Aging in humans is associated with a reduction in total brain volume (Hedman et al., 2012), being the reduction of gray matter volume more pronounced than that of white matter (Ge et al., 2002). The annual rate of volume loss is about 0.2% after 35 years of age, which increases to 0.5% at age of 60, becoming over 0.5% after 60 (Hedman et al., 2012). Age-dependent cortical thinning, which is observed across most brain regions, is more accentuated in the frontal lobe (Raz et al., 1997; Salat et al., 2004). Although decline in hippocampus CA1 region, hilus, and subiculum, volumes appear to be a consequence of neuronal loss, in other regions, like the frontal cortex, the number of neurons are relatively preserved (August et al., 2022).

The brain undergoes a myriad of changes during aging that can lead to an atrophic aged brain. Neurons show dendrite retraction, especially in the prefrontal cortex and hippocampus. With increasing age, dendrites shrink, their branching become less complex, and lose spines (Dickstein et al., 2013), affecting inter-neuronal connections. There are also age-dependent changes in the synthesis of neurotransmitters. For instance, older people brains synthesize less dopamine, and have fewer receptors (Yin et al., 2014). Older adults with mild cognitive impairment appear to have less serotonin (Yin et al., 2014), which could potentiate memory loss. Functional effects are especially relevant for processes that require a high degree of synaptic plasticity. The number of synapses is reduced, which can affect learning and memory, facilitating cognitive decline. Although synaptic changes are selective and mild, they appear to affect cognitive decline beyond their changes in structure and neurotransmitters. The myelin wrapping of axons thins out (Bender et al., 2016), reducing the speed of propagation of action potentials, and impacting neuronal communication. Neurogenesis also declines with age (Li Puma et al., 2020); interestingly, in humans and rodents, physiological strategies that boost neurogenesis, such as regular exercise, can improve cognitive function (Ma et al., 2017). Studies on adult hippocampal neurogenesis in AD mouse models show a severe decline of neurogenesis before cognitive impairment (Li Puma et al., 2020).

Neurons are post-mitotic cells, arrested in the G0 phase of the cell cycle. Therefore, recognition of neuronal senescence cannot be performed on basis of proliferation arrest. A growing body of evidence indicates that cell senescence-like changes in neurons can be observed in aging brains. These characteristics involve shortening of telomeres (Ain et al., 2018), senescence-associated secretory phenotype (SASP), altered morphology and proteostasis, propensity to undergo apoptosis, autophagy impairment, accumulation of lipid droplets, increased activity of senescence-associated-b-galactosidase (SA-b-gal), and epigenetic alterations, including DNA methylation, chromatin remodeling, and histone post-translational modifications that affect gene expression (Di Micco et al., 2021; Sikora et al., 2021; Sahu et al., 2022).

Oligodendrocytes are the myelinating cells of the CNS, generated from oligodendrocyte progenitor cells (OPCs), also known as NG2 cells, because they express the proteoglycan, NG2 (Bradl and Lassmann, 2010; Kuhn et al., 2019). The density of NG2(+) cells in white matter is 1.5-fold higher than that in gray matter (Dawson et al., 2003) and their morphology in the mature CNS vary from region to region (Hughes et al., 2013). NG2 cells maintain their migrating, proliferating, and differentiating capacities throughout life to generate oligodendrocytes and regenerate myelin (Young et al., 2013). In the white matter of the adult mouse brain, NG2 cells are continuously generating mature, myelinating oligodendrocytes, whereas in the gray matter, they mostly generate postmitotic NG2 cells (Dimou et al., 2008). In adult mice, NG2 cells proliferate and differentiate in response to several stimuli such as voluntary physical activity in a brain region-dependent manner. Furthermore, the physical activity-associated cognitive improvement is abolished when NG2 cells differentiation is prevented (Eugenin von Bernhardi and Dimou, 2022).

In addition to increasing the speed of propagation of the action potential, oligodendrocytes provide lactate and pyruvate as metabolic support for neurons (Lee et al., 2012; Franklin and Ffrench-Constant, 2017; Rawji et al., 2023). These functions are lost with the loss of oligodendrocytes observed in demyelinating diseases like multiple sclerosis (MS) resulting in lack of trophic support, leading to slowing of axonal conduction of action potentials, and irreversible degeneration of demyelinated axons (Franklin and Ffrench-Constant, 2017).

The decline in white matter volume with normal aging (Taubert et al., 2020), is thought to contribute to age-dependent cognitive impairment. In aging rodent and nonhuman primates, electron microscopy revealed myelin disintegration and decompaction with accumulation of electron-dense cytoplasm within some sheaths, suggestive of myelin breakdown (Peters and Sethares, 2003; Pannese, 2011). Some myelinated nerve fibers in white matter degenerate and disappear, whereas others show degeneration restricted to myelin sheaths, but axons remain intact (Peters, 2009). Remyelination of these axons results in thinner myelin sheaths with shorter internodes than that observed in young fibers (Peters and Sethares, 2003). This remyelination depends on oligodendrogenesis in the spinal cord, where half of newly generated cells expressed NG2 (Lasiene et al., 2009). Multiphoton live imaging of the upper layers of the cortex in mice has confirmed the occurrence of degeneration and decrease of internodes with advancing age (Hill et al., 2018).

Single-cell RNA sequencing revealed age-associated transcriptomic changes in mice brains. Aging oligodendrocytes show downregulation of genes encoding myelin proteins (Mog, Plp, and Cnp) and cholesterol synthesis pathway (Hmgcs1). By contrast, aging oligodendrocytes upregulate genes involved in ribosome biogenesis (Rpl6, Rps29, and Rpl23a) and immune-response (C4b and Il33) (Ximerakis et al., 2019; Rawji et al., 2023).

Adult rodent NG2 cells progressively loss their differentiation and proliferation potential which results in the slowing of remyelination capacity with aging (Neumann et al., 2019). They fail to be recruited into the lesion area and they are not able to differentiate into oligodendrocytes (Sim et al., 2002). This loss of function with aging is relevant for understanding that in chronically demyelinated MS lesions there is a decrease of the number of NG2 cells and their differentiation into oligodendrocytes (Neumann et al., 2019). Furthermore, NG2 cells microenvironment stiffens with age, and this stiffness appears to be sufficient to cause age-related loss of function of NG2 cells (Segel et al., 2019). Bulk RNA sequencing analysis revealed that the decline in functional capacity is associated with hallmarks of stem cell aging, such as mitochondrial dysfunction, decreased metabolic function, inflammasome signaling, increased DNA damage, dysregulated nutrient sensing, autophagy, and the unfolded protein response (Neumann et al., 2019; Rawji et al., 2023). Proteomic analysis reveals that the amount of myelin-associated proteins, and proteins associated with oxidative phosphorylation, inflammatory responses and actin cytoskeletal structure are upregulated by aging (de la Fuente et al., 2020). By contrast, enzymes related with transcription factors, cell cycle proteins, and biosynthesis of cholesterol, essential for the production of myelin, are downregulated (de la Fuente et al., 2020). Interestingly, regenerative capacity of aged NG2 cells can be enhanced with youthful systemic milieu containing monocytes, fasting, metformin treatment, or reduction of stiffness of the extracellular matrix. All these stimuli will improve remyelination in aged animals (Neumann et al., 2019; Segel et al., 2019).

Astrocytes, not only support neurons, but serve also a wide range of functions that involve ion buffering, water and ion homeostasis, neurotransmitter recycling, formation, maturation, maintenance, pruning and remodeling of synapses, blood–brain barrier formation, inflammation regulation, and interoception. They are located in close contact with all nervous system structures to accomplish metabolic and homeostatic functions indispensable for the proper functioning of neuronal circuits in the brain. They extend processes that occupy non-overlapping domains, into the vicinity of synaptic clefts. They enwrap presynaptic and postsynaptic neuronal regions, forming the so called “tripartite synapse.” In response to synaptic activity (Gomez-Gonzalo et al., 2017) or specific stimuli like hypercapnic acidosis in the brainstem (Beltran-Castillo et al., 2017), they show a calcium dependent release of gliotransmitters (Glu, Gaba, D-serine, ATP) which regulate the excitability and the efficacy of synapses (Araque et al., 2014). Since one astrocyte can contact thousands of synapses, they could recruit and enhance the activity of distant neurons and synapses in brain circuits (Halassa and Haydon, 2010). Together with microglia, astrocytes contribute also to inflammatory processes (Giovannoni and Quintana, 2020).

Astrocyte phenotype changes in response to brain injury, ischemia, infection, neuroinflammation or neurodegenerative diseases, which leads to “reactive astrocytosis.” Reactive astrocytosis involves morphological changes such as hypertrophy, increased astrocyte proliferation, up- and down-regulation of several genes, and particularly, after acute CNS trauma, are associated with glial scar formation (Liddelow et al., 2017; Lopez-Teros et al., 2022). Reactive astrocytosis englobes different astrocyte responses depending on the nature of insults (Zamanian et al., 2012). For example, neuroinflammation and stroke ischemia in mice result in upregulation of genes, whereas 50% are upregulated by both insults, the rest of genes are specific for each injury type. In brain inflammation, reactive astrocytes upregulate complement cascade which are deleterious for synapses, and compatible with a neurotoxic phenotype. By contrast, in ischemia, reactive astrocytes upregulate many neurotrophic factors suggesting a neuroprotective phenotype (Zamanian et al., 2012; Liddelow et al., 2017). As observed in cytotoxic microglia,. inflammatory astrocytes lose several astrocyte properties (Kigerl et al., 2009; Zamanian et al., 2012; Liddelow et al., 2017) such as promotion of neuronal survival, outgrowth, synaptic functions, and phagocytosis. They can induce death of neurons and oligodendrocytes in vitro, and of axotomized neurons in vivo (Liddelow et al., 2017). Increased numbers of activated astrocytes are also observed in human neurodegenerative diseases (Liddelow et al., 2017).

As part of astrocyte/microglia interaction, microglia modulate astrocytes (Liddelow et al., 2017). In wild-type (WT) mice, LPS injection activates astrocytes, whereas in Csf1r−/− mice, which lack microglia cells, LPS fails to the activation of astrocytes (Liddelow et al., 2017). Furthermore, activated microglia cells induce differentiation of astrocytes by secreting interleukin 1α (IL1α), tumor necrosis factor α (TNFα), and complement 1q (C1q), which together are necessary and sufficient to induce the activation (Liddelow et al., 2017). It is worth noting that C1q is part of the first component of the C1 complex, which is bound by antigen/antibody complexes, neuronal blebs, apoptotic cells, fibrillary β amyloid (Aβ) or phospho-tau, to activate the classical pathway of complement. C1q appears to have a role of tagging weak synapses to be engulfed by microglia (Gomez-Arboledas et al., 2021).

Reactive astrocytes can lead to beneficial or detrimental effects. For instance, they can produce a glial scar as a defensive barrier against inflammatory cells or pathogens, promote neuronal survival, detect signals of brain damage, or secrete cytokines and chemokines. However, they also can inhibit cell migration and axonal regeneration (Escartin et al., 2021; Lopez-Teros et al., 2022). The astrocyte’s responses to brain aging are heterogenous. Proliferation-competent glial cells can undergo senescence both in vitro and in vivo, and contribute to neuroinflammation in the aging brain. Astrocytes can become reactive or senescent with aging, depending on stressful stimuli, contributing to the loss of cognitive function through inflammatory mediators (Cohen and Torres, 2019; Escartin et al., 2021; Lopez-Teros et al., 2022; Gaspar-Silva et al., 2023). Astrocyte senescence implies permanent cell cycle arrest, increased cell size, and several characteristics that involve a secretory profile called senescent associated secretory phenotype (SAPS) (Rodier and Campisi, 2011), the presence of DNA damage or “scars,” changes in heterochromatin called Senescence-Associated Heterochromatin Foci (SAHF) (Narita et al., 2003), increment of b-galactosidase enzyme activity (Dimri et al., 1995), lipofuscin accumulation, and a decrease in lamin B1 (Shimi et al., 2011).

In humans, astrocyte- and oligodendrocyte-specific genes, but not neuron-specific genes, show significant age-dependent changes in their regional expression patterns, particularly in the hippocampus and substantia nigra. By contrast, microglia- and endothelial-specific genes increase in all brain regions (Soreq et al., 2017). A hallmark of brain physiological aging, the increase in astrocyte reactivity, assessed through the increase in glial fibrillary acidic protein (GFAP) labeling, is mainly observed in the hippocampus, frontal, temporal, and entorhinal cortex in humans (Nichols et al., 1993; Porchet et al., 2003). Similar findings have been observed in aged mice and rats (Nichols et al., 1993; Rodriguez et al., 2014; Boisvert et al., 2018). In mice, astrocytes undergo various age-dependent morphological and molecular changes in specific brain regions (Boisvert et al., 2018). Such changes involve astrocyte hypertrophy, rearrangement of cytoskeleton, GFAP upregulation, and inflammatory phenotype. Aged astrocytes from the hypothalamus and cerebellum, but not from the cortex, show significant increase in the expression of genes associated with inflammatory response, astrocyte reactivity (GFAP and Serpin3n), synapse elimination pathways (complement C3 and C4b), and downregulation of cholesterol synthesis enzymes (Boisvert et al., 2018).

RNA sequencing (RNAseq) analysis of the differentially expressed astrocytes genes along the mouse lifespan revealed that in aged astrocytes there is a significant increase in the proportion of reactive astrocytes exhibiting a neuroinflammatory phenotype (Clarke et al., 2018) in a brain region-dependent phenomenon. Hippocampal and striatal astrocytes upregulate a higher number of reactive astrocyte genes than that observed in cortical astrocytes. Furthermore, the LPS-induced transformation of astrocytes is increased by aging. Likely, the aging effect may be associated to the 300-fold increase in C1q observed in aging mouse and human brains (Stephan et al., 2013) and the upregulation of inflammatory genes in aging microglia, in special because the aging-dependent upregulation of reactive astrocyte genes was significantly reduced in mice lacking microglia-secreted cytokines (IL1α, TNF, and C1q) (Clarke et al., 2018).

These results show that microglia play a role in astrocyte activation. Aged astrocytes contribute to create an environment favoring synapse elimination and neuronal damage, likely promoting aging-associated cognitive decline (Boisvert et al., 2018). In fact, astrocytes undergo age-dependent changes in gene expression that make specific brain regions more vulnerable to age-associated synapse loss and neuroinflammation. If astrocyte heterogeneity defines the susceptibility of different brain regions to certain insults, and the way brain regions age are still open questions.

In contrast to neurons and astrocytes, microglia have mesodermal origin. Microglia are derived from myeloid precursor cells from the yolk sac, which differentiate in tissue-resident macrophage precursors that migrate into the CNS where they finally become resident microglia at an early embryonic state.

The microglia, as “resident macrophages” (Rivest, 2009), constitute the main defense system of the CNS (Tremblay et al., 2011). Microglia are the main contributors to synaptic pruning in the brain, although astrocytes can participate in this process, particularly, during development and early postnatal life. In adult animals, failure in microglia regulation of pruning can exacerbate synaptic loss leading to memory deficits observed in neurodegenerative diseases (Wu et al., 2015; Hong et al., 2016; Liddelow et al., 2017; Gomez-Arboledas et al., 2021).

Microglia surveil systematically the nervous tissue and detect a wide spectrum of signals associated with autoimmune damage, infections, ischemia, trauma, and toxins (Rivest, 2009). Depending on the nature of these signals, they can trigger an integrative microglial response to maintain brain homeostasis, modifying their morphology and functional properties, phagocytizing, and degrading potentially harmful endogenous and exogenous compounds. Thus, surveillance microglia can adopt an activated phenotype, which can be pro- or anti-inflammatory. Depending on their phenotype, microglia are able to release a broad spectrum of molecules including inflammatory cytokines, such as interleukin 1β (IL1β), interleukin 6 (IL6), tumor necrosis factor α (TNFα), and interferon γ (IFNγ), reactive oxygen species (ROS) (Qin et al., 2005; Block et al., 2007; Li et al., 2007; Kettenmann et al., 2011; Welser-Alves and Milner, 2013; Hickman et al., 2018), and nitric oxide (NO) (von Bernhardi and Eugenin, 2004; Li et al., 2007; Nakajima et al., 2007; Sierra et al., 2007; Tichauer and von Bernhardi, 2012; Welser-Alves and Milner, 2013; von Bernhardi et al., 2015). Interestingly, microglia can modulate regulatory cytokines such as interleukin 10 (IL10) and transforming growth factor β (TGFβ), among other trophic factors. In addition to be the main source of inflammatory molecules, they also release gliotransmitters (Glu, ATP and D-serine) in the CNS (Imura et al., 2013; Beltran-Castillo et al., 2018; Dos-Santos-Pereira et al., 2018).

Microglia undergo senescence, showing characteristic morphological and functional features (Paolicelli et al., 2022), that differ from those found in activated or quiescent microglia (Hart et al., 2012; Spittau, 2017; Wendimu and Hooks, 2022). A feature of abnormal aged dystrophic microglia is the reduction or absence of cytoplasmic processes leading, occasionally, to the formation of spheroids, a striking sign of dystrophia with partial fragmentation of their cytoplasm (Mecca et al., 2018; Edler et al., 2021; Javanmehr et al., 2022; Munoz-Castro et al., 2022; Paolicelli et al., 2022; St-Pierre et al., 2022). Double immunofluorescence, against ionized calcium binding adaptor molecule 1(Iba1)/cluster of differentiation 68 (CD68) and Iba-1/major histocompatibility complex (MHC) class II to identify microglia, together with electron microscopy, revealed that isolated Iba-1(+) fragments persisted connected to each other by CD68(+) or MHCII(+) segments of the microglial process. That is, apparent fragments by light microscopy, indeed still are connected to the microglial soma when observed by electron microscopy (Tischer et al., 2016).

Since microglia is the main generator of inflammatory cytokines and oxidative mediators in the CNS (Pawate et al., 2004; Qin et al., 2005; Hayashi et al., 2008), changes in the regulation of aged microglia may lead microglial activation from neuroprotective to deleterious (von Bernhardi et al., 2015). Hence, microglial dysregulation arises as a key element for inducing chronic neuroinflammation (Tremblay et al., 2011). On the other hand, neuroinflammation promotes neurotoxicity by inducing cytotoxic activation of microglia (Ramirez et al., 2008), inhibition of Aβ clearance (von Bernhardi, 2007), and synergistic deleterious effects promoting neuronal death (Nguyen et al., 2002; von Bernhardi and Eugenin, 2012). Thus, as a global effect of aging, microglia switch from a neuroprotective to a more cytotoxic phenotype (Banati et al., 1993; von Bernhardi et al., 2010, 2015, 2019; Streit et al., 2021). Accordingly, microglia from older individuals present morphological evidence of activation compared with young ones (Streit et al., 2004; von Bernhardi, 2007; von Bernhardi et al., 2010, 2011), elevated basal levels of inflammatory cytokines, such as IL6 and IL1β (Ye and Johnson, 1999; Sierra et al., 2007), decreased signaling by suppressor of mothers against decapentaplegic 3 (Smad3)-TGFβ in inflammation (Tichauer and von Bernhardi, 2012), decreased Aβ-induced phagocytosis (Floden and Combs, 2011), and increased ROS production (Tichauer and von Bernhardi, 2012) and oxidative stress (von Bernhardi and Eugenin, 2012; Tichauer et al., 2014). This microglia activation has been described in the aging CNS (von Bernhardi, 2007; von Bernhardi et al., 2010; Villeda et al., 2014), as well as, in various pathologies, including cerebrovascular disease and AD (Hickman et al., 2008; Murgas et al., 2012; von Bernhardi and Eugenin, 2012; Trougakos, 2019). Changes in the expression of receptors relevant for cellular communication and inflammatory activation (Gu et al., 2019) underly microglial dysfunction (Tarkowski et al., 2002; Tse and Herrup, 2017).

Microglial toxicity is modulated by astrocytes through a cross-regulation that includes TGFβ and IL1β (Tichauer et al., 2007; Orellana et al., 2013), the neuroprotective response being especially conspicuous in acidic microenvironments (Uribe-San Martin et al., 2009). Hypercapnic acidification has also effects on phagocytosis (Eugenin et al., 2016) and synaptic function, inducing the release of D-serine by astrocytes (Beltran-Castillo et al., 2017, 2018), which could also participate in aging. Multiple changes associated with glial dysregulation in aging (von Bernhardi, 2007; Hickman et al., 2018) result in the impairment of neuronal function, including the production of soluble mediators by activated cells in aging (Conboy et al., 2005, 2013; Villeda et al., 2014), which affect synaptic function (Johnson-Venkatesh and Umemori, 2010; Schafer et al., 2013; Cerpa et al., 2016), and the regulation of cellular activation by TGFβ and mTOR, the mammalian target of rapamycin (Flores and von Bernhardi, 2012; Herrera-Molina et al., 2012; Switon et al., 2017).

The development of methodology to cultivate microglia from aged mice (von Bernhardi et al., 2011), allowed us to determine that Aβ phagocytosis declines in microglia obtained from adult and aged mice (Alarcon et al., 2005; Cornejo and von Bernhardi, 2013). The reduced uptake is related to the decreased expression of Scavenger Receptor A (SRA), which is relevant for Aβ phagocytosis and appears to mediate the enhancement of Aβ cytotoxicity (Murgas et al., 2012), and for shaping glial inflammatory activation by regulating the secretion of cytokines, ROS and reactive nitrogen species, and the activation of signaling pathways associated with inflammation (Godoy et al., 2012; Murgas et al., 2014). SRA expression is regulated by TGFβ-Smad signaling (Cornejo et al., 2018), which is altered in aging and AD (Tichauer et al., 2014).

Recently, it has been proposed that a subpopulation of microglia can have a protective effect in neurodegenerative diseases, in particular AD and MS (Keren-Shaul et al., 2017; Krasemann et al., 2017). Using single cell transcriptomics, Keren-Shaul et al. (2017) recognized a microglia sub-population, which was denominated disease associated microglia (DAM), also known as activated response microglia (ARM) or microglial neurodegenerative phenotype (MGnD) (Keren-Shaul et al., 2017; Sobue et al., 2023). DAM microglia were found near Aβ plaques in human and mice brains showing intracellular phagocytosed Aβ, suggesting their capacity to restrict Aβ plaque formation by degrading Aβ (Keren-Shaul et al., 2017). Physiologically, DAM activation is achieved via a 2-step process. The first step is a triggering receptor expressed on myeloid cells 2 (TREM2)- independent process, in which microglia transits toward to a stage 1 DAM, characterized by a reduced expression of homeostatic microglia checkpoint genes such as Cx3cr1 and P2ry12/P2ry12, and upregulation of B2m, and AD-associated genes such as Tyrobp and Apoe (Xu et al., 2022). The second step is a TREM2-dependent process, in which stage 1 DAM cells are transformed into stage 2 DAM cells, characterized by the upregulation of Cst7, Lpl, and Trem2 genes (Xu et al., 2022). It is thought that the inhibition of some of these microglia specific inhibitory checkpoints (such as Cx3cr1), which disinhibits DAM activation, could be an important therapeutic target (Keren-Shaul et al., 2017). Another study identified a similar subpopulation of microglia, named neurodegeneration-associated microglia (MGnD) controlled by TREM2-APOE pathway and post-transcriptionally regulated by microRNA(miR)-155 in the surrounding of plaques (Krasemann et al., 2017). TREM2 activates APOE pathway, which transforms a homeostatic into a neurodegenerative microglia phenotype after phagocytosis of apoptotic neurons. Interestingly, targeting the TREM2-APOE pathway can restore the homeostatic phenotype of microglia in amyotrophic lateral sclerosis (ALS) and AD mice models, and prevented neuronal loss in an acute neurodegeneration model (Krasemann et al., 2017).

In murine model of ALS, in which human superoxide dismutase 1 (SOD1) with the G93A mutation is expressed under control of the cistronic human SOD1 promotor, the lack of expression of CX₃CR1 accelerates ALS-like disease progression leading to a rapid and increased neuronal cell death (Boillee et al., 2006; Cardona et al., 2006; Liu et al., 2019). In this murine model of ALS, the absence of CX₃CR1, increases the activation of NFκB, and impairs the autophagy-lysosome degradation pathway and the autophagosome maturation (Liu et al., 2019) resulting in an intense damage of motoneurons. Thus, the CX₃CL1/CX₃CR1 pathway has anti-inflammatory and neuroprotective effects and appears to play an important role in maintaining autophagy activity.

The Cx3cr1 human polymorphism I249/M280 is present in about 20% of the population. The protein product of this polymorphism exhibits reduced affinity for CX₃CL1 resulting in a reduced regulatory effect on microglia. Accordingly, the hypofunctional variant of CX₃CR1 has been associated with a reduction of the life span of ALS patients (Lopez-Lopez et al., 2014). Thus, ALS patients, who carry one or two copies of the CX3CR1-Val249Ile allele, have a rapid course of the disease and a shorter survival than patients who are carriers of the WT Cx3cr1 (Lopez-Lopez et al., 2014).

MS is a chronic, inflammatory, predominantly immune-mediated disorder of the CNS showing focal lesions in white matter of the brain and spinal cord, characterized by prominent demyelination ensuing axonal damage and glial scar formation, leading to loss of motor and sensory function (Karussis, 2014; Oh et al., 2018; Dobson and Giovannoni, 2019; Absinta et al., 2020). In MS patients, Cx3cr1-Val249Ile polymorphisms revealed that Cx3cr1 Ile249 Thr280 haplotype could endow a protective effect by impairing the switch of MS from the relapsing–remitting type (RRMS) into the secondary progressive type (SPMS) (Stojkovic et al., 2012; Arli et al., 2013). Patients having the variant in both alleles (homozygosity) have a higher risk for disability (Arli et al., 2013).

Current immune therapy is useful for ameliorating RRMS, but not for preventing its progressive forms (Oh et al., 2018; Dobson and Giovannoni, 2019; Absinta et al., 2020). Among several processes, remyelination is required for normalization of neural functions (Kotter et al., 2011; Lampron et al., 2015; Rivest, 2015; Mendiola et al., 2022). Unfortunately, remyelination in MS fails or is incomplete (Oh et al., 2018; Dobson and Giovannoni, 2019).

Chronic microglia activation is related with MS disease progression (Absinta et al., 2020). Besides, a switch in microglia from a pro-inflammatory to a regulatory phenotype is associated with the beginning of remyelination (Miron et al., 2013). The regulatory phenotype favors oligodendrocyte differentiation and is observed in injured CNS regions of aged mice in which remyelination is enhanced, like in MS lesions (Miron et al., 2013).

It has been suggested that CX₃CL1/CX₃CR1 axis is a relevant regulator of the clearance of myelin debris by microglia (Lampron et al., 2015; Rivest, 2015; Mendiola et al., 2022). Experimental autoimmune-independent demyelination without BBB disruption followed by complete remyelination can be attained with cuprizone, a copper chelating toxin that induces apoptosis of oligodendrocytes and can be administered through the diet (Lampron et al., 2015; Rivest, 2015; Mendiola et al., 2022). In WT mice, cuprizone induces massive demyelination as consequence of oligodendrocytes apoptosis during the first 3 weeks (wk.) accompanied by recruitment of NG2 cells, astrogliosis, and microgliosis (Hiremath et al., 1998; Gudi et al., 2014). Cuprizone removal from diet is followed by complete remyelination within 1–3 wk. In CX₃CR1^−/− mice, inflammatory response was like that observed in cuprizone-challenged WT mice. However, cuprizone-induced demyelination was followed by uncomplete remyelination. Microglia in CX₃CR1^−/− mice exhibited an impaired migration into the corpus callosum and phagocytosis, and consequently, persistent myelin debris and defective axonal remyelination characterized by aberrant myelin patterns. Unlike WT microglia that showed abundant phagocytic inclusions, CX₃CR1^−/− microglia lack them (Lampron et al., 2015). Induction of experimental autoimmune encephalomyelitis (EAE) on a mouse model of CX₃CR1 hypofunction by replacement of the normal mo Cx3cr1 locus for the hu Cx3cr1-I249/M280 variant, revealed exacerbated functional signs of EAE, with more severe inflammation and neuronal loss (Cardona et al., 2018).

Transgenic mice expressing the polymorphic hu Cx3cr1-I249/M280 variant, and fractalkine-deficient (Cx3cl1^−/−) mice showed exacerbated cuprizone-induced demyelination in the anterior corpus callosum 4 w, compared with that observed in CX₃CR1^−/− deficient (Cx3cr1^GFP/GFP) and WT mice (Mendiola et al., 2022). Microgliosis and CD68 (phagocytic microglia marker) were similar in hu Cx3cr1-I249/M280, Cx3cl1^−/−, and Cx3cr1^GFP/GFP mice, but higher than those in WT mice after cuprizone treatment (Mendiola et al., 2022). Notoriously, after 1 wk. removal of cuprizone, only Cx3cr1^GFP/GFP and WT mice showed significant remyelination (Mendiola et al., 2022). WT mice showed significant remyelination at the anterior and posterior corpus callosum. In hu Cx3cr1-I249/M280 mice, significant remyelination was only observed in the anterior corpus callosum, whereas in Cx3cr1^GFP/GFP mice, only in the posterior corpus callosum. By contrast, Cx3cl1^−/−mice did not show significant early remyelination (Mendiola et al., 2022).

Using EAE as a rat model of MS, it was shown that 12 days after the inoculation of myelin basic protein (MBP), CX₃CL1 and CX₃CR1 mRNA and protein levels were increased in dorsal root ganglia (DRG) and spinal cord (Zhu et al., 2013). The increase was associated with thermal sensory abnormalities, suggestive of neuropathic pain, and correlated with neurological impairment (Zhu et al., 2013). These results are indicative of the role of CX₃CL1/CX₃CR1 axis as a critical pathway involved in the MS-induced neuropathy.

Studies in humans of the genotype distribution of CX₃CL1 variants and their association with disease, also show a potential link between CX₃CR1-Thr280Met, a variant associated with cell migration deficit, and a higher risk of human age-dependent macular degeneration (AMD) (Tuo et al., 2004; Chan et al., 2005; Combadiere et al., 2007). Evidence supporting a role for CX₃CR1-Val249Ile variant in AMD is controversial (Combadiere et al., 2007). Homozygosity for the Thr280Met allele is a more consistent finding in AMD than of other diseases (Combadiere et al., 2007). In the macula of AMD, photoreceptors show signs of degeneration, retinal pigment epithelium is disrupted and cells expressing CX3CR1 are found in the outer retina, in the subretinal space, in the perivascular vicinity, and in choroidal neovascularization. In addition, deposits of CX₃CR1 were found in drusen spots (Combadiere et al., 2007). Because they show a reduced affinity for CX₃CL1, deletion of CX₃CR1 has been experimentally used to simulate the main effect of these variants. Interestingly, aged albino mice with Cx3cr1 deletion exhibit, in regions of retinal degeneration, accumulation of microglia in the subretinal space associated with drusen-like yellowish-white dots, and increased choroidal neovascularization, histological findings like those found in patients with AMD (Combadiere et al., 2007).

Retinitis pigmentosa is the more common non-syndromic inherited retinal dystrophy that include several heterogeneous retinal neurodegenerative conditions in which mutations in photoreceptor or retinal pigment epithelium genes result in progressive degeneration of photoreceptors (Hartong et al., 2006; Murro et al., 2023). Retinitis pigmentosa is characterized by a primary degeneration of rod photoreceptors that progresses to the loss of cone photoreceptors, followed by an aberrant remodeling of retina resulting in disconnection of neural retina from photoreceptors, associated with changes from the molecular to tissue levels (Hartong et al., 2006; Zieger et al., 2014; Murro et al., 2023).

Retinal degeneration 10 (rd10) is a mouse model of autosomal recessive retinitis pigmentosa, consisting in the spontaneous mutation of the rod-phosphodiesterase-6b (Pde6b) gene, that leads to rod degeneration and later to cone degeneration (Chang et al., 2002). Analysis of mRNA and protein levels of CX₃CL1 revealed that in both WT and rd10 mice, a 100 kDa membrane bound CX₃CL1 form and a cleaved soluble 85 kDa CX₃CL1 form were present at the postnatal day 5 (P5). At P10, a 95 kDa form is accumulated, whereas the 85-kDa form is decreased. In older animals, the 95 kDa form became principal in wt retina, whereas in rd10 retinas, there is a significant increase of soluble 85 kDa form. Retinas of rd10 mice had significantly lower levels of total CX₃CL1 protein (from P10 onwards) and lower CX₃CL1 mRNA levels (from P14) than those observed in WT animals. In situ hybridization histochemistry and immunofluorescence using transgenic Cx3cl1cherry mice showed that neurons of the inner retina layers were the main sites of fractalkine synthesis in WT and rd10 mice (Zieger et al., 2014). In the rd10 mouse model, CX₃CL1 was detected in apoptotic photoreceptors, suggesting a potential role of this chemokine on the recruitment of microglia into the outer nuclear layer of the retina (Makabe et al., 2020).

Microglia have been implicated in many degenerative eye disorders, including retinitis pigmentosa (Wang and Cepko, 2022), in which activated microglia phagocytose death photoreceptors and their debris and release pro-inflammatory factors that contribute to retinal degeneration (Zabel et al., 2016). In fact, in CX₃CR1 deficient (Cx3cr1^GFP/GFP) rd10 mice, microglial infiltration into the photoreceptor layer was greater, and the photoreceptor atrophy and apoptosis were higher than those observed in heterozygous Cx3cr1^GFP/+ rd10 littermates. Furthermore, CX₃CR1 deficient microglia showed increased phagocytosis and increased expression of inflammatory cytokines associated with the increase of activation markers. Activation of CX₃CL1/CX₃CR1 axis in the rd10 retina via exogenous intravitreal delivery of recombinant CX₃CL1 decreased microglial infiltration, phagocytosis, and activation together with reduction of morphological damage and improvement of photoreceptors functionality (Zabel et al., 2016).

Furthermore, synthetic progestin ‘norgestrel’ has been proven as a neuroprotective agent in the retinitis pigmentosa, likely throughout the increase of growth factors, such as basic fibroblast growth factor (bFGF) and leukemia inhibitory factor (LIF) in the retina, favoring the upregulation of pro-survival and downregulation of apoptotic pathways and reduction of microglial pro-inflammatory activity (Doonan et al., 2011; Roche et al., 2016). In primary cultures, rd10 microglia promote neuronal cell death. Norgestrel, in contrast, reduce pro-inflammatory activation and prevent neuronal cell death, reducing the expression of cytokine, chemokine, and danger-associated molecular pattern molecule (DAMP) in the rd10 retina. Furthermore, norgestrel upregulates CX₃CL1/CX₃CR1 signaling in the rd10 mouse, 1,000-fold at the RNA level (Roche et al., 2016). Norgestrel’s neuroprotection would be mediated by its actions on photoreceptors, which are induced to release CX₃CL1, and this chemokine in turn, would refrain harmful microglia activity (Roche et al., 2017).

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder and the main cause of dementia. Clinical symptoms include memory loss, learning difficulties, and problem-solving impairment. The main risk factor is aging; approximately 5–8% individuals older than 65 are affected, increasing to 35–50% in adults older than 85. The prevalence of AD for women is higher than in men by 19%. Risk factors associated with development of AD involve genetic predisposition (familial early-onset forms), allele ApoE-4 for apolipoprotein E, age, sedentarism, hypertension, diabetes, and metabolic syndrome, among others.

AD is characterized by the extracellular accumulation of Aβ_1-42 (senile plaques) and intraneuronal aggregates as neurofibrillary tangles (NFTs) of hyperphosphorylated microtubule associated protein tau (MAPT) (Jellinger, 2020). The time course of AD progression indicates that Aβ deposition begins around 25 years before the onset of clinical diagnoses of AD, and is followed by neurofibrillary tangles (NFT) formation (Saito and Saido, 2018; Jellinger, 2020; Yin et al., 2021). Aβ is generated from the amyloid precursor protein (APP) cleavage by a two-step process: the β-site APP cleaving enzyme 1 (BACE1, or β-secretase), cleaves APP releasing a globular extracellular soluble protein and the membrane-anchored C-terminal fragment; later, γ-secretase (Presenilin) subsequently cleaves this fragment to excise Aβ (a peptide mostly of 39–43 amino acids). In addition to the protein aggregates, AD is characterized by the activation of microglia and astrocytes, neuroinflammation, and the progressive degeneration of neurons, with a predominant early damage of the entorhinal cortex and the hippocampus.

The participation of the CX₃CL1/CX₃CR1 axis on the course of AD differ in different AD mouse models. In APP/PS1 mice, a well-known mouse model for AD, the level of CX₃CR1 does not differ, but the level of CX₃CL1 is slightly reduced compared with WT littermates. In APP/PS1 mice, Cx3cr1 deletion reduces plaques, either because an increased clearance or a reduced deposition of Aβ plaques. However, rTg4510 mice, an AD model with inducible overexpression of human mutant tau (P301L), show a 5-fold increase in CX₃CR1 and a slight increase in CX₃CL1 levels compared with the WT littermates (Nash et al., 2013). In the hu Tau model for AD, Cx3cr1^−/− accelerates the onset of tauopathy and behavioral deficits (Bhaskar et al., 2010; Lee et al., 2010; Cho et al., 2011). Furthermore, the increased presence of sCX₃CL1 ameliorates the severity of tauopathy in rTg4510 mice. Three months before rTg4510 mice show significant neuron loss, mice were injected with recombinant adeno associated virus (rAAV), expressing sCX3CL1, into both hippocampi. Three months after viral injection, 6 months old mice expressing sCX₃CL1 show a significant decrease in tau pathology, an increased density of NeuN(+) cells, reduced hippocampus volume loss, and reduced microglia compared with the control condition (Nash et al., 2013). Memory and learning performance of WT and rTg4510 mice were assessed in the radial arm water maze. The injection of rAAV coding for sCX₃CL1 did not significantly improve the behavioral deficit observed in the rTg4510 mice compared with non-transgenic mice (Nash et al., 2013).

To reveal the effect of CX₃CL1/CX₃CR1 axis on Aβ-induced toxicity, neuron–microglia primary cultures were exposed to hu Aβ_1-42. WT neurons released CX₃CL1 in response to hu Aβ_1-42. Administration of 2 μM hu Aβ_1-42 in microglia-depleted Cx3cr1^−/− mixed cortical and hippocampal cultures produced lower release of LDH revealing decreased cell death than that observed in WT cultures (Dworzak et al., 2015). Furthermore, Aβ_1–42 treatment impaired synaptic transmission, decreasing α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)-dependent miniature excitatory postsynaptic currents (mEPSCs), suggesting both a pre- and post-synaptic compromise (Dworzak et al., 2015).

Conversely, cell loss in the purified microglia cell culture from Cx3cr1^−/− mice were higher than the loss observed in WT microglia cultures. The opposite cytotoxic response induced by Aβ observed in Cx3cr1^−/− microglia and neurons suggest that CX₃CR1 endows each cell type with different effector functions.

Combining Cx3cr1 deletion, a triple-transgenic mouse model of AD (3xTg-AD: PS1M146V Kin, transgenic APPSwe and tauP301L), and two-photon microscopy, it was possible to evaluate simultaneously neuronal death and microglial cell behavior during a 28-d imaging in living mice (Fuhrmann et al., 2010). A significant 2% neuronal loss in cortical layer III was observed in 4-6-month-old 3xTg-Cx3cr1^+/− mice. By contrast, Cx3cr1^+/−, Cx3cr1^−/− or 3xTg-Cx3cr1^−/− mice, did not exhibit neuron loss over the same imaging period (Fuhrmann et al., 2010). Thus, Cx3cr1 deletion appears to be neuroprotective. Additionally, in 3xTg-Cx3cr1^+/− mice, microglia, showing an increased migration, were recruited to the sites of neuron loss before the death of neurons. Interestingly, microglia Aβ-phagocytosing activity was not modified by Cx3cr1 deletion (Fuhrmann et al., 2010). This suggests that expression of CX₃CR1 by microglia is required for neuron elimination in the context of an AD model (Fuhrmann et al., 2010), but the cellular mechanisms underlying such neuron–microglia interaction remain unknown.

Deletion of Cx3cr1 also appears to be neuroprotective in other AD mice models. For instance, in APP/PS1 and R1.40 mice, APP/PS1 mice co-express hu APP carrying the K670M/N671L familial AD mutation and presenilin 1 with the L166P familial AD mutation leading to primarily Aβ_1-42 oligomer aggregation. The R1.40 transgene is a full genomic copy of hu APP carrying the K670M/N671L familial AD mutation that leads to gradual deposition of Aβ_1-40 oligomer. In 4-month-old APP/PS1 mice and in 20-24-month-old R1.40 mice, Cx3cr⁻ deletion reduced Aβ deposition and augmented microglia accumulation around Aβ deposits. In both models, Cx3cr1 deletion reduced immunodetection of CD68(+) microglia, whereas qRT-PCR revealed that CX₃CR1 deficiency reduced TNFα and the chemokine (C-C motif) ligand 2 (CCL2) mRNA levels, and increased IL1β mRNA. In addition, in both AD models, in vitro and in vivo microglial phagocytosis assessment revealed enhanced Aβ uptake (Lee et al., 2010), suggestive of enhanced Aβ clearance and reduced inflammation (Lee et al., 2010).

In the brain of APP/PS1 mice heterozygous for Cx3cr1 (APP/PS1- Cx3cr1^+/−), Aβ level and senile-like plaque deposition are reduced in comparison with age-matched APP/PS1 mice. Reduced Aβ level in the brain was associated with high level of the neuronal-expressed Aβ-degrading enzymes (insulysin and matrix metalloproteinase 9) and with improved performance in the Barnes Maze cognitive test (Hickman et al., 2019). Barnes maze assesses memory and spatial learning placing a rodent on a circular arena, where, moved by its instinctive aversion to open spaces and its natural preference for dark and sheltered spaces, mice select 1 out of 20 equally distributed holes on the arena periphery, which is the unique hole equipped with the animal’s home cage.

PD is the second most common neurodegenerative disease after AD, affecting about 1% of the population older than 60 years (de Lau and Breteler, 2006). PD is characterized by the gradual loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc), the remaining SNpc neurons can show intracellular protein aggregates called Lewy bodies, which are mainly constituted by α-synuclein (SNCA). The cellular pathogenesis of PD is associated with mitochondrial dysfunction, oxidative stress, neuroinflammation, proteasomal dysfunction, impaired autophagy, increased protein aggregation and enhanced apoptosis (Shan et al., 2011; Angelopoulou et al., 2020). PD patients exhibit a constellation of motor and non-motor signs and symptoms. Rest tremor, rigidity, bradykinesia, and loss of postural reflexes are considered as cardinal signs (Jankovic, 2008). Motor signs also include, among others, shuffling gait with decreased arm swing, difficulty arising from chair, turning in bed, hypomimia affecting facial gestural expression and communication, dysarthria, dysphagia, micrography, glabellar reflex, blepharospasm, dystonia, and sialorrhea. Non-motor signs and symptoms include cognitive impairment, bradyphrenia, tip-of-the-tongue (word finding) phenomenon, depression, apathy, anhedonia, fatigue, other behavioral and psychiatric problems including dementia, anosmia, ageusia, pain (shoulder, back), paresthesia, dysautonomia (orthostatic hypotension, constipation, urinary and sexual dysfunction, abnormal sweating, seborrhea), weight loss, REM sleep disorders, sleep fragmentation, and restless legs syndrome (Jankovic, 2008). Current therapies provide symptomatic relief, but fail to stop disease progression (Angelopoulou et al., 2020).

Administration of 1-methyl-4-phenyl- 1,2,3,6-tetrahydropyridine (MPTP), or its active metabolite, 1-methyl-4-phenylpyridinium (MPP+), has been extensively used to model PD in non-human primates and rodents. In mice receiving IP MPTP injection, Cx3cl1 or Cx3cr1 deletion worsened the dopaminergic neuronal loss and increased microglial activation in the SNpc (Cardona et al., 2006).

The effects induced by IP injection of MPTP varies on different mice strains. For instance, IP injection of MPTP in C57BL/6 mice, but not into BALB/c (BALB) mice, induced behavioral dysfunction, activated microglia/astrocytes, and increase of IL10, IL12, IL13, IFNγ, and CCL2 in cerebrospinal fluid (CSF). It has been proposed that differences in the immunological background affect the outcome. B6 mice show a bias toward Th1 acquired immune responses, whereas BALB mice toward Th2, which are characterized by their distinct cytokine profiles (Yasuda et al., 2008). IL2, IFNγ and TNFα are Th1 cytokines and IL4, IL5, IL6, IL10, and IL13 is Th2 cytokines. Activation of Th1 promotes cell-mediated immunity and activation of Th2 promotes humoral immunity (Yasuda et al., 2008).

Unilateral injection of MPP+ into the rat SN increase the expression of CX₃CL1, and CX₃CR1 in the ipsilateral (injected) but not in the contralateral SN (Shan et al., 2011). In contrast, both proteins are weakly expressed in sham animals. On the other hand, unilateral injection of CX₃CL1 into the rat SN increased microglial activation, dopaminergic neuron loss and motor dysfunction. After 14 days of MPP+ treatment, almost all CX₃CR1(+) cells in the injected SN expressed CD11b, also called integrin alpha-M, ITGAM, or integrin alpha-X, ITGAX, is commonly used to identify macrophages and microglia and showed an activated microglia-like phenotype. Furthermore, daily ICV injection of anti-CX₃CR1 neutralizing antibodies for a week after injection of MPP+ reduced, in a dose-dependent way, the MPP + -induced rotation behavior. In addition, microglia activation, dopaminergic degeneration, and CX₃CL1-induced behavior abnormalities induced by the injection of CX₃CL1 into the SN were prevented by ICV administration of anti-CX₃CR1 neutralizing antibody 6 and 1 h before CX₃CL1 injection (Shan et al., 2011). Moreover, ICV administration of minocycline, a selective microglia inhibitor, prevented CX₃CL1-induced rotation behavior and reduced dopaminergic neuron loss in a dose-dependent way (Shan et al., 2011), supporting the involvement of microglia in the dopaminergic toxicity induced by MPP+ and CX₃CL1.

Mice with a Cx3cr1 deletion have reduced dopamine levels in the striatum, independently of the neurotoxin administration. After 7 days of intranasal inoculation of MPTP, mice of various genotypes (Cx3cr1^+/+, Cx3cr1^+/−, Cx3cr1^−/−) showed a reduction in tyrosine hydroxylase (TH)- and dopamine transporter (DAT)-positive cells in the striatum. Whereas, in the SN, Cx3cr1^+/+ and Cx3cr1^+/− genotypes showed a reduction of TH-positive cells, but in Cx3cr1^−/− mice, no change of TH- or DAT-positive cells after intranasal MPTP inoculation was observed (Tristao et al., 2016).

The effects of Cx3cr1 deletion on MPPT-treated mice do not depend only on the brain region analyzed (SN vs. striatum), but also depend on the via of MPPT administration (Tristao et al., 2016). After 3 and 7 days of intranasal MPPT, Cx3cr1 deletion did not affect microglial activation and astrogliosis in the striatum. Although, Cx3cr1 deletion at day 3 did not affect the loss of dopamine neurons in the striatum, it showed neuroprotective effects on day 7. In contrast, after intraperitoneal MPTP treatment, Cx3cr1 deletion did not affect dopaminergic degeneration or astrogliosis but increased microglial activation in the striatum and SN (Tristao et al., 2016).

Another model of PD consists in the unilateral stereotaxic administration of the catecholaminergic neurotoxin 6-hydroxydopamine (6 − OHDA), in the rat striatum or SN. 6 − OHDA does not cross the BBB and induces dopaminergic degeneration essentially by increasing ROS and inflammation (Blesa et al., 2012). The administration of recombinant CX₃CL1 via an osmotic minipump for 28 days starting 7 days after the 6-OHDA injection into the same striatum, prevented activation of microglia, and showed neuroprotective effect, reducing the loss of striatum neurons that resulted in the reduction in volume of the striatum. Application of CX₃CL1 also reduced dopaminergic cell loss and inflammation in the SN (Pabon et al., 2011).

Transgenic PD models have been developed by overexpression of mutant genes for autosomal dominant genes such as α-synuclein and leucine rich repeat kinase 2 (LRRK2) and KO or Kdown models for autosomal recessive genes, such as Parkin, DJ-1, phosphatase and tensin homolog (PTEN)-induced novel kinase 1 (PINK1) (Dawson et al., 2010; Blesa et al., 2012; Angelopoulou et al., 2020). Increase of α-synuclein promotes neuroinflammation, astrocytosis, microgliosis, activation of microglia, and NFκB activation, leading to an increased release of inflammatory cytokines, promoting neuroinflammation and worsening dopaminergic neurodegeneration (Angelopoulou et al., 2020). Microglia activation results in an increased phagocytosis of α-synuclein, which exacerbates the production of ROS and inflammatory activation (Theodore et al., 2008; Angelopoulou et al., 2020).

CX₃CL1 has diverse effects on neuroinflammation, and degeneration in α-synuclein models of PD (Angelopoulou et al., 2020). CX₃CL1 can be neuroprotective, inhibiting motor impairment and dopaminergic cell loss in the SN and striatum of rats injected with rAAVs coding for overexpressed hu α-synuclein (Nash et al., 2015). Overexpression of hu α-synuclein or hu α-synuclein A53T in WT (Cx3cr1^+/+) or null (Cx3cr1^−/−) mice display very low levels of CX₃CL1. Interestingly, overexpression of hu α-synuclein induced microgliosis, neuroinflammation, and dopaminergic neuronal death in SN, which are of similar magnitude in Cx3cr1^+/+ or Cx3cr1^−/− mice. However, the overexpression of hu α-synuclein A53T resulted in an exacerbated neurodegeneration compared to hu α-synuclein overexpression. Moreover, α-synuclein A53T-induced neurodegeneration was enhanced in Cx3cr1^−/− mice (Castro-Sanchez et al., 2018).

The native 95 kDa membrane-bound CX₃CL1 (mCX₃CL1) can be transformed into a 65 kDa chemotactic glycoprotein (sCX₃CL1, which contains the extracellular N-terminal CKD and MS domains), by metalloproteinases such as the TNFα converting enzyme (TACE, ADAM17) (Garton et al., 2001; Tsou et al., 2001), ADAM10 (α-secretase) (Hundhausen et al., 2003, 2007), matrix metalloproteinase 2 (MMP-2) (Bourd-Boittin et al., 2009) or the Cathepsin S (CatS) produced by microglia (Clark et al., 2009). CX₃CL1 is also a substrate for β-secretase (BACE1), and γ-secretase, involved in the APP cleavage to generate Aβ (Fan et al., 2019).

Age-dependent changes in CX₃CL1 shedding can be the result of changes in the amount of protease available to process mCX₃CL1, as consequence of changes in their synthesis or degradation. Alternatively, age-dependent changes in the enzymatic efficiency of proteases may be caused by changes in the amount of endogenous inhibitors or enzyme enhancers (Postina, 2012; Mathews and Levy, 2016; Shi et al., 2018; Sharma et al., 2022).

ADAM10 and ADAM17 are the best characterized members of the disintegrin and metalloprotease (ADAM) family (Black and White, 1998) involved in the processing of membrane-associated proteins such cytokine receptors, chemokines, adhesion molecules, and growth factors. Their targets include TNFα, TGFβ, Notch, CX₃CL1, APP, the low-density lipoprotein receptor-related protein 1 (LRP1), and the anti-aging protein α-klotho (Blobel, 1997; Peschon et al., 1998; Borrell-Pages et al., 2003; Liu et al., 2009; Chen et al., 2014; Brifault et al., 2017).

The enzymes, ADAM 10, and ADAM 17 are implicated in the α-secretase non-amyloidogenic APP processing, leading to the production of soluble APP (APPsα), which exhibits properties of neuroprotection, memory-enhancer, and regulation of neuronal excitability, synaptogenesis, and synaptic plasticity (Turner et al., 2003; Reinhard et al., 2005; Zheng and Koo, 2006). Mass spectrometric analysis revealed that 23-month-old rats showed a reduction of soluble APPsα in cisternal CSF compared with that observed in 3- and 13-month-old animals (Anderson et al., 1999). The reduction of APPsα has a positive correlation (r = 0.52–0.57, p < 0.001) with the performance in spatial memory tasks observed in young and aged rats (Anderson et al., 1999). Accordingly, ADAM 17 activity in cortex and hippocampus changes with aging (Bertoldi et al., 2017). Aged rats show a reduced ADAM 17 activity, whereas the amyloidogenic BACE activity is like that in young rats (Bertoldi et al., 2017). This age-dependent reduction in ADAM 17 activity is accompanied with a reduced performance in aversive memory test (Bertoldi et al., 2017). By contrast, daily moderate treadmill exercise for two weeks had not impact on ADAM 17 and BACE activities at various ages (Bertoldi et al., 2017).

CatS is a member of the family of cysteine lysosomal proteases preferentially expressed in macrophages and microglia and released from them in response to neurotrophic factors and inflammatory mediators. It plays a role, together with ADAM 17 and ADAM10, in CX₃CL1 shedding (Petanceska et al., 1996; Liuzzo et al., 1999a,b; Turk et al., 2000; Nakanishi, 2003; Wendt et al., 2008).

Age-dependent changes in the protein expression of CatS has been detected by Western Blot in cerebral cortex, cerebellum, brainstem, and spinal cord. In 1-wk-old mice, only pro- CatS is found. Starting at 6 months, age-dependent upregulation of prepro-form, pro-form, and mature form of CatS is evidenced (Wendt et al., 2008). Furthermore, analysis of gene expression by DNA microarrays reveals that CatS gene expression was higher in 25-month-old mice than that observed in 5-month-old mice (Park et al., 2009).

Double immunofluorescent labeling revealed that most CatS positive cells in the CNS are microglia (PT66), followed by astrocytes (GFAP) and a few neurons (Neu) or endothelial cells (von Willebrand factor). Aging upregulates the number and the labeling intensity of CatS positive cells. The number of glia-like cells as well as neurons expressing CatS in cortex and brainstem clearly increases in the aged mouse brain (Wendt et al., 2008).

Age-dependent changes in the activity of CatS are not only due to the increased, but also to age-dependent changes of the level of the cysteine-protease inhibitor cystatin C (CysC, also known as γ-trace), an endogenous inhibitor of CatS, a member of the endogenous cysteine-protease inhibitors (Barrett et al., 1986; Mathews and Levy, 2016), which is found in neurons, astrocytes, endothelial, and microglia cells (Yasuhara et al., 1993; Palm et al., 1995; Miyake et al., 1996). Besides of inhibiting CatS, CysC also inhibits the cysteine-proteases cathepsin (Cat) B, Cat H, Cat K, and Cat L (Turk et al., 2008). In turn, CysC is targeted and inactivated by aspartyl-protease Cat D and the serine-protease elastase (Lenarcic et al., 1991; Abrahamson et al., 1991a,b).

Hu CysC is a 120 aa protein which is preceded by a26 aa amino terminal that endows it with the capacity of being secreted (Turk and Bode, 1991; Turk et al., 1997). The novo synthesized CysC can access the endosomal-lysosomal compartment via an intracellular pathway or most of it is secreted to the extracellular compartment (Zucker-Franklin et al., 1987; Chapman et al., 1990; Barka et al., 1992; Tavera et al., 1992; Paraoan et al., 2001) and, only a fraction remains within exosomes (Ghidoni et al., 2011). Secreted CysC can be found in cell surfaces or in the extracellular matrix (Calkins and Sloane, 1995; Sastre et al., 2004; Kolodziejczyk et al., 2010), where it can inhibit the activity of released cathepsins, or alternatively, it can be internalized into other cells (Merz et al., 1997; Ekstrom et al., 2008). CysC is internalized via endocytosis, allowing its access into the endosomal-lysosomal compartment (Merz et al., 1997; Pierre and Mellman, 1998; Ekstrom et al., 2008) where it can inhibit lysosomal cathepsins and be degraded by aspartyl-protease Cat D (Wallin et al., 2013).

CysC secretion appears to be age-dependent since CysC levels in serum correlates with age in dogs, cats, and humans (Ohara et al., 2012; Ghys et al., 2015; Iwasa et al., 2022). Changes of CysC brain expression and secretion are found in several neurological disorders, and also correlates with animal models of neurodegeneration (Mathews and Levy, 2016). However, up to date, conclusive studies on age-dependent changes in CysC expression are lacking.