Lymphocyte trafficking into the inflamed tissues represents a critical event in the pathogenesis of autoimmune diseases. Based on in vitro and in vivo evidence, leukocyte migration through the vascular wall is described as a sequential process including distinct adhesive events: tethering, rolling, chemoattractant-dependent activation of integrins, slow rolling, firm adhesion (also called arrest or sticking), crawling, and diapedesis (Ley et al., 2007; Vestweber, 2015). Specificity and diversity in leukocyte-endothelial cell interactions are generated by different combinations of interchangeable ligand-receptor pairs for each step of the adhesion cascade. The molecular specificity directing leukocytes to the site of inflammation is regulated by adhesion molecules such as selectins, integrins, and members of the immunoglobulin superfamily. A critical step in the transition from leukocyte rolling to arrest is the induction of firm adhesion by chemokines expressed on activated endothelium. Chemokine receptors on the surface of rolling immune cells bind to their cognate ligand, triggering a G protein-dependent signaling that leads to the activation of membrane-expressed β₁ and β₂ integrins (Montresor et al., 2012). At the BBB level, leukocyte extravasation follows the standard paradigm of cell migration, and the inhibition of adhesion mechanisms using different therapeutical approaches proved to be beneficial in several models of brain inflammatory conditions (Yednock et al., 1992; Piccio et al., 2002; Constantin, 2008; Rossi et al., 2011; Fabene et al., 2013; Zenaro et al., 2015; Farinazzo et al., 2018).

Adhesion molecules expressed by brain endothelial cells have been studied for three decades for their role in lymphocyte trafficking during EAE and MS. P-selectin is upregulated on CNS vasculature during EAE (Piccio et al., 2002, 2005; Döring et al., 2007), whereas E-selectin was found expressed on autoptic cerebral micro-vessels of MS patients (Washington et al., 1994). In support of these data, in vitro studies have shown that P-selectin glycoprotein ligand (PSGL)-1, the major P-selectin ligand, is involved in rolling of CD8+ T cells isolated from MS patients and contributes to the transendothelial migration of MS-derived CD4+ T cells (Battistini et al., 2003; Bahbouhi et al., 2009). Moreover, MRI targeting of P-selectin with iron oxide-conjugated antibodies, to evaluate the modulation of its expression on the endothelium during inflammation, was found useful as a predictive method of EAE activity (Fournier et al., 2017). However, blockade of P- and E-selectin had no therapeutic effect on EAE models (Engelhardt et al., 2005; Döring et al., 2007). Intriguingly, combined blockade of P-selectin and α4 integrins resulted in significantly better clinical outcome than anti-α4 integrin alone in EAE mice, suggesting that selectin blockade may have some therapeutic effect in MS as well (Kerfoot et al., 2006). Together, these data suggest that further studies are needed to better understand the potential pathological role of endothelial selectins in MS.

Cellular adhesion molecules (CAM) belonging to the immunoglobulin superfamily such as vascular CAM-1 (VCAM-1), intercellular CAM-1 (ICAM-1), activated leukocyte CAM (ALCAM), and platelet endothelial CAM-1 (PECAM-1) are upregulated on the CNS vasculature, enhancing leukocyte adhesion and migration to the brain in both EAE and MS (Steffen et al., 1994; Cayrol et al., 2008; Steiner et al., 2010; Greenwood et al., 2011; Wimmer et al., 2019). The involvement of VCAM-1 and ICAM-1 is confirmed by the high expression of very late antigen (VLA)-4 and lymphocyte function-activated antigen (LFA)-1, their main ligands, respectively, on the immune infiltrates populating MS lesions (Cannella and Raine, 1995; Bö et al., 1996). ICAM-1 expression on neuro-vasculature was found to strongly correlate with relapses in RRMS patients, while the blood levels of its soluble form is associated with the degree of BBB impairment and disease activity in MS subjects (Cannella et al., 1990; Hartung et al., 1995). Although VLA-4 can also bind fibronectin with low affinity, its high affinity ligand VCAM-1 is believed to have a central role in encephalitogenic T cell migration into the CNS, as suggested by the strong blocking effect of anti-α4 integrin and anti-VCAM-1 antibodies on EAE development, corroborated by the therapeutic success of Natalizumab in MS patients (Yednock et al., 1992; Baron et al., 1993; Chan and Aruffo, 1993; Steffen et al., 1994; Miller et al., 2003; Belachew et al., 2011). Moreover, ALCAM and its counter-ligand CD6 were correlated to an increased risk of MS development (Wagner et al., 2014), further emphasizing the relevance of CAMs in MS.

Chemokines are fundamental molecules for integrin activation and leukocyte migration into the CNS (Constantin, 2008). C-C motif chemokine ligand 19 (CCL19) and CCL21, two chemokines able to induce integrin activation, were found to be expressed on CNS endothelial cells of post-capillary venules in both healthy and EAE mice, and these data correlated to C-C chemokine receptor type 7 (CCR7) expression on perivascular T cells (Alt et al., 2002). However, cerebral expression of CCL19 and CCL21 was not found in MS subjects, suggesting that different chemokines may be responsible for triggering intravascular integrin activation during human disease (Kivisäkk et al., 2004). Several studies have instead identified a constitutive expression of C-X-C motif chemokine ligand 12 (CXCL12) on BBB endothelial cells in EAE mice and MS patients (Krumbholz et al., 2006; McCandless et al., 2006, 2008). Particularly, a CXCL12 expression gradient was detected along the abluminal BBB surface in healthy animals and during early EAE, as well as in uninflamed regions of the MS brain and in control subjects. However, a preferential luminal localization of CXCL12 was found at EAE peak and in active MS lesions, suggesting a role for this chemokine in integrin activation and intravascular leukocyte adhesion (McCandless et al., 2006, 2008). Indeed, in both MS and EAE, the intravascular expression of CXCL12 was associated to the presence of intraluminal CXCR4+ adhering leukocytes as well as to CXCR4 expression on perivascular infiltrating leukocytes (McCandless et al., 2008). Previous studies have also suggested an involvement of CXCR4 in the confinement of infiltrated T cells at the perivascular level, promoting their compartmentalization around blood vessels and preventing widespread parenchymal infiltration (McCandless et al., 2006; Siffrin et al., 2009). Moreover, CXCR7, an alternative receptor for CXCL12, is present on endothelial cells within the CNS and its expression is increased in EAE at sites of inflammatory infiltration (Cruz-Orengo et al., 2011). Notably, CXCR7 was shown to be critical in mediating CXCL12 internalization and redistribution at CNS endothelial barriers and CXCR7 inhibition ameliorated EAE and reduced leukocyte infiltration into the CNS parenchyma (Cruz-Orengo et al., 2011). CCL2, CCL4, and CCL5 were also found to be expressed on cultured brain endothelial cell suggesting that multiple chemokines may contribute to integrin activation and subsequent arrest in inflamed CNS vessels, even if their expression on the BBB and potential role during in vivo CNS inflammation remains to be elucidated (Andjelkovic et al., 1999; Quandt and Dorovini-Zis, 2004).

During inflammatory conditions, leukocyte adhesion on vascular wall per se may lead to increased endothelial permeability (Figure 1). Accordingly, the overexpression of endothelial CAMs during neuroinflammation was shown to promote not only leukocyte adhesion, but also BBB dysfunction and increased permeability. Indeed, ICAM-1 cross-linking leads to reorganization of endothelial cytoskeleton and tight junctions phosphorylation and destabilization, favoring leukocyte adhesion and increasing BBB leakage (Durieu-Trautmann et al., 1994; Adamson et al., 1999). In this context, Rho/ROCK and Rac downstream signals are central modulators of endothelial alterations induced by ICAM-1 and VCAM-1 crosslinking during leukocyte adhesive contacts. Particularly, ICAM-1 engagement elicits RhoA activation and intracellular calcium increase with subsequent reactive oxygen species (ROS) production, actomyosin contraction, and formation of stress fibers (Etienne et al., 1998; Etienne-Manneville et al., 2000; Wang and Doerschuk, 2000). On the other hand, VCAM-1 clustering triggers Rac1 activation and intracellular calcium release, allowing ROS generation, transient disruption of adherent junctions and focal adhesion formation (Lorenzon et al., 1998; Matheny et al., 2000; van Wetering et al., 2003). The relevance of RhoA signaling was confirmed in vivo, since its blockade attenuated EAE severity by reducing leukocyte migration into the CNS (Greenwood et al., 2003). This activating signaling within the cerebral endothelium deserves further investigation as a potential molecular target to prevent leukocyte CNS invasion and safeguard BBB integrity in MS. In addition to ICAM-1 and VCAM-1, ALCAM was also found to be present on CNS endothelium during MS and EAE and its expression was associated to T cell migration (Cayrol et al., 2008). However, other studies performed in a chronic EAE model demonstrated that ALCAM maintains BBB integrity by controlling tight junction stability (Lécuyer et al., 2017), suggesting a dual function of ALCAM in neuroinflammation as well as BBB homeostasis (Cayrol et al., 2008; Lyck et al., 2017). Similarly, it was shown that PECAM-1 favors T cell diapedesis across the BBB during neuroinflammation, but also stabilizes BBB integrity (Graesser et al., 2002; Wimmer et al., 2019). Together, these findings suggest that, in addition to the inhibition of leukocyte trafficking into the CNS, BBB stabilization and tight junctions recovery may also represent a therapeutic strategy in brain inflammatory and demyelinating diseases (Spencer et al., 2018).

Multiple sclerosis has been described as a T cell-mediated autoimmune disease and its widely used model, EAE, is characterized by a Th1 and Th17-driven autoimmune and neuroinflammatory process (Segal, 2019). The study of recruitment kinetics of Th1 and Th17 cells into the CNS during EAE led to the publication of some contrasting reports, probably due to the in vivo phenotype plasticity of these cells (Geginat et al., 2014). Whereas some EAE studies indicated Th1 lymphocytes as the earliest CD4+ T cells infiltrating the CNS, followed by secondary Th17 migration (O’Connor et al., 2008), more recent findings suggested that Th17 are the pioneer lymphocytes with a later accumulation of Th1 cells during EAE development (Murphy et al., 2010). Accordingly, in vitro migration studies support a preferential capacity of Th17 cells to migrate across the BBB when compared to Th1 lymphocytes (Kebir et al., 2007), but the precise kinetics of Th cell infiltration into the CNS in MS and EAE still remains a matter of debate.

To date, Natalizumab, a monoclonal antibody that blocks activated T cell trafficking targeting the α₄ chain of VLA-4 integrin, is one of the most efficient treatments for MS (Polman et al., 2006), demonstrating the involvement of VLA-4 in leukocyte migration into the CNS (Schneider-Hohendorf et al., 2014). However, while the blockade of α₄ integrin completely abolishes Th1 recruitment and partially reduces Th17 cell accumulation in the spinal cord of EAE mice, it fails to prevent Th17 lymphocyte entrance into the cerebrum in an experimental EAE model with atypical brain manifestations (Rothhammer et al., 2011). Under flow adhesion assays using an in vitro BBB model confirmed that VCAM-1, the main VLA-4 ligand, is required for the arrest of murine Th1 cells (Steiner et al., 2010). β2 integrins have also a role in firm adhesion and crawling of Th1 lymphocytes on the BBB in in vitro models, whereas LFA-1 expression on human Th17 cells has a critical contribution to transmigration across interferon (IFN)-γ activated BBB, suggesting that both VLA-4 and LFA-1 integrins control Th lymphocyte trafficking into the brain in MS (Kebir et al., 2009; Steiner et al., 2010). Intriguingly, in MS patients treated with Natalizumab, Th cells overcome α₄ blockade by upregulating alternative adhesion receptors, such as PSGL-1 and melanoma CAM (MCAM) (Schneider-Hohendorf et al., 2014). MCAM binds to laminin 411, a component of the endothelial basement membrane, facilitating T lymphocyte penetration into the parenchyma during EAE (Sixt et al., 2001; Wu et al., 2009). Recently, the dual immunoglobulin domain containing CAM (DICAM) was shown to be preferentially expressed on Th17 cells contributing to their migration and pathogenic capacity during EAE. Interestingly, DICAM is particularly expressed on circulating CD4+ T lymphocytes of MS patients and, together with its ligand α_Vβ₃, is upregulated on lesion-associated BBB, further suggesting a role for this molecules in Th cell migration during EAE and MS (Du et al., 2016; Charabati et al., 2022).

Interferon-γ, the main Th1-related cytokine, is massively produced during neuroinflammatory conditions. The analysis of blood, CSF, and brain tissue samples from MS patients clearly supported the pathogenic role of Th1 cells and their signature cytokine in disease development. Particularly, increased frequency of IFN-γ-producing myelin-reactive T lymphocytes in the blood of MS subjects correlated with the active phase of disease and the worsening of neurological symptoms (Arbour et al., 2003; Moldovan et al., 2003). IFN-γ+ T cells were also found to be enriched in the CSF of MS patients compared to healthy controls (Wallström et al., 2000). Accordingly, IFN-γ levels were significantly elevated in the blood, CSF and CNS lesions of subjects with MS (Mycko et al., 2003; Arellano et al., 2017). The pathogenic effect of IFN-γ is corroborated by an early pilot study evaluating the efficacy of recombinant IFN-γ administration to RRMS patients, who showed a significantly higher exacerbation rate with respect to pre- and post-treatment rates (Panitch et al., 1987).

The overall picture emerging from in vitro and in vivo observations is that IFN-γ alone is able to regulate the expression of several surface molecules on brain endothelium including MHCI, MHCII, programmed death-ligand (PD-L)1 VCAM-1, mucosal vascular addressin (Mad)CAM-1 and ICAM-1 (Brown et al., 2007; Sonar et al., 2017; Figure 2 and Table 1). Interestingly, time-lapse confocal microscopy experiments showed that IFN-γ, rather than inducing ICAM-1 overexpression, determined a rapid re-localization of this molecule from the basal to the apical side of endothelial cells, potentially promoting leukocyte adhesion (Sonar et al., 2017). Notably, in vitro and in vivo data suggested that IFN-γ promote BBB leakage by inducing STAT-1 expression and cytoskeleton remodeling affecting tight junction protein organization (Sonar et al., 2017; Bonney et al., 2019). Also, exposure to IFN-γ promotes endothelial permeability through zonula occludens (ZO)-1 and claudin (CLDN) 5 delocalization and adherent junction molecule VE-cadherin perturbation in an in vitro system using bEnd3.1 cell line, and these changes were associated to a relevant increase of actin stress fibers and cytoskeletal contraction (Sonar et al., 2017; Liu et al., 2018). In vitro inhibition of Rho kinase (ROCK) was sufficient to restore the integrity of brain endothelial cells during IFN-γ treatment by blocking the junctional localization of phosphorylated myosin light chain (Bonney et al., 2019), a phenomenon linked to weakened cell-cell interactions (Hirano and Hirano, 2016). In line with this, in vivo ROCK inhibition induced the upregulation of occludin and ZO-1 tight junction proteins on cerebral vessels, making this drug a valuable candidate to tackle BBB breakdown in MS (Yan et al., 2019).

However, the detrimental effect of IFN-γ on BBB integrity was challenged by Ni et al. (2014), whose findings suggested a role for IFN-γ in inducing the clinching of brain endothelial junctions. In particular, these authors showed that IFN-γ treatment of bEnd3.1 cells enhances their paracellular tightness in a dose-dependent manner and dramatically reduces splenocyte transmigration, with CLDN 5 being essential to link cytokine-dependent signaling to junctional strengthening (Ni et al., 2014). Additionally, exposure of murine brain endothelial cells to IFN-γ decreased CXCR7 expression and led to a reduction of CXCL12 internalization (Cruz-Orengo et al., 2011). Interestingly, blocking CXCL12 sequestration may allow CXCL12 to tether T cells to the perivascular space, preventing infiltration and having a beneficial effect on EAE (Cruz-Orengo et al., 2011). In support of these results, EAE induction in mice expressing IFN-γ receptor exclusively on endothelial cells was sufficient to significantly mitigate brain inflammation, resulting in a lower incidence of atypical symptoms, compared to IFN-γ receptor deficient animals (Ni et al., 2014). Notably, studies performed almost three decades ago showed that IFN-γ signaling plays a protective role during neuroinflammation, as proved by exacerbation of EAE symptoms in IFN-γ^–/– mice (Ferber et al., 1996; Willenborg et al., 1996) or disease alleviation upon IFN-γ administration, supporting the idea that IFN-γ promotes BBB stability (Voorthuis et al., 1990; Naves et al., 2013). However, more recent studies showed that IFN-γ signaling is required to trigger spinal cord inflammation and classic EAE, whereas IFN-γ receptor deficiency promotes the development of atypical EAE symptoms, associated to Th17 cell migration in the brainstem and cerebellum (Lees et al., 2008; Stromnes et al., 2008; Pierson et al., 2012). Together, these data suggest that regional CNS responses to IFN-γ, including BBB inflammation and breakdown, determine lesion localization patterns during EAE development.

IL-17A, IL-17F, IL-22, and IL-26 represent a group of cytokines produced by Th17 cells and are classically associated to the CNS inflammatory milieu during MS (Figure 2 and Table 1). Indeed, increased levels of IL-17 mRNA and IL-17 secreting CD4+ T cells were detected in the blood, CSF, and brain lesions of MS patients (Matusevicius et al., 1999; Lock et al., 2002; Tzartos et al., 2008; Durelli et al., 2009). Particularly, IL-17A is elevated in the CSF of RRMS subjects and correlates with CSF/serum albumin quotient, an index of BBB permeability. In this regard, treating RRMS patients with secukinumab, an anti-IL-17A antibody, reduced MRI lesion activity compared to placebo, supporting a role for IL-17 in disease development (Havrdová et al., 2016).

Cytokines produced by Th17 cells may directly affect BBB integrity and function. Indeed, in vitro stimulation of human brain endothelial cells (HBEC) with IL-17A diminishes the expression of tight junction protein-encoding genes Pecam1, Cdh5, and Tjp1 and leads to the disassembly of ZO-1 and reduction of occludin and CLDN 5 protein expression (Rahman et al., 2018; Setiadi et al., 2019). Similarly, in vitro data obtained using the mouse bEnd.3 cell line showed that IL-17A activates endothelial contractile machinery through a ROS-dependent pathway, inducing occludin downregulation, ZO-1 disorganization, and ICAM-1 upregulation (Huppert et al., 2010) (Figure 2 and Table 1). In line with this, inhibition of IL-17A in a model of EAE led to a reduced leukocyte infiltration of the CNS, lower CNS oxidative stress and milder clinical disease, further demonstrating a role for IL-17 in BBB breakdown (Huppert et al., 2010; Setiadi et al., 2019). Moreover, IL-17, conversely to IFN-γ, induces upregulation of CXCR7 on primary mouse brain endothelial cells (Cruz-Orengo et al., 2011) potentially promoting leukocyte migration, as also suggested by in vivo data demonstrating that activation of CXCR7 reduces abluminal CXCL12 concentration and leads to an increased leukocyte entry in the inflamed CNS during EAE (Cruz-Orengo et al., 2011). IL-22 produced by Th17 cells can also affect BBB permeability, as shown by studies performed with primary endothelial cells from human CNS tissue specimens (Kebir et al., 2007; Figure 2 and Table 1). Remarkably, the receptors for IL-17 and IL-22 are absent in CNS samples from healthy human subjects, but they are strongly expressed on CNS endothelial cells within MS lesions, suggesting that the human BBB is highly responsive to the detrimental effect of these two cytokines (Kebir et al., 2007). In agreement with these data, both IL-17 and IL-22 can boost Th cell transmigration across brain endothelium in vitro, although IL-17 is more effective in inducing the production of endothelial pro-inflammatory molecules (Kebir et al., 2007; Wojkowska et al., 2017). However, differently from IL-17, which downregulates tight junction proteins, the mechanisms underlying IL-22 effect on the BBB remain uncertain.

Unlike IL-17 and IL-22, IL-26 showed a protective effect on BBB integrity (Broux et al., 2020; Figure 2 and Table 1). Brain endothelial cells treated with IL-26 showed a general downregulation of pro-inflammatory pathways and upregulation of Tjp1, Ocln, and Cldn18 gene transcripts. Additionally, IL-26 treatment of EAE mice resulted in a reduced infiltration of pathogenic T lymphocytes and disease attenuation, clearly demonstrating a protective effect of IL-26 at the BBB level (Broux et al., 2020). Overall, the existing literature on MS and EAE point to Th17 cytokines as key players in modulating BBB permeability and leukocyte trafficking during CNS autoimmune inflammatory conditions.

The discovery of lymphoid follicles in the leptomeninges of MS patients and EAE mice redirected the attention from the BBB dysfunction to meningeal inflammation and its potential role in disease pathogenesis (Magliozzi et al., 2004; Serafini et al., 2004). In RRMS and PPMS, the leptomeningeal area seems to be predominantly characterized by a chaotic accumulation of inflammatory cells, whereas in SPMS, immune cells were found in highly organized tertiary lymphoid structures, pointing to leptomeningeal inflammation as a pathogenic driver for disease development (Choi et al., 2012; Howell et al., 2015; Magliozzi et al., 2018). As observed in EAE, MRI investigations corroborated by histopathology studies highlighted a direct correlation between leptomeningeal contrast enhancement, subarachnoid space (SAS)-localized inflammation and subpial cortical pathology in MS patients (Lucchinetti et al., 2011; Choi et al., 2012; Howell et al., 2015; Harrison et al., 2017; Magliozzi et al., 2018; Bergsland et al., 2019; Zurawski et al., 2020). At early stages of human disease, meningeal inflammation is mainly detected in the proximity of areas with altered BBB, cortical demyelination, and gray matter lesions and seems to develop before the emergence of white matter plaques (Lucchinetti et al., 2011). Similarly, inflammatory features appear in the cerebral and spinal cord meninges before EAE onset (Brown and Sawchenko, 2007; Shrestha et al., 2017). Finally, a recent ultra-high field MRI study demonstrated that the cerebral leptomeningeal contrast enhancement magnitude, which reaches a peak of intensity during the acute phase of EAE, is associated to the clinical signs and high inflammatory cell density in EAE mice (Pol et al., 2019). Together, these data underline the importance of the leptomeningeal BCSFB as one of the earliest routes of entry for peripheral immune cells in both MS and EAE.

The SAS is a surgically easily accessible area, and several in vivo imaging studies have described the central role of pial vessels during autoreactive T cell recruitment into the CNS (Zenaro et al., 2013; Rossi and Constantin, 2016). Given their permissive endothelial layer and the lack of astrocyte sheathes typical of the BBB, leptomeningeal pial vessels represent a preferential CNS gateway for pioneer circulating autoreactive CD4+ T cells during the preclinical phase of EAE (Bartholomäus et al., 2009). Th cells reach the maximum accumulation in the leptomeninges at EAE peak and migrate in the SAS attracted by the CXCL9-11 and CCL5 chemokines (Bartholomäus et al., 2009; Lodygin et al., 2013; Schläger et al., 2016). The endothelium of inflamed pial vessels expresses P- and E-selectins (Kerfoot and Kubes, 2002; Piccio et al., 2002; Kivisäkk et al., 2003). Particularly, P-selectin was found to be a key molecule involved in leukocyte adhesive interactions in brain pial vessels during the early stages of EAE (Kerfoot and Kubes, 2002; Kerfoot et al., 2006). In agreement with these data, intravital microscopy studies performed in the leptomeningeal microcirculation have shown that a properly glycosylated PSGL-1 is the major P-selectin ligand for efficient tethering and rolling of activated T cells (Piccio et al., 2002, 2005). Likewise, CD8+ T cells from MS patients preferentially exploit PSGL-1 for their rolling on inflamed pial vessels endothelium, further indicating a role for PSGL-1 in activated T cell adhesion in brain pial vessels (Battistini et al., 2003). Interestingly, T cell immunoglobulin and mucin domain 1 (TIM-1) was shown to cooperate with PSGL-1 in the mediation of Th1 and Th17 tethering and rolling on inflamed pial vasculature, contributing to the CNS infiltration of autoreactive T cells and to the development of EAE (Angiari and Constantin, 2014; Angiari et al., 2014). Moreover, integrins VLA-4 and LFA-1 were found to be involved in rolling and G protein-mediated firm adhesion of peripheral leukocytes on inflamed endothelium of cerebral pial vessels, potentially leading to vascular disfunction and leakage (Kerfoot and Kubes, 2002; Piccio et al., 2002; Kerfoot et al., 2006; Figure 1).

Once migrated in the leptomeninges, autoreactive CD4+ T cells are reactivated by local antigen presenting cells and directed to the underlying CNS parenchyma (Bartholomäus et al., 2009; Lodygin et al., 2013; Schläger et al., 2016). However, how autoreactive T cells access the parenchyma from the SAS is still unclear. In this respect, it was suggested that, similar to the blood circulation, CSF dynamics could operate as a physical transport mechanism facilitating cell migration into the inflamed areas of the CNS (Schläger et al., 2016). A potential molecular mechanism controlling T cell migration into the parenchyma could be mediated by laminin 111 (α1β1γ1 or laminin 1), which is preferentially expressed by the leptomeningeal cells and in close contact with astrocyte end feet (Sixt et al., 2001; Agrawal et al., 2006). Also, glia limitans breakdown and microglia activation, both potentially due to meningeal inflammation, may promote autoreactive T cell entry in the cortex and spinal cord during MS and EAE (Gardner et al., 2013; Magliozzi et al., 2018, 2019; James et al., 2020). Furthermore, astrocytes can also contribute to T cell trafficking during EAE. In this context, Th1 and Th17 cytokines induce region-specific astrocyte expression of VCAM-1 and CXCR7, modulating local astrocyte-dependent immune cell trafficking into the CNS during EAE (Williams et al., 2020).

Due to its location within the cerebral ventricles and the presence of fenestrated capillaries, the choroid plexus is another important gateway involved in CNS immunosurveillance (Nathanson and Chun, 1989). However, the poor vascular selectivity is compensated by the expression of apical tight junctions and basolateral adherens junctions in the epithelial layer of ependymal cells that strictly controls immune cell trafficking from the choroid plexus stroma into the CSF (Maxwell and Pease, 1956; van Deurs and Koehler, 1979; Lippoldt et al., 2000; Ayub et al., 2021). Indeed, the murine choroid plexus epithelium constitutively expresses ICAM-1, VCAM-1, MadCAM, P-selectin, and CCL20, which may explain, at least in part, why memory CD4+ T cells are the most represented leukocytes within the CSF under physiological conditions (Steffen et al., 1996; Wolburg et al., 1999; Kivisäkk et al., 2003; Reboldi et al., 2009; de Graaf et al., 2011; Figure 3). Additionally, recent in vitro studies suggest that, once migrated in the choroid plexus stroma, human CD4+ T cells could further migrate into the CSF by binding ICAM-1 expressed at the luminal side of epithelial cells (Nishihara et al., 2020).

The onset of CNS-targeted autoimmune responses during EAE induces a prominent upregulation of ICAM-1 and VCAM-1 on choroid plexus epithelium, further favoring leukocytes trafficking through this anatomical route (Steffen et al., 1996; Figure 3). When exposed to IFN-γ, the murine choroid plexus epithelium becomes activated, producing chemokines (CCL2, CCL5, CXCL9, CXCL10, CX3CL1, and M-CSF) and upregulating cell adhesion receptors (ICAM-1 and VCAM-1) and MHC-II molecules (Figure 3; Kunis et al., 2013). On the other hand, exposing choroid plexus epithelial cells to IL-17 results in further upregulation of CCL20, a chemokine involved in the recruitment of CCR6+ encephalitogenic Th17 cells during early EAE (Reboldi et al., 2009; Kunis et al., 2013; Figure 3). Notably, CCR6+ CD25- CD4+ pro-inflammatory lymphocytes are present in the CSF of MS patients starting from the earliest disease episode, supporting the pivotal involvement of the evolutionarily conserved CCL20-CCR6 axis in the establishment of a pathological loop amplifying local CNS immune responses (Reboldi et al., 2009; Kara et al., 2015). Interestingly, CD4+ memory T cells expressing CXCR3, a receptor highly present on Th1 cells, were found enriched in the CSF compared to circulating levels (Kivisäkk et al., 2002). However, no differences in CXCR3 expression were detected between MS and control subjects, suggesting a potential role for this chemokine receptor in T cell intrathecal residency (Kivisäkk et al., 2002).

The analysis of postmortem MS brain samples revealed an impairment of epithelial CLDN 3 at the level of the choroid plexus. Accordingly, mice lacking CLDN 3 displayed an earlier EAE onset due to higher levels of infiltrated leukocytes in the CSF (Kooij et al., 2014). The early accumulation of activated T cells in the choroid plexus during EAE suggests a crucial role for this CNS compartment in disease pathogenesis (Brown and Sawchenko, 2007; Murugesan et al., 2012). Throughout the EAE course, the choroid plexus appears enlarged, morphologically altered, and expresses immune-related genes including proinflammatory cytokines, co-stimulatory molecules for T cells and adhesion molecules, suggesting its potential capacity to recall and activate migrating autoreactive T cells (Murugesan et al., 2012). Also, high-throughput studies showed that the majority of T cell receptor repertoire inside the choroid plexus is specific for CNS antigens in mice that were immunized with spinal cord homogenates (Baruch et al., 2013). In addition, recent data demonstrated that the choroid plexus is characterized by the constitutive presence of CNS antigens and antigen presenting cells, that can rapidly react to inflammatory signals, promoting T cell proliferation and a second wave of leukocyte migration into the CNS (Strominger et al., 2018). In agreement with studies performed in mouse models, results obtained on post-mortem choroid plexus samples found increased T cell migration and HLA-DR expression on the choroid plexus stroma of patients with MS compared to control subjects, suggesting that choroid plexus may represent a site for lymphocyte entry in the CSF and antigen presentation also during human disease (Vercellino et al., 2008).

Recent MRI data showed a significant increase of choroid plexus volume in EAE mice compared to untreated animals at baseline, as well as in RRMS patients compared to healthy controls, correlating to acute disease episodes and disability worsening (Fleischer et al., 2021). Notably, Natalizumab treatment proved effective in reducing choroid plexus enlargement in MS patients at follow-up, while dimethyl fumarate, which does not interfere with leukocyte recruitment into the CNS, had no effect (Fleischer et al., 2021). Altogether, these data suggest a role for leukocyte migration through the choroid plexus in promoting MS pathology.