. The activation of these receptors induces microglial modifications, and the secretion of different molecules, including growth factors, cytokines, and chemokines. The latter can be either anti-inflammatory (e.g., TGF-β) or pro-inflammatory ones (e.g., TNF-α, IL-6, MP-1, IL-1β), in parallel with the M2 or M1 phenotype, respectively [4
Microglia also represent antigen-presenting cells (APCs), as they can express MHCII protein in specific circumstances, similarly to astrocytes. Indeed, it is worth noting that, even though in a very low number, T-lymphocytes can be found in brain parenchyma, and can participate in local immune response [5
Astrocytes participate in neuroinflammatory phenomena in several ways. In particular, when stimulated by cytokines, they change their phenotype and secrete different substances including nitric oxide, cytokines, metalloproteinases, and growth factors [6
]. Similar to microglia, they can also be directly activated by PRR.
One of the main characters of innate immunity is represented by the BBB integrity. It is worth noting that astrocytes also play a key role in this respect; not only do astrocyte end-feet surround the intraparenchymal vessel, forming the so-called glia limitans
, but they also secrete substances inducing the expression of tight junctions (TJs) in endothelial cells [7
]. An impairment of the BBB can allow the access of large molecules into the brain parenchyma, which can per se cause damage and contribute to triggering neuroinflammation [8
3. The Locus Coeruleus
The LC is the main noradrenergic nucleus of the brain. According to the classification of catecholaminergic nuclei by Dahlström and Fuxe, LC corresponds to the A6 nucleus, which is placed in the dorsomedial part of the lateral nuclei column of the reticular formation of the pons [9
]. The LC is also strictly connected with the so-called nucleus sub-coeruleus, which is placed ventrally and caudally to the main LC aggregate [10
]; thus, these nuclei are often considered as part of the same structure and, in this review, we will refer to LC including both of them. LC is a tube-shaped neuronal aggregate that is placed right below the floor of the fourth ventricle at the level of the pons and extends from the posterior commissure rostrally, up to the caudal border of the pons. It is formed by a number of neurons ranging from approximately 25,000 to 50,000 in humans. The main type of neurons of LC is NA neurons, which can be further divided into two sub-types based on their shape and size. In particular, the prevalent ones are medium-size (35–45 um body diameter) neurons, each possessing several dendrites and a large axon, while a smaller type of NA neurons possesses spindle-like soma (approximately 15–20 um diameter) from whose extremities two tufts of dendrites emerge [10
]. The smaller neurons are more represented in the sub-coeruleus component of LC, while the larger neurons are almost exclusively placed in the main component of LC [10
]. Each of the axons originating from the medium-size LC neurons can extend for up to several cm, and it can send collaterals reaching different parts of the brain [11
]. Furthermore, these axons are covered by varicosities, which represent structures from which NA can be released and affect neighboring targets through a “volume transmission”, i.e., a paracrine-type of neurotransmission [12
]. At the same time, the same axons also possess a specific synaptic boutons contributing to classical synapses. Thus, a single LC neuron can simultaneously affect different parts of the brain, and this is indeed one of the most important and specific features of LC neurons.
Finally, LC NA neurons can also express co-transmitters, such as galanin, which can exert modulatory effects in post-synaptic target neurons [15
]. Another typical feature of LC NA neurons is the intracellular accumulation within their cell bodies of neuromelanin (NM), a by-product of NA that can bind metal ions, and thus is considered to play a protective role towards neurons themselves, at least at the early stages of its accumulation [16
]. Neuromelanin is indeed the pigment that confers the “coeruleus” (i.e., cerulean, in latin) color to the LC; it accumulates within the LC during the whole life-span, up to reaching a plateau around 60 years of age, when almost all LC NA neurons contain NM [17
]; it is worth mentioning that its paramagnetic features have recently allowed the identification of LC in vivo in humans through specific magnetic resonance imaging sequences (see the review by [18
The LC receives afferents from a variety of structures. Interestingly, a hierarchical distribution of afferent fibers on LC neurons has been proposed, as projections from selected structures, including the prepositus hypoglossi
and paragigantocellular nuclei, end directly in the cell body of LC medium-size neurons [19
], while most afferents, originating from several, mainly sub-cortical, structures of the brain end in their distal dendrites [19
LC neurons send their efferents to many cortical and subcortical structures. In particular, all cortical regions receive fibers from the LC, and in the case of the limbic cortex, such an innervation is particularly dense [21
LC plays a key role in several physiological functions, such as attention, memory encoding, orientation to novelty, and the sleep/wake cycle [22
]; furthermore, it strongly modulates neuronal plastic mechanisms in physiological [23
] and pathological conditions [24
Apart from such functional effects, LC terminals also play an important role in regulating the integrity of the neurovascular unit [26
], and significantly modulate neuroinflammation, as will be reviewed in detail in the second part of this review.
4. Locus Coeruleus Degeneration in Alzheimer’s and Parkinson’s Disease
It is now well known that LC is significantly degenerated in Parkinson’s disease and Alzheimer’s disease. The earliest studies showing a marked neuronal loss in the LC of autopsies of patients with PD date back to the 1970s. Oleh Hornykiewicz himself, who is the discoverer of the occurrence of dopaminergic (DA) cell loss in the Substantia Nigra pars compacta (SNpc) in PD, described, in parallel with such SN alterations, an even more pronounced NE neuron loss in the LC in those same patients [27
]. In PD patients, the number of remaining LC neurons never overlapped with the number measured in any one of the neurologically intact subjects, thus Hornykiewicz himself postulated that, in PD, LC degeneration was even as pathognomonic as the nigrostriatal dopaminergic degeneration. Such an observation was further extended by the same group, and others, in the following years, as well as by profiting from more sophisticated/quantitative histological approaches [28
]. The neuronal alteration typical of PD and of other synucleinopathies is the occurrence, within the SNpc and in other brain regions, of the so-called Lewy bodies (LBs), which are neuronal inclusions formed by α-synuclein (α-syn) deposits, leading to frank cell death. In 2003, a seminal autoptic study by Braak et al. showed that LBs accumulate within the LC years before being observed at the level of SNpc [31
] and that LC neuronal loss also precedes the degenerative phenomena of the SNpc by years.
In light of the early observations quoted above, several authors tried to reproduce the LC lesion in animal models of PD (such as those in which SNpc dopaminergic (DA) loss is induced by 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine -MPTP-, or by substituted amphetamine administration, in rodents or primates) in order to evaluate whether LC degeneration might concur to PD pathogenesis, rather than being just an epiphenomenon. In the early 1990s, Colpaert’s group showed that LC lesion by 6-hydroxy-dopamine in primates and by the systemic administration of N-(2-chloroethyl)-N-ethyl-2-bromobenzylamine (DSP-4) in mice
could strongly potentiate nigrostriatal DA damage induced by MPTP [32
]. Fornai et al. significantly extended these findings by showing the following: (a) that LC lesion makes otherwise sub-toxic doses of methamphetamine toxic for DA SNpc neurons [34
] and significantly potentiates nigrostriatal loss induced by systemic methamphetamine administration in mice
]; (b) that the potentiating effects of LC lesion on DA damage in these rodent models of PD was not due to a change of MPTP/MPP+ or methamphetamine pharmacokinetics, or to an impairment of DA loss recovery, but rather to a direct potentiation of the neurotoxic effects/mechanisms of MPTP/methamphetamine themselves [34
]. The fact that such an effect of LC pre-lesion was obtained in different animal species [34
] and using different DA neurotoxins has been interpreted as a proof that the role of LC degeneration on the pathophysiology of nigrostriatal DA loss occurring in PD could be a sound phenomenon, which could also be extended to the human disease, according to the temporal sequence of events in which LC degeneration precedes DA loss [38
Concerning AD, several reports obtained in small casistics dating back to the early 1980s already showed a significant neuronal loss in the LC of patients at advanced disease stage. In particular, LC degeneration was analyzed by Tomlinson et al. [39
], Mann et al. [40
], and Bondareff et al. [42
], among others, and all of them showed a significant LC neuronal loss, which was proposed to more significantly affect the rostral extent of the nucleus [28
]. The abovementioned studies were performed, however, in small casistics of subjects, all of which were affected by severe dementia, and in which diagnostic criteria were significantly different from the current ones. Only very recently has such evidence eventually been confirmed and extended in a seminal paper [43
], in which the authors profited from large case series, from more detailed diagnostic criteria, and from stereological analysis of brain specimens, the latter providing a precise estimate of the absolute number of neurons in LC. In particular, Kelly et al., in 2017 analyzed post-mortem the brains of subjects who had been followed-up in detail in terms of neuropsychological and clinical features during their life [43
]. They showed the following: (a) in patients with mild cognitive impairment (MCI) due to AD (i.e., who bear the pathological features of AD, but have only isolated episodic memory impairment and lack any need of support for daily life activities, and will eventually develop dementia due to AD), there is already a significant neuronal loss in the LC; (b) such a neuronal loss is much higher in AD dementia patients; and (c) the number of NA LC neurons is directly related to the performances in several cognitive tasks and to the global cognitive score [43
As mentioned, in PD, the degeneration of LC is associated to the accumulation of inclusions of α-synuclein [31
]. Conversely, recent studies have clearly shown that, in the LC of AD patients, there is a massive accumulation of hyperphosphorylated Tau (p-Tau) [44
]. In particular, in 2011, Braak et al. [44
] analyzed hundreds of brains of subjects showing different degrees of AD pathology and showed that the progressive accumulation of p-Tau in LC precedes by years the occurrence of neurofibrillary tangles (NFT) deposits in the entorhinal cortex (which was classically considered, up to that study, as the first cortical site involved by Tau pathology in AD). These authors also showed that, in AD, such early involvement of LC is related mainly to the accumulation of p-Tau (also defined by the authors as “pre-tangles”) within LC neuron perykaria, and that this also progressively extends to LC terminals, and eventually leads to frank LC neuronal loss due to NFT formation.
The effects of an LC lesion in transgenic models of AD have been investigated by several authors. The lesion of LC by DSP-4 has been shown to exacerbate dramatically the deposition of amyloid and cognitive impairment in these AD models [45
]. Most of them also addressed the modulatory role of LC in the neuroinflammatory phenomena involved in amyloid plaques deposition; they will be described in detail below.
7. Discussion and Conclusions
The LC loss occurring in AD and PD, apart from contributing to some of the signs and symptoms occurring in the two NDDs (e.g., sleep/waking alterations, memory/executive function complaints, impairment of attention, and hypotensive states), is likely to contribute dramatically to the pathogenesis of the two disorders, as shown by several studies in experimental animals. An increase in neuroinflammation has been shown to play a potential key role in such a contribution, and this might be especially true concerning the potentiation of amyloid burden in AD. Several experimental models showed a specific strong role of LC in neuroinflammation through modulation of microglia and astroglia activity (summarized in Figure 1
). There is also evidence for a significant role of complement-mediated neuronal death in AD pathogenesis, but to the best of our knowledge, the role of LC on this phenomenon has not been explored in detail yet.
The fact that the role of LC in neuroinflammation has been confirmed in different models and by a variety of approaches (again, especially concerning AD) suggests that this phenomenon is likely to also occur in humans. However, experimental models obviously bear intrinsic limitations. For instance, one could not completely exclude that the direct neurotoxic effect of DSP-4, or other toxins used to induce LC lesion, may already contribute by itself, at least in part, to the neuroinflammation increase observed after LC lesion.
Thus, it appears mandatory to also obtain direct confirmations of the role of LC in neuroinflammation in PD/AD in patients in vivo. Potentially, nowadays, there are a variety of experimental tools that can be used in humans to assess neuroinflammation, as well as tools that allow estimating in vivo LC integrity. In particular, CSF analysis is currently performed in most subjects with suspect degenerative dementia, in order to assess the profile of amyloid/p-tau/total tau in AD/MCI patients, as well as in subjects affected by other NDDs; in this same biological matrix, the concentration of different types of inflammatory/anti-inflammatory cytokines can also be directly assessed, which allows an estimation of their concentration in the brain. Again, as mentioned in paragraph 5, nowadays, it is possible to directly estimate in vivo, in humans, the burden of microglia-related neuroinflammation by TSPO-ligand PET tracers. These approaches have shown increased neuroinflammation in the brain of PD and AD patients [57
], but they have not yet been put in relation with LC markers obtained in the same subjects.
In humans, LC features can be estimated non-invasively by specific MRI T1-wighted sequences and ad-hoc post-processing analysis. These MRI tools, the development of which started in the last decades, have been progressively refined, up to recent, more sophisticated approaches allowing to estimate not only the LC cell density, but also its volumetric features [131
]. By the same token, PET tracers specific for noradrenergic transporters have recently been developed [132
]. Finally, it has been possible to quite reliably directly assess the levels of amyloid and tau-related pathology in vivo in patients through PET or CSF analysis for almost 10 years, and these are nowadays assessed almost routinely in memory clinics to confirm AD phenotype in single patients [133
]. Thus, it is auspicable that, in the future, the link between the loss of LC integrity and neuroinflammatory burden, in relation to AD (and PD) pathology, could be tested directly in patients in vivo.
Apart from the direct assessment of such a pathogenetic link, early identification of the combination of these biomarkers (i.e., LC parameters, neuroinflammation, and amyloid/tau/syn parameters) in a patient might theoretically even allow to directly intervene at the early stages of NDD in order to slow its progression. More in detail, one might try to intervene (a) on the mechanisms causing LC degeneration and/or (b) to replace pharmacologically the noradrenergic tone, which is dramatically reduced in LC target structures after LC degeneration.
Concerning the first aspect, i.e., trying to intervene as soon as possible on the degenerative processes causing NA neuronal loss, it is worth noting that the molecular mechanisms involved in the degeneration of LC in PD and AD are likely significantly different from one another. In fact, LC degeneration in PD is mainly due to the accumulation of α-syn, up to LB formation [31
], while in AD, it degenerates after the progressive accumulation of p-Tau within NA cell bodies, up to frank accumulation of NFT within the axons and cell bodies [44
]. Unfortunately, the precise mechanisms through which both pathological processes occur are not clear yet (a detailed description of them is beyond the aim of this review), and thus it is unlikely that in the near future there will be promising therapeutic tools aimed at halting LC degeneration.
Conversely, a promising approach might be represented by replacing the impaired NA tone early in the course of NDD. In line with this, there are several pieces of evidence obtained in experimental models in vitro and in vivo, in which the effects of adrenergic agonists on neuroinflammation have been assessed, and some of them were obtained in the context neurodegenerative phenomena. NE induces the expression of several anti-inflammatory genes (including IL-10, heat shock protein 70, and PPARγ) in both neurons and glia [95
]. Furthermore, NE has been shown to decrease, through β-AR activation, the microglial activity of nuclear factor kappa-light-chain enhancer of activated B cells (NF-kB), which is involved in the transcription of pro-inflammatory molecules such as IL-8 and TNF- α [136
]. The latter is indeed reduced by NE in microglia [46
More specifically, in the context of AD pathogenesis, in rats pre-treated with DSP-4 and submitted to cortical microinfusion of Aβ1-42, it has been shown that the increased levels of neuronal iNOS and of IL1β expression in microglia were dramatically attenuated by co-injection of the β2-AR agonist isoproterenol and of NE [101
]. In APP
-transgenic mice, NE application has been shown to enhance microglial migration and Aβ clearance [46
]. In in vitro experiments in which microglia was exposed to Aβ, NE application prevented the Aβ-related production of chemokines and cytokines and the induction of pro-inflammatory genes such as TNF-α, iNOS, CCL-2
, and MCP-1
, and at the same time, β2-AR stimulation increased Aβ phagocytosis and Aβ-related microglial migration [46
Less clear evidence for a beneficial effect of AR stimulation on the neuroinflammatory phenomena occurring in PD models is available; however, recent exciting data on a potential beneficial effect on PD of β-AR agonists have been obtained based on data analysis of clinical records available for the entire Norwegian population by the Norwegian National Registry and the Norwegian Prescription Database [138
]. In this study, Mittal et al. retrospectively extrapolated a potential significant protective effect of treatment with β2-AR agonists on PD. Even though they also showed that β2-AR agonists reduce the expression of synuclein, in the same paper, the authors themselves also discussed the potential role of β2-AR protective effects in light of the anti-inflammatory role [138
Thus, there are at least some hints, concerning both NDDs, for a potential beneficial role of β2-AR agonists on the neuroinflammatory mechanisms involved in degenerative phenomena, and drugs with such mechanisms might be worth testing. Furthermore, one might also assess potential beneficial effects of an increase of NE in the brain; in line with this, it is worth mentioning the study by Gutierrez et al. [113
], who showed that reboxetine (a selective blocker of NA reuptake) was able to reduce neuroinflammation and neurodegeneration in the 5xFAD mouse model of AD. With this purpose, the administration of synthetic NA precursors, such as L-threo-3,4-dihydroxyphenylserine (which is able to selectively increase NA levels in the brain), might also represent an interesting therapeutic approach to be tested in the future [98
]. Finally, it is worth mentioning that VNS, which is a therapeutic tool already approved in humans for the treatment of specific types of epilepsy and severe depression [140
], is known to exert an indirect strong activating effect on LC [128
] and reduces glial activation and α-syn accumulation LC-lesioned animals [129
]. It is worth noting that there have been already proposals for the use of VNS in AD and PD patients as well [142
], and this might represent indeed a useful approach for an early NE-related modulation of neuroinflammation in these NDDs.