Neuroinflammation in neurodegenerative disorders: the roles of microglia and astrocytes

With the increase in life expectancy, the global socioeconomic impact of neurodegenerative diseases, including Alzheimer’s disease (AD), Parkinson’s disease (PD), and amyotrophic lateral sclerosis (ALS), is increasing considerably [ 1 ]. However, the pathological mechanisms underlying neurodegenerative diseases are not fully understood. Several factors including genetic, environmental, and endogenous factors are involved. Abnormal protein dynamics, oxidative stress with reactive oxygen species, mitochondrial dysfunction, DNA damage, dysfunction of neurotrophins, and neuroinflammatory processes are considered to be common pathophysiological mechanisms [ 2 ]. Neuroinflammation is a defense mechanism that initially protects the brain by removing or inhibiting diverse pathogens [ 3 ]. This inflammatory response can have beneficial effects by promoting tissue repair and removing cellular debris. Sustained inflammatory responses, however, are detrimental, and they inhibit regeneration [ 4 , 5 ]. Inflammatory stimulation can persist due to endogenous (e.g., genetic mutation and protein aggregation) or environmental (e.g., infection, trauma, and drugs) factors [ 6 , 7 ]. The persistent inflammatory responses involve microglia and astrocytes and can lead to neurodegenerative diseases [ 4 ]. Two categories of cells populate the central nervous system: neurons and glial cells [ 8 ]. Glial cells do not produce electrical impulses, and they were considered as supporting cells for neurons. It has been revealed that glial cells are superior to neurons in cellular diversity and function [ 9 ]. Glial cells, including astrocytes, oligodendrocytes, and microglia, can regulate neuronal activity [ 8 , 10 ]. Microglia and astrocytes serve diverse functions including innate immune responses in the brain. Traditionally, both can be classified into two opposing phenotypes: neurotoxic and neuroprotective. Microglia are divided into the M1 (classical activation) and M2 (alternative activation) phenotypes based on their activation status [ 6 , 11 ]. Similar to the microglia, astrocytes can produce pro-inflammatory or immunoregulatory mediators according to the phenotype of the polarization status [ 7 ]. However, microglia and astrocytes are considered to have multiple reactive phenotypes related to the type and stage of neurodegenerative diseases and the regional location [ 12 – 14 ]. Furthermore, the changes in phenotypes of microglia and astrocytes, their loss of neuroprotective functions, and their gain of neurotoxic functions are complicated and may differ with the stage and severity of neurodegenerative diseases. Therefore, the simple dichotomized classification cannot reflect the various phenotypes of microglia and astrocytes [ 12 ]. For these reasons, the use of the M1/M2 and A1/A2 nomenclature was limited in this manuscript, appearing only directly from the references where they were used. However, they should be considered as being on a spectrum, rather than being two distinct populations. This complexity could be the reason why trials of anti-inflammatory drugs have, to date, failed to show significant therapeutic effects. Here, we review the roles of inflammatory responses in neurodegenerative diseases, such as AD, PD, and ALS, focusing on the roles of microglia and astrocytes and their relationships. Recommendations for the success of clinical trials are also made. In addition, biomarkers to measure neuroinflammation and studies on drugs that can modulate neuroinflammation are also discussed. Microglia Microglia are ubiquitously distributed in the brain and are the principal innate immune cells and the first responders to pathological insults [ 15 , 16 ]. The proportion of microglia ranges 5–12% of the total cell population in the mouse brain depending on the location, and they have diverse morphologies: compact round, longitudinally branched, and radially branched [ 17 ]. They are involved in homeostasis and host defense mechanisms by participating in three essential functions [ 18 ]. The first function is detecting changes in their environment using their sensomes, which are encoded by various genes [ 19 ]. The second is the physiological housekeeping function, which includes migrating to injured sites, remodeling synapses, and maintaining myelin homeostasis [ 18 , 20 ]. The third is protecting against injurious stimuli, including pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs). Cellular receptors such as toll-like receptors (TLRs), nuclear oligomerization domain-like receptors, and viral receptors are expressed on microglia, and can recognize PAMPs and DAMPs [ 6 , 7 ]. In response to such stimuli, microglia produce proinflammatory cytokines, such as tumor necrosis factor (TNF)-α, interleukin (IL)-1β, IL-16 and chemokines, including the C-C motif chemokine ligand 2 (CCL2) and IL-18, to recruit additional cells and remove pathological agents [ 6 , 18 ]. However, although neu

AD is the most common form of dementia and is pathologically characterized by extracellular accumulation of amyloid-beta (Aβ)-containing plaques and development of intracellular neurofibrillary tangles composed of hyperphosphorylated tau protein [ 16 , 59 ]. In addition, neuroinflammation contributes to the pathogenesis of AD [ 16 ], as inflammatory responses have been repeatedly demonstrated in AD. For example, researchers have found higher TNF-α (pro-inflammatory cytokine) and lower TNF-β (anti-inflammatory cytokine) levels in the CSF of mild cognitive impairment patients who progressed to AD, compared with the controls who did not progress to AD [ 60 ]. Some cytokines including IL-1β, IL-6, and TNF-α have slowly increased levels from the early stage of the disease, while the levels of other cytokines including IL-18, MCP-1, and IP-10 can peak at a certain stage of the disease [ 61 ]. Although current publications have inconsistent results, it has been clear that neuroinflammation occurs early in AD, and may trigger the progression of this disease. Morphological changes of microglia and astrocytes surrounding the senile plaques are also indicative of the neuroinflammatory response [ 6 ]. Both microglia and astrocytes interact with Aβ. Dysfunctions of microglia and astrocytic metabolism can result in the accumulation of Aβ [ 18 , 62 ]. Aβ in turn activates microglia and astrocytes through TLRs to release neuroinflammatory mediators that promote neurodegeneration [ 4 , 6 ]. Microglia can be neuroprotective by degrading and removing Aβ and tau [ 63 , 64 ]. However, the persistent interaction between Aβ and Aβ-induced pro-inflammatory cytokines overwhelms the clearance ability of microglia [ 18 ]. Increases in the size and number of Aβ plaques during the late-onset form of AD may reflect the decreased clearance ability of microglia [ 65 ]. Microglia that surround the Aβ plaques are generally of the neuroprotective phenotype, labeled as Ym1, at the beginning of Aβ pathology; but they later switch to the neurotoxic (pro-inflammatory) phenotype during the advanced stage of the disease [ 28 , 66 ]. The pro-inflammatory cytokines decrease the phagocytic activity of microglia, and they are also likely to transform microglia into the pro-inflammatory phenotypes. In addition, the pro-inflammatory microglia increase the phosphorylation of tau and exacerbate tau pathology [ 67 ]. This indicates that microglia in AD are involved in the Aβ and tau pathologies. In a recent study, microglial activation, as measured by 11 C-(R)PK11195, decreased longitudinally in patients with mild cognitive impairment and increased longitudinally in AD patients [ 68 ], suggesting that two peaks of microglial activation may be present in AD. Although the 11 C-(R)PK11195 cannot differentiate between the phenotypes of microglial activation, the first and last peaks may be neuroprotective and pro-inflammatory, respectively, as microglia are known to change from the neuroprotective to the pro-inflammatory activation phenotypes during aging [ 68 , 69 ]. In addition, the microglial activation is significantly correlated with amyloid deposition (measured by 11 C-PIB PET) [ 68 ], but not significantly correlated with tau accumulation (measured by 18 F-AV-1451 PET) in AD [ 23 ]. This difference can be explained by the stage of the disease and the regional location in the brain. CSF sTREM2, a marker of microglial activation, is also increased in AD patients compared with the healthy controls [ 25 ]. However, several studies on sTREM2 have contradictory findings and the discriminative power of sTREM2 for AD is low. Further studies are required to establish the utility of sTREM2 in clinical practice [ 70 ]. Pro-inflammatory reactive astrocyte phenotypes have been linked to synaptic degeneration and glutamate dysregulation [ 13 , 41 ]. Knockout of astrocytic glutamate transporters EAAT1 (glutamate/aspartate transporter, GLAST) and EAAT2 (glutamate transporter-1, GLT-1) caused excitotoxicity and synaptic hyperexcitability in an AD model [ 71 – 73 ]. In the AD mouse model, calcineurin (Ca 2+ /calmodulin-dependent phosphatase), a nuclear factor of the activated T-4 cell signaling pathway, has been found to link between astrocyte activation and hyperexcitability during AD [ 72 ]. The astrocytes alter their function and morphology during AD, and may have different functions as AD progresses [ 43 ]. Most clinical trials for anti-inflammatory drugs in AD patients, including aspirin, prednisone, naproxen, diclofenac, indomethacin, and celecoxib, have failed to show definite improvements, even revealing some detrimental effects [ 74 – 79 ], although some have shown modest beneficial results. A subgroup analysis showed that mild-to-moderate AD patients who were APOE ε4 carriers benefitted from ibuprofen over the 12-month trial duration [ 80 ]. Another study revealed protective effects of indomethacin in mild-to-moderate AD patients over 6 months, but the patient

ALS, also called Lou Gehrig’s disease, is an adult-onset progressive neurodegenerative disease in which motor neurons are selectively affected [ 110 ]. The etiology of most ALS patients remains unidentified. Only less than 10% of cases are due to mutations of specific genes, including superoxidase dismutase 1 ( SOD1 ), C9orf72 , TDP43 , and FUS [ 18 ]. Neuroinflammation is a pathological mechanism common to ALS patients with and without genetic mutations, which is characterized by the infiltration of activated microglia and astrocytes. The activated microglia and astrocytes that produce pro-inflammatory cytokines are upregulated in post-mortem tissues of ALS patients [ 41 , 111 , 112 ]. A PET study has demonstrated increases in activated microglia [ 11 C-(R)PK11195 PET] and astrocytes ( 11 C-DED PET) in living ALS patients [ 113 , 114 ]. In addition, the CSF sTREM2 level is significantly higher in sporadic ALS patients with varying disease severity than controls [ 115 ]. In particular, the CSF sTREM2 level is highest in the early-stage ALS, and in late stage, higher levels of CSF sTREM2 are associated with slower disease progression [ 115 ]. Prolonged high levels of CSF sTREM2 may be indicative of a neuroprotective phenotype. Toxicity caused by mutant SOD1 , the most common form of inherited ALS, is mediated by direct damage that is incurred within the motor neurons, microglia, and astrocytes [ 110 ]. The activated pro-inflammatory microglia and astrocytes produce toxic factors that cause the initial damage and disease progression. The G930A- SOD1 transgenic mouse model has demonstrated the ability of microglia to switch from the neuroprotective to the pro-inflammatory phenotype from the onset of the pathology [ 28 , 116 ]. The SOD1 -mutant microglia isolated from mice with early-stage ALS express higher levels of M2 microglia phenotype markers and lower levels of pro-inflammatory microglia markers, compared with the SOD1 -mutant microglia isolated from mice with end-stage ALS [ 117 ]. Altogether, as ALS progresses, the function of neuroprotective microglia may decrease and the proportion of pro-inflammatory phenotypes may increase. The C3-expressing pro-inflammatory astrocytes and astrocytic NLRP3 inflammasomes have been found to be upregulated in post-mortem ALS patients [ 41 , 46 , 118 ]. The activation of astrocytes in ALS decreases their protective effects and increases their detrimental effects [ 119 ]. Astrocytes with SOD1 mutations have been reported to release soluble factors toxic to motor neurons [ 120 ]. IL-1α, TNF-α, and C1q released from microglia drive astrocytes to the neurotoxic phenotype, while reducing reactive astrocytes by inhibiting these factors attenuates the disease progression in the G93A- SOD1 mouse model [ 121 ]. However, little is known about the neuroprotective phenotype of astrocytes in the pathogenesis of ALS. The ablation of NOX3 or NF-kB improved motor neuron survival in the G930A- SOD1 transgenic mouse model [ 28 , 122 ]. In addition, the administration of minocycline in G930A- SOD1 transgenic mice selectively attenuated the expression of markers for the pro-inflammatory microglia, inhibited the upregulation of NF-κB in the primary culture of microglia, and delayed the pathogenesis [ 28 , 123 ]. Cromolyn, which induced the activation of neuroprotective microglia in the AD mouse model, demonstrated a neuroprotective effect in the G93A- SOD1 transgenic mouse model by delaying the disease onset and reducing the motor impairment [ 124 ]. Recently, masitinib, an oral tyrosine kinase inhibitor, has shown beneficial effects in ALS patients over 48 weeks [ 125 ]. Masitinib reduces the microgliosis and the emergence of aberrant glial cells in the G93A- SOD1 transgenic mouse model [ 126 ]. Regulatory T-lymphocytes (Tregs) can augment IL-4 expression, induce the M2-phenotype microglia, and delay the progression of the disease [ 127 ]. Upregulation of Tregs can be achieved using dimethyl fumarate (Tecfidera), and a phase II trial of Tecfidera is being conducted in patients with sporadic ALS [ 128 ]. In addition, the infusion of tocilizumab in ALS patients reduced neuroinflammation [ 129 ], and it is now in a phase II clinical trial in ALS patients ( NCT02469896 ), results of which are expected to be announced soon.