The central nervous system (CNS) has long been considered an immune-privileged compartment due to the presence of the blood-brain barrier (BBB). The BBB is a multicellular structure composed of endothelial cells, pericytes, astrocytes, neurons, and microglial cells, which collectively act as a barrier to prevent the undesired infiltration of peripheral immune cells into the cerebral parenchyma [1], [2]. However, the immune-privileged status of the CNS is not absolute, as various physiological and pathological factors can alter BBB permeability [2]. Among the physiologically relevant mediators capable of modulating BBB integrity are bradykinin, histamine, thrombin, and a broad spectrum of pro-inflammatory cytokines [3]. Neuroinflammation is an inflammatory response within the CNS, driven by molecules secreted by astrocytes, microglial cells, perivascular cells, and, in later stages, by immune cells recruited from the peripheral circulation [3], [4]. In this inflammatory state, the excessive production of pro-inflammatory cytokines, reactive oxygen species (ROS), and second messengers—such as nitric oxide (NO) and prostaglandins—pushes the CNS into a state of detrimental neurodegenerative stress [5]. Regardless of its origin—whether infectious, neoplastic, autoimmune, or neurodegenerative (including genetic) diseases—neuroinflammation is a common denominator across these conditions [6]. In this context, the migration of immune cells to inflamed sites [1], [3], is a crucial process orchestrated by chemokines (CKs) and chemokine receptors (CKRs), which generate and respond to chemotactic signals [7]. This review focuses on the CXCL10 (C-X-C motif chemokine ligand 10)–CXCR3 (C-X-C motif chemokine receptor 3) axis, given its central role in regulating cell migration during neuroinflammation. Additionally, we will examine current experimental and therapeutic strategies targeting the CXCL10-CXCR3 axis, with the goal of slowing or halting neurodegenerative conditions affecting the CNS.

The cDNA of human CXCR3 was partially cloned from a CD4⁺ T lymphocyte library in 1996, with one of its first characterized features being its inducibility in response to IL-2 (Interleukin-2) [8]. Two years later, the same receptor was identified in mice, cloned from a thymocyte library (αβTCR⁺, but CD4⁻ and CD8⁻) [9]. The CXCR3 gene is located on the X chromosome in both humans and mice. In humans, it is specifically mapped to the Xq13.1 region, while in mice, its orthologous gene is found at D|X 44.58 cM [10]. Structurally, the human CXCR3 gene consists of three exons, with a single intron between the first two exons [11]. The encoded protein is a G-protein-coupled receptor (GPCR) with seven transmembrane domains, a total length of 368 amino acids (aa), and a molecular weight of 45.65 kDa [11], [12]. Within the GPCR superfamily, CXCR3 belongs to Class A (the rhodopsin-like subfamily) [13]. Additionally, it is also referred to as GPR9 (G protein-coupled receptor 9) or CD183 [14]. While mice possess only a single isoform of this receptor, humans express two splicing variants: CXCR3-B and CXCR3-alt [11]. In 2003, the CXCR3-B isoform was first identified, leading to the renaming of the originally characterized transcript as CXCR3-A [15] The CXCR3-B isoform arises from an alternative splicing process that includes exon 2, whereas CXCR3-A excludes it. This occurs due to the utilization of a different splicing acceptor site, located 233 bp upstream of the one used for CXCR3-A [15]. As a result, CXCR3-B features an N-terminal extracellular domain that is 52 amino acids longer than that of CXCR3-A [15]. One year after the discovery of CXCR3-B, the CXCR3-alt variant was identified. Unlike CXCR3-B, this variant results from a post-translational exon skipping mechanism [16]. The post-transcriptional processing of CXCR3-alt causes a downstream frameshift, maintaining sequence homology with CXCR3-A up to glutamine at position 209 (Q209). However, due to the deletion of bases 696–1032, CXCR3-alt is a truncated variant that possesses only five transmembrane domains instead of the typical seven [16]. As a consequence of alternative splicing, the stability of the CXCR3-alt transcript is significantly compromised. However, this does not lead to a loss of function, as CXCR3-alt has been shown to play a crucial role in inducing chemotaxis in response to CXCL11 (C-X-C motif chemokine ligand 11 [16].

In addition to all T-lymphocyte subpopulations (from T-naïve to memory clones [8], including Th17 cells [17], γδT cells [18], and Tregs [19]), CXCR3 is also largely expressed on Natural Killer (NK) cells [8] and B cells [20]. There is a strong correlation between CXCR3 expression and differentiation of T lymphocytes into Th1, cytotoxic phenotypes. This correlation is so profound that the master regulator of this process, t-bet (T-box expressed in T-cells), also acts as a transactivator of the CXCR3 gene [21]. In the immune compartment, neutrophils [22] and eosinophils [23] also express the receptor in conditions of inflammation, as well as dendritic cells (DCs) and macrophages [24]. Outside of the immune system, CXCR3 expression has been detected on cultured neurons [25], astrocytes and Purkinje cells [26], endothelial cells, smooth muscle and Kupffer cells [24]. Furthermore, CXCR3-B expression has been observed in other tissues and organs, including the heart, skeletal muscles, liver and kidneys. This widespread expression indicates that CXCR3 signalling may have diverse functions beyond its role in immune responses, potentially influencing tissue development and organ homeostasis [15].

At the molecular level, the signal transduction pathways of CXCR3 isoforms differs significantly. Downstream signalling via CXCR3-A relies on the activation of a PTX (Pertussis toxin)-sensible Gαi (Inhibitory guanine nucleotide binding protein). It leads to the engagement of cascade-activated kinases, starting from Src (Src Proto-oncogene), then following with RAS (Rat Sarcoma Virus) and MAPK (Mitogen-activated protein kinase) cascade, culminating in the phosphorylation and activation of MEK (also known as Mitogen-activated protein kinase kinase, or MAP2K) and ERK (Extracellular signal-regulated kinase) proteins. RAS is also capable of contextually activating the PI3K (Phosphatidylinositol-3-kinase) - Akt (Ak strain transforming) pathway [27], [28], [29], [30]. CXCR3 is also able to trigger the activation of PLC-β (Phospholipase C beta) via a Gαq (Guanine nucleotide binding protein q). PLC-β, through PKC (Protein kinase C), mediates the proteolytic cleavage of PIP2 (Phosphatidylinositol-2-phosphate) into DAG (Diacylglycerol) and IP3 (Inositol trisphosphate). IP3, in turn, induces an increase in intracellular calcium (Ca2+) concentration, released by endoplasmic reticulum (ER) [31]. Signalling via the CXCR3-A isoform is usually associated with cell proliferation and survival, the promotion of angiogenesis, and a notable reduction in the rate of apoptosis. In contrast, PTX-insensitive downstream signalling via CXCR3-B relies on the activation of Gαs (Stimulating guanine nucleotide binding protein), which is – following activation of the adenylate cyclase – responsible for an increase in intracellular cAMP (Cyclic adenosine monophosphate) levels and PKA (Phosphokinase A) activity [15]. Additionally, in response to its ligands, CXCR3-B can trigger p53-independent upregulation of the cell cycle inhibitor p21Cip/WAF1. It inhibits the phosphorylation of pRB (Retinoblastoma protein) catalysed by the complex Cdk2 (Cyclin dependent kinase 2) and Cyclin-E, hindering the G1-S transition [15], [32]. Notably, CXCR3-B is also able to interact with CXCL4 (C-X-C motif chemokine ligand 4), a strong angiostatic factor [32]. Altogether, signalling via the CXCR3-B isoform is usually associated with a physiological control mechanism that favours cell cycle arrest, apoptosis and angiostasis [15], [31] [Fig. 1].

In addition to CXCL4, the three main CXCR3 ligands are CXCL9 (C-X-C motif chemokine ligand 9), CXCL10 and CXCL11. In humans, the genes encoding these three CKs are located on chromosome 4, which harbours the largest cluster of inflammatory CKs in the human genome [33]. In mice, this cluster is located on chromosome 5 [33]. The first CK capable of interacting with CXCR3 was CXCL10, discovered in 1985. It was cloned from the IFNγ (Interferon-gamma) stimulated U937 monocytic cell line and subsequently named IP-10 (Interferon-gamma induced protein 10) [5]. Thereafter, a large plethora of cells were identified as capable of producing CXCL10 after exposure to IFN (α, β or γ) or LPS (Lipopolysaccharide), including keratinocytes, fibroblasts, astrocytes, endothelial and mesenchymal cells, neutrophils, activated T cells [34] and neurons [35]. This clearly suggests a widespread expression of this chemokine [6], [34]. CXCL9, originally known as MIG (Monokine Induced by Interferon-gamma), was first identified in 1990 by Farber and colleagues [2]. It was discovered through a cDNA library generated from murine RAW 264.7 macrophages cultured in medium enriched with lymphokines produced by mitogen-stimulated splenocytes [2]. Three years later, the human gene was identified in a cDNA library from human THP-1 monocytic cells stimulated with IFNγ [3]. In 1998, CXCL11 was first identified and is recognized as the ligand with the highest affinity for CXCR3 [4]. Originally named as ITAC-1 (IFN-inducible T-cell alpha chemoattractant), the CXCL11 gene was isolated from a cDNA library obtained from human astrocytes stimulated with IFNγ [4]. Interestingly, in 2006, it was discovered that CXCL11 can bind to a second receptor, CXCR7 (C-X-C motif chemokine receptor 7), which is highly expressed on tumour cells, and plays a role in transmitting survival and proliferation signals [36]. Notably, C57BL/6 mice are incapable of synthesizing CXCL11 due to a deletion within the coding region of its gene. The main sources for CXCL9 and CXCL11 production are astrocytes [4], monocytes, APCs (Antigen presenting cells), endothelial cells and fibroblasts, which cooperate to generate a chemotactic gradient that favours the recruitment of inflammatory cells [37].

The dynamics of ligand-receptor interactions are tightly regulated by structural features. Within the extracellular region of CXCR3, which includes the N-terminal domain and three extracellular loops (ECL1–3), the N-terminal domain and ECL-3 play a crucial role in docking with CXCL10 and CXCL11 [7]. An especially intriguing structural feature is ECL-2, whose absence completely abolishes receptor activation in response to any ligand. This finding highlights its essential role in GPCR activation[7]. At the intracellular level, pathway activation—such as ERK phosphorylation and calcium mobilization—is primarily governed by the carboxy-terminal domain in the cytoplasm and a highly conserved tripeptide sequence (DRY) located within the third transmembrane α-helix [38]. The ligand-receptor interaction is further stabilized by aromatic and electrostatic interactions between amino acid residues in the N-loop of CKs [39]. Specifically, in CXCL10, two hydrophobic pockets have been identified as key contributors to bond stabilization [40]. At the electrostatic level, the positively charged amino acids in the N-terminal domain of chemokine ligands interact with negatively charged amino acids in the N-terminal region of the receptor, as well as in ECL-2 and ECL-3 [41]. Another key factor in stabilizing these interactions is the presence of phosphorylation sites at the contact regions between the ligand and receptor [42]. The three-dimensional (3D) structure of CKs is crucial for receptor binding. This structure is maintained by disulfide bonds involving cysteine residues, which give chemokines their characteristic folding pattern. If these disulfide bonds are disrupted, the ligand irreversibly loses its ability to dock with the receptor[39]. According to the "2-step model" of ligand-receptor interaction, the process begins with a recognition phase, in which the N-loop of the chemokine is identified by the N-terminal tail of the receptor and its extracellular loops [41], [43]. This is followed by the binding phase, in which the chemokine's docking site is correctly positioned within its flexible N-terminal domain to facilitate stable interaction [43]. Once ligand recognition and binding occur, CXCR3 undergoes internalization, a process mediated by the adaptor protein β-arrestin. This internalization is critically influenced by threonine residues in the carboxy-terminal tail of the receptor [38]. Additionally, post-translational modifications (PTMs) of CXCR3 are essential for optimizing its structural properties and interactions with ligands. For example, sulfation of tyrosines at positions 27 and 29 (Tyr27 and Tyr29) plays a key role in signal transduction activation [42], as well as the N-glycosylation of asparagines Asn22 and Asn32 [44].

The existence of three ligands for the same receptor raises the question if and how they can differentiate their actions. In some cases, this differentiation depends on a phenomenon called "ligand dominance” [45]. For example, in certain CNS infections, CXCL10 exhibits clear dominance over the other two ligands [45]. This is particularly evident in two murine models of Dengue virus [46] and West Nile virus [47] infections. Although the inflammatory response triggered by these pathogens involves CXCL9, CXCL10, and CXCL11, only CXCL10 appears to be indispensable. In CXCL10-deficient mice, the infection fails to induce massive T lymphocyte recruitment, a critical process for preventing fatal encephalitis[46], [47]. Neither CXCL9 nor CXCL11 can compensate for the absence of CXCL10 in these settings. In other cases, mechanisms of true "ligand collaboration" come into play, where the three CKs function together to fine-tune immune responses [45]. In a murine model of cerebral malaria caused by Plasmodium berghei, the absence of either CXCL9 or CXCL10 results in an attenuation of the severity of the syndrome, even when the other ligand is still present [48], [49]. The likely reason for this is their distinct cellular sources: CXCL9 is primarily produced by cerebrovascular endothelial cells, whereas CXCL10 is secreted by neurons. Together, these molecules facilitate the recruitment of immune cells when the parasitic infection spreads to the nervous tissue [48]. Lastly, a situation of "ligand antagonism" has also been observed [45]. The first evidence of this was verified in a murine model of rejection after allograft heart transplantation[50]. When CXCL9⁻/⁻ mice were used as donors, there was a decrease in the number of Th1 cells and cytotoxic T lymphocytes that were activated and contributed to rejection. This is likely due to the fact that allogeneic DCs serve as the primary reservoir of CXCL9. This is likely since allogeneic DCs are the main reservoir of CXCL9. On the other hand, in CXCL10⁻/⁻ recipient mice, the rate of graft rejection increased, which was directly associated with the enhanced recruitment of CD8⁺ T lymphocytes [50]. These findings underscore that the functional consequences of CXCL9, CXCL10, and CXCL11 interactions with CXCR3 are highly context-dependent, varying according to the type of immune challenge the host is facing.

Although our understanding of the CXCL10-CXCR3 axis is still evolving, the functional consequences of CXCL10-CXCR3 interaction extends far beyond immune cell recruitment. As such, CXCL10-CXCR3 axis functions also as pivotal regulator of transcriptional programming, metabolic reprogramming, and fate specification across adaptive and innate immune cells in neuroinflammation. This axis does serves as a critical interface between chemotaxis and cellular adaptation, ensuring that recruited cells are not only delivered to inflamed tissue but also primed for the metabolic [51], [52] and transcriptional demands imposed by the inflammatory microenvironment [53]. Furthermore, a growing body of evidence now highlights its critical role in the pathophysiology of CNS disorders. Following a general overview of its function, the second part of this review will explore current insights into its involvement in several prevalent CNS pathologies, including infectious, tumoral, neurodegenerative, and autoimmune diseases. To provide a comprehensive and translational perspective, particular emphasis will be placed on in vitro models, in vivo animal models, and ex vivo patient-derived material, which have been instrumental in elucidating the role of the CXCL10-CXCR3 axis in the progression of neurodegenerative pathologies.

Lymphocytic choriomeningitis virus

Lymphocytic choriomeningitis virus (LCMV) is a non-cytolytic murine virus capable of triggering a substantial inflammatory response in the meninges and choroid plexus, typically leading to the death of infected mice within a week [54]. Although LCMV is primarily a rodent-borne virus, documented cases in the literature described human infections, which manifest as acute febrile neurological syndromes and aseptic meningitis [55], [56]. Due to its unique pathogenesis, LCMV serves as an ideal model for studying abnormal immune cell trafficking in the CNS, as the fatal outcome in mice is entirely attributed to the host's inflammatory response, rather than direct viral cytopathic effects [57]. The presence of elevated levels of CXCL10 and CXCR3 mRNA in the brains of infected mice has led to the hypothesis that these molecules are involved in the pathogenic process [58], [59]. Assensio and Campbell verified that cerebral expression of CXCL10 in mice infected with the LCMV-Armstrong viral serotype (LCMV-A) increases by approximately 35-fold compared to the non-infected controls [59]. Of particular interest was the assessment of a substantial overlap between the cerebral regions overexpressing CXCL10 (meninges, choroid plexus, olfactory bulbs, and ventricles) and those in which the viral was most prevalent [59]. Using GKO-IFNγ gene-disrupted and athymic nude mice, Assensio and Campbell documented a marked reduction in CXCL10 expression, which was ultimately abolished when the immune system was completely compromised [59]. Further supporting the role of the CXCL10-CXCR3 axis in the disease process, Christensen et al. demonstrated that LCMV-Traub serotype (LCMV-T) infection leads to a 63 % increase in CXCR3 expression on CD8 + T lymphocytes from the spleen [57]. Strikingly, at the cerebrospinal fluid (CSF) level, CXCR3 expression on CD8⁺ T cells reached up to 93 % [57]. This massive infiltration of T lymphocytes into the CNS leads to fatal meningitis in all wild-type (WT) mice, causing death within 10 days post-infection. However, when CXCR3-deficient mice were infected, most survived, exhibiting no signs of meningitis, at least within the 20–35 day post-infection window [57]. Having demonstrated that the absence of CXCR3 does not affect the activation capacity of CD8 + T lymphocytes, it was possible to conclude that mouse survival depends on two pivotal factors: a significant deceleration in immune cell recruitment to the meninges and the ependyma, as well as a distinct arrangement of the infiltrate [57]. Simultaneously, yet separately, De Lemos and colleagues arrived at the same conclusions, further validating the crucial role of CXCR3 in regulating leukocyte trafficking across brain tissues [60]. Recognizing the crucial involvement of CXCR3, the role of its most recognized ligand, CXCL10, was further investigated using CXCL10-deficient mice [61], [62]. Similar to CXCR3-deficient models, mice lacking CXCL10 exhibited a significantly prolonged survival compared to WT mice [61], [62]. Notably, 40 % of CXCL10-deficient mice achieved complete viral clearance and survived without any signs of meningitis [61]. Contrary to what has been discussed thus far, it appears that the absence of CXCR3 expression is unable to alter the lethal course of neurological pathology in mice infected with LCMV-A. Hofer and colleagues confirmed the overexpression of CXCR3 and CXCL10 following intracranial injection of LCMV-A; however, the absence of CXCR3 did not attenuate lethality, suggesting that other compensatory mechanisms may drive disease progression in this context [63]. Improvements were not observed even in IFNγR (Interferon-gamma receptor)-knockout mice (KO), despite the fact that the defect of IFNγ prevents the expression of CXCL10 [63]. Similarly, the distribution of T lymphocytes (CD4 + and CD8 +) remained unchanged [63]. Further efforts to elucidate the reasons behind such a profound disparity need to be extended; nonetheless, the authors hypothesize that subtle differences in viral dissemination and antiviral dynamics between the two strains may be at play [63].

Human immunodeficiency virus

Even though combination antiretroviral therapy (cART) ensures life expectancy for Human immunodeficiency virus (HIV)-positive individuals comparable to that of the general population and effectively mitigates hallmark AIDS (Acquired immune deficiency syndrome) manifestations, such as opportunistic infections [64], it is estimated that approximately 50 % of treated individuals experience neurological complications. They are collectively categorized as HIV-associated neurocognitive disorders (HAND) [65]. HAND encompasses a broad spectrum of clinical manifestations, ranging from mild impairments in motor and memory functions that minimally affect daily activities to severe HIV-associated dementia (HAD). HAD is characterized by profound motor and cognitive deficits, which can result in significant functional decline and, in some cases, lead to mortality within a year [66]. The mechanisms by which HIV infiltrates the CNS are multifaceted. These include the infection of vascular pericytes [67], which surround the cerebral vasculature and contribute to the formation of the BBB, as well as the infection of CD16 + monocytes recruited into the brain parenchyma [68], [69]. Within the CNS, the primary viral reservoir is constituted by microglial cells, which drive a pervasive state of neuroinflammation that culminates in aberrant pruning of synaptic terminals [70], [71]. Additionally, approximately 20 % of astrocytes are estimated to harbour latent HIV infection, further amplifying local inflammatory processes and facilitating viral dissemination [72]. The persistence of latent infection in these cells, even in the context of cART administration, makes the complete eradication of HIV from the CNS virtually impossible [73]. Although latent, the presence of HIV sustains the low-level production of viral proteins, notably gp120 (Glycoprotein 120) [74] and TAT (Trans-activator of transcription)[75]. These proteins are key drivers of chronic inflammation, as they enhance the expression of pro-inflammatory cytokines [75], [76], and compromise the integrity of the BBB [77]. A strong correlation exists between HIV infection and the production of CXCL10 in the CNS. Notably, both the viral gp120 [78] and TAT [79] protein have been shown to upregulate CXCL10 expression in mice and SIV (Simian immunodeficiency virus)-infected macaques [80]. Elevated levels of CXCL10 are also observed in the CSF and brain tissue of HIV-positive individuals, with significantly higher concentrations detected in those diagnosed with HAD [81], [82]. CXCL10 levels are markedly elevated in the CSF during the early stages of HIV infection, correlating with high viral RNA loads and leukocyte counts. Although these levels tend to decline, particularly with the administration of cART, they remain significantly elevated in patients with HAND. In this population, a linear relationship has been demonstrated between plasma CXCL10 levels and the severity of neuropsychiatric manifestations [83]. Finally, it has been demonstrated that latently infected astrocytes are the primary contributors to the production of this CK [84]. The mechanisms through which a clinically inactive HIV infection can still induce neurological damage, leading to severe dementia, remain poorly understood. Mehla et al. infected human astrocytes and mixed neurons with HIV-1, collected the supernatant, and appraised the migratory capacity of human peripheral blood mononuclear cells (PBMCs) [85]. Antibodies against CXCR3 have proven to be an effective treatment, resulting in the near-complete abolishment of the migratory capacity of PBMCs. Similar findings were observed with Bryostatin treatment, a protein kinase inhibitor and PKC activator, as well as a well-known neuroprotective factor. Its protective action relies on a negative regulation of the CXCL10 secretion by virus-infected astrocytes [85]. Molecular dynamics underlying the modulation of CXCL10 expression include p38, JNK 1/2 (Jun N-terminal kinase 1/2), and Akt pathways, all converging to the activation of NF-κB (Nuclear factor kappa-light-chain-enhancer of activated B cells) and STAT1 (Signal transducer and activator of transcription 1). NF-κB translocates into the nucleus of the astrocytes and, among other genes, it initiates the transcription of the CXCL10 gene, whose promoter hosts a κB binding site [86]. They confirmed that HIV-1 infection-induced increase in CXCL10 secretion promotes the homing of lymphocytes and monocytes into the brain, a mechanism that underlies a prominent neurotoxicity and has been identified as a potential contributor to neurological disorders [86] [Fig. 2a]. Another model has been proposed to explain the neurotoxic effects of CXCL10, whose transcription is induced by HIV proteins. Incubating foetal neurons with CXCL10 leads to a release of Ca2+ from cellular reservoirs, including the ER. The Ca2+ is taken up by the mitochondria, and the resulting increase in its concentration triggers the release of cytochrome-C that acts as an activator for Apaf-1 (Apoptotic protease activating factor-1). Upon binding with pro-caspase 9 and oligomerizing, Apaf-1 forms the apoptosome complex. Within this complex, caspase-9 becomes activated and subsequently cleaves pro-caspase-3 into its active form, which induces programmed cell death and neuronal loss [87] [Fig. 2b].

Zika virus and Guillain-Barré syndrome

A recent study also highlighted the pivotal role of the CXCL10-CXCR3 axis in the resolution of Zika virus (ZIKV) infection through its involvement in the recruitment of CD8⁺ T lymphocytes to the brain [88]. ZIKV is a member of the Flaviviridae family, typically causing mild and self-limiting illness in adults [89]. However, its ability to cross the placental barrier and its broad organotropism were associated with severe neurological complications in the foetus [90], [91]. Using CXCR3-deficient mice, Nazerai et al. demonstrated that CXCR3 expression on T lymphocytes is essential for an effective immune response to eliminate ZIKV. In its absence, the amount of ZIKV-specific CD8⁺ T cells in the brain was reduced and even when it increased to levels comparable to those in WT mice, the delayed recruitment and activation of T cells allowed the ZIKV a wider window for replication [88]. A potential correlation has also been proposed between ZIKV infection and peripheral neurological manifestations within the spectrum of Guillain-Barré syndrome (GBS), an autoimmune disorder affecting peripheral nerves and dorsal root ganglia [92]. It is characterized by progressive motor and sensory deficits primarily driven by demyelinating damages [93]. Parallel investigations reported elevated levels of CXCL10 in the CSF of patients, alongside strong expression of this chemokine in the peripheral nerve vascular endothelium. CXCR3, in turn, is upregulated in cell types implicated in nerve injury, including macrophages, the resident immune cell population, infiltrating T lymphocytes and plasma cells [94]. These findings have been substantiated in the experimental autoimmune neuritis (EAN) mouse model, which recapitulates key features of the human disease and is induced by immunization with peripheral myelin proteins [95], [96]. Although further studies are required before these findings can be translated into clinical practice, there is compelling evidence suggesting that targeting this axis may represent a promising therapeutic strategy. In an unpublished 2012 study, the use of a CXCR3 antagonist appeared to promote significant recovery in mice affected by a particularly aggressive variant of EAN induced by immunization with bovine myelin proteins and leading to a rapid and monophasic demyelinating polyneuritis [94]. Nonetheless, it can here again be noted that signalling through the CXCR3-CXCL10 axis can either be necessary/beneficial (ZIKV) or detrimental (GBS), underscoring the need for agonistic or antagonistic therapeutic interventions dependent on the pathology, which unfortunately – as of to date – need further development and pre-clinical validation before their true value can be explored in clinical trials [97].

Severe acute respiratory syndrome coronavirus 2

Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) is the etiological agent responsible for Coronavirus Disease 2019 (COVID-19), a condition that, following its rapid spread in early 2020, led to a full-scale global pandemic [98]. The ability of SARS-CoV-2 to disseminate throughout the body underlies its capacity to trigger a sudden and massive release of pro-inflammatory cytokines, a phenomenon known as cytokine storm. One of the most common clinical manifestations is acute respiratory distress syndrome (ARDS), which represented the leading cause of death among patients, especially during the early months of the pandemic [98]. However, additional symptoms considered more distinctive have also been identified, including anosmia (loss of smell) and ageusia (loss of taste), both of which immediately pointed the scientific community toward the viral neurotropic potential [99]. SARS-CoV-2 has been shown to invade olfactory sensory neurons, the only neuronal population in direct contact with the external environment, due to their expression of viral entry mediators such as angiotensin-converting enzyme 2 (ACE2), neuropilin-1 (NRP1), and transmembrane protease serine subtype 2 (TMPRSS2) [100]. This observation laid the foundation for in-depth investigations into the link between pro-inflammatory cytokine production and associated neurological manifestations. Attention rapidly turned to CXCL10, one of the key chemokines induced in response to SARS-CoV-2 infection, whose circulating levels were found to correlate directly with both viral load and COVID-19 severity [101]. Oliviero et al. highlighted a hypothetical parallel between the mechanism through which SARS-CoV-2 induces such phenomena, and the demyelinating mechanisms described for another murine coronavirus, the John Howard Mueller strain of mouse hepatitis virus (JHMV) [102]. JHMV infection is characterized by the early induction of two pro-inflammatory cytokines, IFNγ and CXCL10, the latter playing a key role in recruiting T lymphocytes necessary for viral clearance. The uncontrolled inflammatory response also contributes to demyelinating damage that may persist chronically, ultimately exerting an encephalitogenic effect. Altogether, these findings suggest that the CXCL10–CXCR3 axis may play a pivotal role in the molecular mechanisms underlying SARS-CoV-2 neuropathogenesis [103].

Neuroborreliosis

Borreliosis, also known as Lyme disease or Lyme borreliosis, is one of the most common anthroponosis on the planet, transmitted to humans through the bite of ticks from the Ixodes genus. The main etiological agents are bacteria of the Borrelia genus, with cases up to 60 % caused by B. burgdorferi [104]. The most common acute clinical manifestations are dermatological including erythema migrans, lymphocytoma, fever, and fatigue [104]. In approximately 40 % of affected patients, the disease involves the nervous system, leading to a condition known as neuroborreliosis (NB) [105]. The symptoms are no longer solely acute (meningopolyradiculonephritis) but can become chronic, manifesting as encephalomyelitis, spastic ataxia, and urinary difficulties [106]. Additionally, 15 % of patients develop cerebrospinal lymphocytic meningitis [105]. Consequently, several studies have been conducted to elucidate the role of the CXCL10-CXCR3 axis in leukocyte recruitment within this pathological context. Among the pioneers in demonstrating an increased pleocytosis in CSF of patients with NB were Lepej and colleagues. They characterized the infiltrate, highlighting a high percentage of memory T lymphocytes (CD4 + CD45RO+), which were also strongly positive for CXCR3 [107]. Furthermore, CXCL10 concentrations were quantified, revealing markedly elevated levels in the CSF compared to serum levels in the same individual or in patients with non-inflammatory neurological disorders [107]. All of this led to an interesting hypothesis: the presence of the bacteria in the CNS stimulates the creation of a chemotactic gradient that attracts T lymphocytes into the CSF. The attempt at bacterial clearance results in a chronic inflammatory state [107]. This hypothesis has been further validated by Moniuszko et al. [108]and Henningsson et al. [109]. These inflammatory responses are even more pronounced in children under 15 years of age. Furthermore, an increased concentration of IL-17 (Interleukin-17) has been detected in approximately 50 % of affected individuals, correlating with higher pleocytosis [109]. By cross-referencing these data, researchers hypothesized for the first time that, in addition to the Th1 compartment, the inflammatory response in neuroborreliosis may also be orchestrated by Th17 lymphocytes [109].

Cerebral Malaria

Malaria is a parasitic disease that remains highly lethal to this day, caused by protozoa belonging to the Plasmodium genus. Among these, the species primarily responsible for the most common and deadly manifestations is P. falciparum, transmitted to humans through the bites of female mosquitoes of the Anopheles genus [110]. One of the most severe complications that can arise following P. falciparum infection is cerebral malaria (CM), which results in severe encephalopathy and can lead to even irreversible comatose stages [110]. Two are the most widely accepted hypotheses for explaining the pathogenesis of this complication. The first theory, the "mechanical hypothesis", aims to elucidate neurological damage starting from vascular impairment. The massive sequestration of red blood cells in cerebral vessels would lead to a congestion, consequently resulting in hypoperfusion and hypoxia [110]. Furthermore, red blood cells are colonized by the Plasmodium parasite, which coats the cell surface with a protein known as PfEMP-1 (P. falciparum erythrocyte membrane protein-1). This protein interacts with some endothelial receptors, such as ICAM (Intercellular adhesion molecule) and CD36, activating pathways of coagulation and inflammation [110]. The second theory, known as the "cytokine storm hypothesis", attributes vascular damage to the elevated circulation of pro-inflammatory cytokines (TNFα, IFNγ, IL-2, etc.), which play a role in the activation of neutrophils and T lymphocytes, widely observed at intra- and perivascular levels [110]. Given the notable presence of CD8 + T lymphocytes, it is reasonable to believe that the chemotactic component could be crucial in the pathogenic mechanism. The quintessential murine model in the study of CM involves infecting mice with PbA (P. berghei ANKA strain), which induces a syndromic condition highly reminiscent of the human counterpart. It is precisely through the utilization of this model that Belnoue and colleagues have dispelled any uncertainty regarding the involvement of CD8 + T lymphocytes in the pathogenesis of cerebral malaria [111]. By depleting this population, using antibodies or by employing mice deficient in their production, they have demonstrated that infection is entirely incapable of inducing CM. Even the mortality within the typical timeframe of CM has been completely abolished: death occurs much later due to anaemia and hyperparasitemia [111]. Complete abrogation of CM has also been achieved through the depletion of NK cells with anti-Asialo GM1(Ganglio-N-tetraosylceramide) antibodies and minimal vascular damage [112]. The study also highlighted the significance of CXCR3 expression on both T lymphocytes and NK cells, which lead to the extensive recruitment of the former. More than 90 % of the lymphocytes found in the brains of mice afflicted with CM are positive for this receptor, while the percentage exceeds 80 % [112]. Hence, NK cells have proven to be a crucial upstream factor for the migration of T lymphocytes through CXCR3 expression. T cells isolated from the spleens of NK-depleted mice were significantly less responsive to CXCL10. Notably, the adoptive transfer of NK cells is sufficient to reverse this phenomenon- but only if they are capable of producing IFNγ [112]. Sequential studies have instead demonstrated that following PbA infection, there is an overexpression of CXCL10 in the brain [113], [114]. This is dependent on IFNγ as a trigger for CXCL10 increase: IFNγ KO mice do not experience any overexpression and do not exhibit any symptoms related to CM [113]. Van den Steen et al. have ultimately dispelled any doubts regarding the centrality of CXCR3 as a determinant factor by comparing C57BL/6 mice, which are susceptible to the disease, with Balb/c mice, which are spontaneously resistant [113]. The upregulation of CXCR3 is significant on CD4 + T lymphocytes in C57BL/6 mice, but largely absent in Balb/c, a characteristic that subsequently limits the capacity to respond to CXCL10 and cause neuroinflammation [113].

Tumors within the CNS are relatively rare; nevertheless, they are associated with a high mortality rate. Furthermore, they represent the foremost cause of cancer-related morbidity among pediatric patients (aged 0–14 years), and the second most prevalent in adolescents (aged 14–19 years) [115]. These tumors comprise a highly heterogeneous group of pathologies due to numerous factors, including anatomical location, cell of origin, degree of malignancy, and, increasingly crucial in the era of omics sciences, distinct types of molecular alterations [115], [116]. The World Health Organization (WHO) has classified over 100 types of CNS tumors [116]. Among the notably infrequent primary tumors, it is important to note that the CNS serves as a site for metastases originating from other malignancies, primarily breast cancer, lung cancer, and melanoma. It is estimated that cerebral metastases are approximately 10 times more prevalent than primary tumors [117]. In 2008, a study was published that evaluated the expression profile of CKRs in human cell lines at various degrees of malignancy [118]. The findings demonstrated that cells of anaplastic astrocytoma (WHO grade III) and those of glioblastoma multiforme (WHO grade IV - GBM) exhibit significantly higher expression levels of CXCR3 compared to untransformed astrocytes—up to fourfold and sevenfold higher, respectively [118]. Moreover, significant increase in CXCL10 expression has also been observed in tumor cells [118]. [Fig. 3a] Given that both the receptor (CXCR3) and its ligand (CXCL10) are concurrently expressed on the same cell, the authors investigated the potential for this pathway to function in an autocrine manner. Indeed, upon adding CXCL10 to the tumor cell culture, they observed an increase in new DNA synthesis, indicating enhanced proliferation. However, neutralizing CXCR3 did not halt this effect, indicating that the replicative ability is not solely dependent on this signaling [118]. This evidence has also been corroborated by Sharma et al., who, through immunohistochemical analysis, demonstrated a clear direct correlation between the production of these cytokines and the degree of tumor malignancy, ranging from pilocytic astrocytoma to GBM [119]. Further confirmation has been obtained from the utilization of 3D cultures, particularly gliomaspheres (GSphs). Since these structures originate from tumor stem cells, they more accurately recapitulate both the genotype and phenotype of the original tumor, providing a more faithful model for studying tumor biology [120], [121]. The expression of CXCR3 is even more pronounced in 3D structures compared to 2D cultures. Notably, isoform A – whose expression level correlates with a higher tumor growth rate – is consistently present in GSphs derived from all tested glioma cell lines as well as human biopsy specimens [120]. Several murine models have been developed to study the biology of these tumors, such as the intracerebral inoculation of the GL261 cell line (murine glioma cells) [122]. Using this model, Liu et al. evaluated tumor progression in the absence of CXCR3, utilizing CXCR3-deficient mice or administering the receptor antagonist, NBI-74330 [120]. In the former scenario, mouse survival was compromised due to malfunctioning of NK cells, which were entirely incapable of infiltrating the tumor. Conversely, with the second strategy, precisely the opposite was observed: in mice where CXCR3 was antagonized, tumor growth was more effectively controlled, leading to increased survival [120]. The dependence of these tumors on the signalling of the CXCL10-CXCR3 axis has also been demonstrated through in vivo treatment with Celecoxib. In this study, wild-type (WT) mice were inoculated with glioma stem cells (GSCs)—neural/progenitor stem cells engineered to overexpress H-RAS and exhibit a homozygous deletion of Ink4a/Arf—and subsequently treated with Celecoxib [123]. The antitumor effect of Celecoxib relies on its ability to induce direct apoptosis in cancer cells through caspase cleavage and restrict replication by inhibiting β-catenin and cyclin D1 [123]. However, what has intrigued researchers the most is its antitumor mechanism that relies on reducing the expression of CXCR3 and CXCL10, further highlighting the critical role of this signaling axis in glioma progression [123]. This applies only partially to GSCs, in which CXCR3 is reduced while CXCL10 is not affected, but fundamentally to cells populating the tumor microenvironment (TME), glioma-associated microglia/macrophages (GAMs) in particular. These cells are Nestin+ , Iba1 + (Ionized calcium-binding adapter molecule 1), CD163 + or CD16 + and exhibit an intermediate M1-M2 phenotype, able to support tumor growth [123]. The most significant reduction in CXCL10 synthesis is observed specifically in GAMs, effectively disrupting their paracrine communication with transformed cells [123]. [Fig. 3b] Given that this molecular axis plays a role in various aspects of tumor progression, further investigations were conducted to determine its potential involvement in metastatic mechanisms. Importantly, the Neta Erez research group has developed a murine model of melanoma in which cerebral metastases spontaneously arise [124]. Using this model, researchers have demonstrated that a specific phenotype of astrocytes, known as metastasis associated astrocytes (MAAs), is capable of generating a chemotactic gradient directed not towards immune cells, but rather directed towards cells of the primary tumor [125]. The tumor cells, in turn, overexpress CXCR3 enabling them to respond to and migrate along this chemotactic gradient toward the metastatic niche where they encounter a favorable microenvironment for engraftment and formation of cerebral metastases [125]. This homing effect is so impactful that by neutralizing CXCL10, through neutralizing antibodies or knock-down approaches, or by knocking down CXCR3, it is possible to significantly impair the migratory capacity of melanoma cells with tropism for brain tissue [125]. While substantial evidence supports a pro-tumoral role for the CXCL10-CXCR3 axis in tumor survival and progression, there are some other studies that highlight its potential to drive an anti-tumoral response, orchestrated by immune cells, capable of limiting tumor development. Recent studies have demonstrated that murine mesenchymal stromal cells engineered for stable CXCL10 production, when injected peritumorally, significantly inhibit the in vivo growth of two murine glioblastoma cell lines, GL261 and CT2A. This effect is attributed to a marked increase in the infiltration of CXCR3 + T lymphocytes into the tumor (CXCR3 + TILs – Tumor infiltrating lymphocytes), which actively eliminate tumor cells [126]. Limited infiltration of CXCR3 + TILs, leading to compromised immuno-surveillance, appears to be driven by reduced production of ligands such as CXCL10 compared to other solid tumors [127]. In other anatomical compartments, the CXCL10–CXCR3 axis has traditionally been regarded as exerting a protective role, primarily due to its chemotactic ability to recruit TILs and other beneficial immune cell subsets, such as NK cells. This is exemplified in primary hepatocellular carcinoma, where disease control appears to be more effective in the presence of the hepatic symbiont Lactobacillus reuteri. This bacterium has been shown to upregulate IFN-γ and CXCL10 expression, whose elevated levels have been associated with improved prognosis [128].

Parkinson’s Disease

Parkinson’s disease (PD) ranks as the second most prevalent neurodegenerative disorder among the elderly population, distinguished by a gradual degeneration of dopaminergic neurons situated within the substantia nigra [129]. The primary hallmark is the emergence of Lewy bodies, insoluble protein aggregates, primarily composed of α-synuclein. The symptoms are predominantly motor in nature, characterized by resting tremor of the upper limbs, bradykinesia, and rigidity. However, the condition can also entail cognitive decline (dementia) and behavioral disturbances, such as depression [129]. It is a condition with a multifactorial etiology that remains incompletely elucidated. Unfortunately, as of today, there are no available therapies capable of reversing the degenerative effects [129]. Limited data are currently available regarding the correlation between the CXCL10-CXCR3 axis and the pathophysiological features of PD. However, some evidence suggests a potential link. First, in the murine MPTP (1-methyl1–4-phenyl1–1,2,3,6-tetrahydropyridine) model of PD, it has been shown that CXCL10 is among the CK whose expression is increased [130]. Other preliminary data supporting this observation dates back to 2010 and pertains to the role of neuromelanin (NM) [131]. This pigment imparts the characteristic brown hue to nigrostriatal neurons and its role remains contentious. Traditionally, NM has been linked to a cytoprotective effect due to its ability to chelate transition metals, particularly iron, thereby mitigating the impact of toxic free radicals [132]. On the other hand, recent studies suggest that CXCL10 may contribute to astroglial activation when not sequestered within the cytoplasm, thereby perpetuating degenerative damage to motor neurons [133], [134]. However, assuming that NM possesses neuroprotective properties, it has been demonstrated that when the human astroglial A172 cell line is inflamed with TNFα and subsequently exposed to NM, there is a notable decrease in CXCL10 secretion. This finding suggests a potential regulatory role of NM in modulating inflammatory responses, possibly counteracting CXCL10-mediated neurotoxicity [131]. The only other available data on the CXCL10-CXCR3 axis in PD pertains to α-synuclein. Tousi et al. observed that exposing the A172 astroglial cell line to α-synuclein led to a notable increase in CXCL10 secretion [135]. Further supporting this link, Rocha et al. published a study demonstrating an apparent correlation between serum CXCL10 concentration and cognitive status in PD patients. This association was characterized by poorer intellectual performance and reduced mental flexibility [136]. Although further studies are needed, a recent bioinformatics analysis reinforced this hypothesis, highlighting that one of the significantly dysregulated hub genes in PD is associated with CXCL10 signaling pathways[137].

Alzheimer’s Disease

Alzheimer's disease (AD) is by far the most prevalent cause of dementia in the elderly and stands as the foremost neurodegenerative condition in terms of morbidity [138]. Clinically, the disease entails a gradual decline in cognitive functions, characterized by initial sporadic episodes of memory loss, disorientation and confusion (MCI – mild cognitive impairment), ultimately leading to a complete loss of individual autonomy. Although various related phenomena are known, such as the deposition of "senile" plaques composed of β-amyloid (Aβ) or hyperphosphorylated tau protein, the underlying causes driving the onset of this multifactorial condition remain elusive. Unfortunately, much like with PD, there are currently no drugs available to halt the progression of the condition [138]. It is now well established that the immune system plays a pivotal role in the pathogenesis of AD. The degeneration of synapses is heavily influenced by an abnormal activation of immune cells, upon contact with oligomers of Aβ, leading to the release of proinflammatory cytokines that perpetuate the damage [138]. Initial studies using immunohistochemistry on post-mortem human brain tissue revealed a strong CXCL10 positivity in periplaque astrocytes [139]. This has also been demonstrated by Lai et al. who incubated cultured astrocytes with Aβ and observed alterations in their activation state. These changes included the upregulation of pro-inflammatory markers, such as glial fibrillary acidic protein (GFAP), along with modifications in their secreted factors. Their findings ultimately confirmed that CXCL10 acts as a key activator of astrocyte migration, prompting these cells to cluster and encircle amyloid plaques [140]. [Fig. 4a] The concentration of CXCL10 in the CSF has been also found to be elevated compared to non-AD patients, particularly in the early stages of AD, when the first signs of MCI become evident [141]. Given that AD poses a significant societal challenge in a human population experiencing increases in life expectancy, complex murine models have been developed to further investigate its pathophysiological mechanisms, for example, causing the overexpression of the APP (Amyloid-beta precursor protein) isoform 695 and the KM670/671NL (Swedish) mutation [142]. These models have been employed to validate findings observed in humans, demonstrating that CXCL10 positivity is significantly higher in the murine model of AD compared to the healthy counterpart. Once again, it is co-localized with Aβ [143]. The same conclusions were subsequently confirmed using two additional models: 3xTg-AD mice, which carries mutations (APPswa, tauP301L, and PSEN1M146V) in three genes encoding for APP, tau, and PSEN1 (Presenilin 1 – also PS1) [144], and 5xFAD mice, which harbours 5 mutations, two (M146L and L286V) in PSEN-1 and three in APP (KM670/671NL, I716V, V717I) [145]. Both models exhibit a significant upregulation in CXCL10 mRNA expression [144], [145]. Another widely utilized model is the "APP/PS1 double-transgenic," which manifests Aβ accumulation and cognitive decline due to the expression of chimeric human/mouse APP (Mo/HuAPP695swe) and a mutant human PSEN-1 (PSEN-1-dE9) [146]. In addition to demonstrating the elevation of CXCL10 production, it was used by Krauthausen et al. to provide the most compelling evidence that the CXCL10-CXCR3 axis is directly implicated in Aβ plaque deposition. [147]. APP/PS1 double-transgenic mice simultaneously knocked out for CXCR3 exhibit an unequivocal reduction in Aβ plaque deposition areas and an approximately 5-fold lower concentration of insoluble Aβ peptides [147]. Another dramatically diminished aspect is the recruitment of activated microglial cells and astrocytes, which are otherwise prominently present near deposited plaques [147]. The enhanced control over the number and volume of plaques in the study is attributed to the more efficient phagocytic capacity of microglial cells. It is indeed demonstrated that the signaling activated by CXCL10, upon interaction with CXCR3 on the surface of microglial cells, can reduce the degradation capacity of precipitated Aβ by approximately 30 %. This function, both in vivo and in vitro, remains almost intact in APP/PS1 CXCR3-/- mice [147]. As an overall outcome, there is a reduction in proinflammatory cytokines within the brain tissue, exactly in contrast to what occurs with neurotrophic factors like BDNF (Brain-derived neurotrophic factors), accompanied by a general improvement in the memory capabilities of the tested mice [147] [Fig. 4b].

Multiple sclerosis (MS) is a disorder that represents the most prevalent cause of non-traumatic disability in the young adult population [148]. The causes leading to the development of this condition are not yet fully elucidated, although associated environmental factors have been identified, including EBV (Epstein-Barr virus) infection, vitamin D deficiency, smoking, and UVB (Ultraviolet B radiation). The strongest genetic association pertains to the HLA-DRB1*15:01 allele (Human Leukocyte Antigen), which increases the risk of developing the condition by 3-fold in heterozygotes and 6-fold in homozygotes [148]. While the autoimmune hypothesis is generally accepted for its etiology, factors such as the absence of a clear autoantigen, against which the immune response is mounted, and the frequent lack of responsiveness to immunosuppressive therapies cast doubt on this conviction [149]. What is certain is that the immune component remains the triggering cause: the pathognomonic signs include marked perivascular inflammation and infiltration, primarily by CD8 + T lymphocytes, B cells, and plasma cells. All of this leads to the disease hallmark par excellence: the development of demyelination plaques, caused by the degradation of axonal myelin [148]. Since the etiology of the damage is to be traced within the immunological compartment, research interest has promptly focused on the investigation of molecules capable of orchestrating the migration of leukocytes, foremost the CXCL10-CXCR3 axis. In 1999, two separate research groups observed an elevation in the number of CXCR3 + T lymphocytes in both the blood and CSF of individuals afflicted with MS. Furthermore, CXCL10 concentration was found to be elevated in the CSF [150], [151]. Examination of post-mortem human brain tissue samples also revealed intense CXCL10 positivity in astrocytes surrounding vessels within the demyelination plaque[152]. The high presence of CXCR3 + T lymphocytes was also noted located directly amidst the demyelination plaques, this serves as further evidence substantiating their role as the instigators of myelin sheath degradation [152]. Simpson et al., using post-mortem cerebral tissue samples, demonstrated that CXCL10 is produced not only by astrocytes but also by macrophages, whereas CXCR3 is strongly expressed by astrocytes. Consequently, not only immune system cells are recruited, but also astrocytes are attracted into the periplaque space, and the activation is not solely astrocytic, rather astroglial [153]. The concentration of CXCL10 in the CSF has subsequently been correlated in a directly proportional manner to the severity of the symptoms [154], [155]. The same has been observed for CXCR3, whose increased expression occurs concurrently with symptom exacerbation [156]. As expected, remission periods are characterized by the return of CXCR3 levels to baseline [156]. Many murine and rat models have been developed to study MS, allowing for the generation of demyelination phenomena that can recapitulate what occurs in human demyelination during MS progression [157]. In two consecutive studies, Liu et al. employed mice infected with the MHV (Murine hepatitis virus), which, when inoculated intrathecally, induces encephalomyelitis and chronic demyelination [158], [159]. Neutralizing anti-CXCL10 antibodies have proven effective in limiting the infiltration of CD4 + T lymphocytes and monocytes into the brain parenchyma, resulting in an improvement in the motor skills of mice. If treatment is discontinued, symptom exacerbation leads to complete paralysis of the mouse musculature [158], [159]. Similar effects are observed in mice genetically incapable of producing CXCL10 (CXCL10-/-) [159]. The experimental autoimmune encephalomyelitis (EAE) model serves as the most widely employed system for investigating demyelination. This model entails the exogenous introduction of an autoantigen derived from myelin which artificially induces immune system sensitization [160]. Building on the well-established chemotactic role of CXCL10, particularly in attracting monocytes and T lymphocytes that infiltrate nervous tissue [126], several approaches for CXCL10 neutralization have been tested, ranging from neutralizing antibodies [161], antisense oligonucleotides (ASOs) [162] to DNA vaccination [163]. In each of these strategies, an improvement in motor symptoms and disease progression were observed. This is mainly due to an efficient decrease in accumulation of CD4 + encephalitogenic T cells in CNS, with a profound modification of pro-inflammatory cytokine secretion and, most of all, of the Th1/Th2 ratio. [Fig. 5] A final model of demyelination involves cuprizone intoxication administered through the diet: it is a copper chelator that, when continuously administered, induces demyelination due to oligodendrocytes apoptosis and glial activation [164]. In CXCR3-/- mice, the course of the pathology and the onset of symptoms are delayed due to a reduction in the expression of pro-inflammatory cytokines, such as CXCL10, and a concurrent decrease in astrocyte and microglia activation [165]. Very similar effects are observed in CXCL10-/- mice, where long-term cell death is reduced, as is the extent of plaques, indicative of their role in supporting the degradation process [166]. The absence of CXCL10 appears to benefit the survival of oligodendrocytes in these mice, yielding a generally neuroprotective effect that is highly advantageous for axonal myelination [166].

Despite all this evidence supporting the idea that, as with previous diseases, inhibiting the CXCL10-CXCR3 axis improves experimental demyelination conditions, the discussion regarding MS remains controversial. Indeed, some published studies suggest that the absence of CXCR3 may worsen the degree of degeneration and the overall condition of the mice [167], [168]. Some authors have attempted to explain this discrepancy. Narumi et al. proposed that CXCR3 is also expressed at the lymph node level, and its absence allows lymphocytes greater freedom to migrate and penetrate deeper into the brain [167]. On the other hand, Muller et al. have argued that CXCR3 serves to keep leukocytes near the cerebral vasculature, facilitating the contact between cytotoxic T lymphocytes and Tregs that limit their cytopathic functions. In the absence of a restriction on dispersion through the brain parenchyma, immune cells would be free to spread and perpetrate increasingly extensive damage [168]. Notwithstanding these objections, the involvement of the CXCL10-CXCR3 axis remains indisputable, though further studies will be necessary.