Blockade of MCAM/CD146 impedes CNS infiltration of T cells over the choroid plexus

Background Very late antigen 4 (VLA-4; integrin α4β1) is critical for transmigration of T helper (TH) 1 cells into the central nervous system (CNS) under inflammatory conditions such as multiple sclerosis (MS). We have previously shown that VLA-4 and melanoma cell adhesion molecule (MCAM) are important for trans-endothelial migration of human TH17 cells in vitro and here investigate their contribution to pathogenic CNS inflammation. Methods Antibody blockade of VLA-4 and MCAM is assessed in murine models of CNS inflammation in conjunction with conditional ablation of α4-integrin expression in T cells. Effects of VLA-4 and MCAM blockade on lymphocyte migration are further investigated in the human system via in vitro T cell transmigration assays. Results Compared to the broad effects of VLA-4 blockade on encephalitogenic T cell migration over endothelial barriers, MCAM blockade impeded encephalitogenic T cell migration in murine models of MS that especially depend on CNS migration across the choroid plexus (CP). In transgenic mice lacking T cell α4-integrin expression (CD4::Itga4−/−), MCAM blockade delayed disease onset. Migration of MCAM-expressing T cells through the CP into the CNS was restricted, where laminin 411 (composed of α4, β1, γ1 chains), the proposed major ligand of MCAM, is detected in the endothelial basement membranes of murine CP tissue. This finding was translated to the human system; blockade of MCAM with a therapeutic antibody reduced in vitro transmigration of MCAM-expressing T cells across a human fibroblast-derived extracellular matrix layer and a brain-derived endothelial monolayer, both expressing laminin α4. Laminin α4 was further detected in situ in CP endothelial-basement membranes in MS patients’ brain tissue. Conclusions Our findings suggest that MCAM-laminin 411 interactions facilitate trans-endothelial migration of MCAM-expressing T cells into the CNS, which seems to be highly relevant to migration via the CP and to potential future clinical applications in neuroinflammatory disorders. Electronic supplementary material The online version of this article (10.1186/s12974-018-1276-4) contains supplementary material, which is available to authorized users.


Background
Multiple sclerosis (MS), a chronic disorder of the central nervous system (CNS), is characterized by inflammatory lesions at early disease stages that are caused by the infiltration of autoreactive immune cells into the CNS [1,2]. The inflammatory infiltrates, containing pathogenic T cells (T helper 1 (T H 1) and T H 17 cells), lead to demyelination and neuronal degeneration resulting in serious physical disability [3,4].
Immune cell infiltration from the periphery into the CNS, characteristic of the common relapsing-remitting form of MS (RRMS), has been successfully targeted by new therapies. In particular, the monoclonal antibody natalizumab, which recognizes the integrin α4 subunit of very late antigen-4 (VLA-4, integrin α4β1, CD49d/ CD29), has proven highly effective in reducing disease progression by preventing leukocyte migration into the CNS [5,6]. VLA-4 plays a key role in the entry of encephalitogenic T cells into the CNS by mediating the initial rolling and adhesion steps of transmigration through interaction with its receptor, vascular cell adhesion molecule-1 (VCAM-1), expressed on endothelial cells of post-capillary venules upon inflammation [7]. Even though VLA-4 has been shown to play a crucial role in leukocyte trafficking over the blood-brain barrier (BBB), animal studies revealed that a subpopulation of T H 17 cells, which represent a major pathogenic T cell population in murine experimental autoimmune encephalomyelitis (EAE) [8][9][10], can invade the CNS in a VLA-4 independent manner. Recent studies suggest that T H 17 cells mainly rely on leukocyte function-associated molecule-1 (LFA-1) mediated firm adhesion to endothelial intercellular adhesion molecule-1 (ICAM-1) to penetrate the choroid plexus (CP) and enter into the CNS parenchyma or the cerebrospinal fluid (CSF) space [11][12][13]. However, LFA-1 is primarily involved in the firm adhesion to endothelium and subsequent para-endothelial migration, and not in the initial rolling and adhesive steps [14]. This poses the question of how the initial adhesive steps are mediated in the encephalitogenic T H 17 subpopulation and raises the possibility that different immune cell-adhesion receptor interactions are employed, depending on the route of entry into the brain and the cellular barrier involved. Three important barriers restrict access of circulating immune cells to the CNS parenchyma: the endothelial BBB in CNS post-capillary venules, the epithelial-cerebrospinal fluid barrier (BCSFB) at the CP, and the arachnoidea formed by leptomeningeal cells [14][15][16].
Apart from LFA-1-ICAM-1-dependent penetration of the CP by T H 17 cells, we have recently shown that adhesion of T H 17 cells to an in vitro model of the BBB in the absence of VLA-4 involves melanoma cell adhesion molecule (MCAM; CD146) [17]. MCAM is an extracellular adhesion molecule broadly expressed on mesenchymally derived tissues [18]; in humans it has also been reported to be expressed on a subpopulation of IL-17-secreting CD4 + and CD8 + T cells (T H 17 and Tc17 cells, respectively) [18][19][20][21][22][23][24][25]. EAE experiments performed in endothelial MCAM knockout mice have shown reduced disease severity and reduced T H 17 infiltration into the CNS, supporting a role for this molecule in the entry of encephalitogenic T cells into the brain [26]. Laminin 411 (composed of α4, β1, and γ1 chains), a major component of endothelial basement membranes [27], has been identified as a ligand for MCAM, and MCAM-laminin 411 interactions have been proposed to facilitate penetration of the BBB [22]. However, whether MCAM can contribute to T H 17 penetration of other CNS barriers remains unclear.
The focus of the current study was, therefore, to elucidate the pathogenic relevance of MCAM compared to VLA-4 in CNS inflammation in vivo by using different murine models of MS (active and spontaneous experimental autoimmune encephalitis (EAE)). Using antibody-mediated blocking of VLA-4 and/or MCAM and T cell-specific ablation of α4-integrin expression, we show here that VLA-4 blockade abrogates the encephalitogenic potential of T cells in a wide-ranging way by blocking migration over endothelial barriers, whereas MCAM targeting primarily restricts migration of T H 17 cells over the CP into the CNS by inhibiting MCAM-laminin α4 interactions at the endothelial CP layer.
All animal experiments were approved by and conducted in accordance with the laws and regulations of the regulatory authorities for animal care and scientific use in North Rhine-Westphalia, Germany (TVA-number 84-02.04.2014.A075).

Antibody treatment for in vivo blocking
The blocking antibodies anti-mMCAM (clone 15) (described in [22]) and α4 integrin neutralizing antibody (clone: PS/2, BioXCell, New Hampshire, USA) as well as the appropriate isotype control antibodies (rat IgG1, clone: HRPN and rat IgG2b, clone: LTF-2; both BioXCell) were used at a concentration of 10 mg/kg body weight. Mice were treated every other day with i.p. injections of the respective antibody from the indicated day on.

Cell culture and transmigration assays
Fibroblasts originated from primary human choroid plexus epithelial cells (labeled as HCPEpiC) that were purchased from ScienCell Research Laboratories (Carlsbad, CA, USA) and were cultured in Epithelial Cell Medium (EpiCM) supplemented with 2% fetal bovine serum (FBS; ProVitro, Berlin, Germany) on poly-L-lysine (2 μg/cm 2 ; ScienCell Research Laboratories). Cultured cells were characterized by real-time quantitative PCR for expression of endothelial and epithelial markers as well as expression of laminin α4.

Blocking antibodies for transmigration assays
The blocking antibodies anti-VLA4 (anti-CD49d/natalizumab, Tysabri; Biogen Idec, Cambridge, MA, USA) and anti-MCAM (clone PRX003; Prothena Biosciences Inc., Dublin, Ireland) were used at a concentration of 10 μg/ ml. CD4 + lymphocytes were pre-incubated for 30 min with the antibodies diluted in PBS before washing out of the blocking antibodies and application to the transmigration assays.

Real-time quantitative PCR
RNA was extracted from confluent monolayers of HCPEpiC-derived fibroblasts or HBMEC using TRIzol reagent (Invitrogen). cDNA was synthesized from 1 μg of total RNA using a standard protocol with random hexamer primers (ThermoScientifc). Real-time qPCR was performed in a StepOnePlus cycler (Applied Biosystems, Darmstadt, Germany) employing endogen control primers for 18sRNA as well as a TaqMan Gene Expression Assays specific for human laminin α4, cytokeratin 18, VE-cadherin, vimentin, or PECAM1 (Applied Biosystems, Darmstadt, Germany).
For human CNS tissue, autopsy material from subjects with MS was obtained from the Netherlands Brain Bank (NBB). Immunofluorescence stainings were performed on 10 μm cryosections and paraformaldehyde-fixed confluent monolayers of fibroblasts (derived from primary HCPE-piC) with a monoclonal anti-human laminin α4 antibody (clone 3D12 [39]) and an AlexaFluor488-labeled goatanti-mouse secondary antibody. In case of the fibroblastic monolayers, an amplification step with TSA Plus Biotin kit (PerkinElmer, Waltham, MA, USA) was performed according to the manufacturer's instructions. Tissues were analyzed using a Zeiss LSM700 confocal microscope. Images were analyzed using Volocity 6.3 software (ImproVision).

Study approval
Studies on human samples were approved by the local ethics committee (University of Muenster: Ethik-Kommission der Ärztekammer Westfalen-Lippe und der Medizinischen Fakultät der Westfälischen Wilhelms-Universität, registration number: 2010-245-f-S). All experiments were performed according to the Declaration of Helsinki.

Statistical analysis
All values are presented as mean ± standard error of the mean (SEM). Statistical significance was determined using Student's t test in the case of normally distributed data; otherwise, a Mann-Whitney test was performed. For comparisons between data of CD4 + subpopulations from the same donor, the paired Student's t test was used. The two-way ANOVA followed by Bonferroni post-test was applied for EAE experiments. P values were considered significant if < 0.05 in all cases. Statistical analyses were carried out using Prism (Version 5, GraphPad, San Diego, CA, USA).

Results and discussion
MCAM blockade delays EAE disease onset in T cell-specific integrin α4 knockout mice (CD4::Itga4 −/− ) To investigate whether encephalitogenic CD4 + T cells without functionally active VLA-4 can employ MCAM to migrate into the CNS, we induced EAE in transgenic mice lacking α4-integrin expression on T cells (CD4::Itga4 −/− ) in the presence or absence of a functionblocking antibody to MCAM. MOG 35-55 immunized CD4::Itga4 −/− mice developed a mild and clinically atypical EAE, as previously reported [12] (Fig. 1a). Application of anti-MCAM to CD4::Itga4 −/− mice every second day after MOG  immunization commencing on day 0 resulted in a delay in disease onset (Fig. 1a, b and Table 1), but no change in disease severity. This suggests that in the absence of α4-integrin, MCAM blocking delays but does not completely prevent CD4 + T H 17 entry into the CNS parenchyma. Former studies have shown that the chemokine receptor CCR6, which like MCAM is selectively upregulated in polarized murine T H 17 cells [10,22], is essential for CNS trafficking of T H 17 cells to the CP. CCR6 + T H 17 + cells are considered to represent the first wave of leukocyte infiltration required for the subsequent recruitment of other inflammatory cells (T H 1 cells) across post-capillary venules and induction of disease symptoms [10]. T H 17 cells are further able to access the CNS parenchyma independently of VLA-4 and MOG  immunized mice lacking α4-integrin expression on T cells (CD4::Itga4 −/− ) exhibit marked immune cell infiltration within the plexus epithelium [12], indicating that in the absence of α4-integrin encephalitogenic T cells are recruited via the choroid plexus into the CNS. These findings and the observed delay in EAE onset upon MCAM blockade suggest that, besides CCR6, MCAM might be crucial for the initial infiltration of encephalitogenic T H 17 cells via the CP. Further, our recent observation of a specific accumulation of MCAM-expressing lymphocytes in the CSF of patients with MS that are under long-term natalizumab therapy (VLA-4 blockade) [17] suggests also that MCAM-expressing lymphocytes migrate over the CP into the CNS.
Blockade of MCAM might possibly impact migration into the CNS via blocking of endothelial MCAM, as entry of encephalitogenic T cells into the brain is reduced in endothelial MCAM knockout mice [26]; however, flow cytometry performed on day 22 post immunization with an anti-mMCAM, antibody recognizing an epitope distinct from the therapeutic antibody, showed that administration of the MCAM neutralizing antibody resulted in lower expression of MCAM on natural killer (NK) cells (Additional file 1: Figure S1a) and CD4 + T cells both in the periphery and the CNS (quantification is shown in Fig. 1c and representative flow cytometric analysis in Fig. 1e), suggesting shedding or downregulation of MCAM on peripheral immune cells.
In mice, contrary to humans, MCAM is strongly expressed on NK cells [40], which might be a reason for the more pronounced downregulation in the percentages of MCAM + NK cells compared to the slight effect on MCAM + T cells. In CD4::Itga4 −/− mice-independent of the treatment, percentages of MCAM-expressing CD4 + cells in the brain correlated with the clinical score (Spearman r = 0.7513; P = 0.01) (Fig. 1d), whereas MCAM-expressing NK cell proportions in the brain did not correlate with clinical score (Spearman r = − 0.4427; P = 0.2) (Additional file 1: Figure S1b), indicating that the severity of EAE depends on the amount of MCAM + CD4 + T cells in the brain. As CD4 + T cells are the primary pathogenic T cells in EAE [41], this suggests that antibody blocking of MCAM affects either MCAM-expressing T H 17 cell numbers in the periphery or their entry into the CNS and, thereby, EAE induction.

Blockade of VLA-4 and MCAM in a spontaneous mouse model of MS ("Devic" mice)
To assess whether anti-MCAM effects are specific to T H 17 versus T H 1-induced inflammation, we utilized a spontaneous mouse model of MS ("Devic" mice), which shows largely optic nerve and spinal cord leukocyte infiltration and is mainly characterized by a T H 1 cytokine profile [31,32,42]. Devic mice are a double-transgenic mouse strain, expressing T and B cell receptors that recognize the same autoantigen (MOG), in which around 50% of the double-transgenic offspring spontaneously develop a paralytic disease reminiscent of a subset of neuromyelitis optica patients. Devic mice were treated every second day from day 16 after birth with either anti-VLA4, anti-MCAM, or both antibodies; control mice were treated with the appropriate isotype control antibodies. VLA-4 blockade largely protected mice from the development of paralytic disease (Fig. 2a and Table 1), whereas blockade of MCAM resulted in reduced disease incidence and severity ( Fig. 2b and Table 1). Blockade of both VLA-4 and MCAM did not show an additive effect compared to VLA-4 blockade alone ( Fig. 2c and Table 1). These results are consistent with the reported high expression of VLA-4 on T H 1 cells and the comparatively low VLA-4 levels on MCAM high T H 17 cells [12,22]. In addition, the data suggest that the effects of anti-MCAM treatment are most pronounced in T H 17-mediated EAE and therefore affect T H 17 trafficking to the brain. Previous experiments employing murine models of MS with encephalitogenic T cells expressing VLA-4 showed that administration of a MCAM blocking antibody affected disease progression during remission [22][23][24]. However, mice with Devic's disease normally do not enter remission, which could explain why we did not observe effects of the MCAM treatment in the late stage of the disease.
In order to corroborate the influence of MCAM in CNS inflammation in the Devic model, we performed histological analyses of the CP tissue to investigate whether MCAM blockade might impact migration into the CNS via blocking of endothelial MCAM. However, endothelial MCAM expressing cells could easily be detected in the CP tissue of mice treated with anti-MCAM blocking antibody or with a synergistic blockade of MCAM and VLA-4 (Fig. 2d), suggesting that this anti-MCAM neutralizing antibody does not have any impact on endothelial MCAM and acts on peripheral immune cells.
To assess the relevance of MCAM expression on T cells for their transmigration across the CP, we performed histological stainings of laminin α4, the proposed major ligand of MCAM, in murine CP tissue. Double staining of CP tissue from healthy control mice for endothelial or epithelial markers together with the MCAM ligand laminin α4 revealed laminin α4 staining exclusively in association with the MECA32 + endothelial cells of the CP (Fig. 3a) located between the plectin + epithelial cell layers (Fig. 3b) [43,44]. Blockade of MCAM might therefore ameliorate CNS inflammation by preventing migration of MCAM expressing cells across the endothelial CP layer on their way into the CNS.
In a proof-of-concept experiment, we transferred fluorescently labeled myelin oligodendrocyte glycoproteinspecific TCR-transgenic T cells (2D2 T cells), differentiated in vitro with a protocol that induces high percentages of MCAM expressing cells [22], into C57BL/6  Figure S2; numbers of transferred cells, that were detected in the CP: mouse 1: 23; mouse 2: 19; mouse 3: 17). Staining of laminin α4 in cortical explants confirmed further the localization of the MCAM ligand to endothelial basement membranes of the CP (Fig. 3c-e).

Enhanced migration of human MCAM-expressing T cells across endothelial and extracellular matrix barriers is dependent on MCAM-laminin 411 interactions
To further investigate whether MCAM has an effect on lymphocyte migration over CP endothelial barriers, we performed in vitro T cell transmigration assays using primary human brain-derived microvascular endothelial cells (HBMECs) and choroid plexus fibroblasts (labeled as "HCPEpiC") (characterization of the cells is shown in Additional file 3: Figure S3). We have previously shown that laminin α4 can be expressed by fibroblasts in different tissues but is deposited in subjacent endothelial basement membranes [27] and that it is also expressed by cultured HBMECs [17]. Immunofluorescent staining confirmed that laminin α4 is present on cultured HCPEpiC-derived fibroblasts in vitro (Fig. 4a), and MCAM blockade indeed restricted transmigration of MCAM-expressing T cells (Fig. 4b). The proportion of MCAM + cells among CD4 + T cells was enriched in the fraction of cells that transmigrated the HCPEpiC-derived fibroblasts layer, compared to the initial (ex vivo) CD4 + T cell fraction, but was significantly reduced when the CD4 + lymphocytes were pre-incubated with a MCAM blocking antibody, either alone or together with anti-VLA-4. Moreover, analysis of fibroblastic layer following the transmigration period, revealed enrichment of MCAM + CD4 + T cells within the cell layer (Fig. 4c), suggesting adherence of MCAM + lymphocytes to laminin α4 in the deposited extracellular matrix during their migration through the cell layer. MCAM blockade led to marked accumulation of MCAM + CD4 + T cells within the cell layer (Fig. 4c), suggesting that the cells can still adhere to the fibroblastic cell layer, but require MCAM-mediated processes for efficient transmigration. This was substantiated by the results of transmigration over both TNF-α inflamed or non-inflamed HBMECs, which also revealed reduced MCAM + CD4 + T cell transmigration over an endothelial layer in presence of a MCAM blocking antibody (Fig. 4d).
To determine whether these in vitro findings are also relevant to the human in vivo system, we stained human CP tissue from one MS patient for laminin α4. We detected high laminin α4 expression in association with the endothelial layer of the CP, whereas no laminin α4 staining was detectable on the epithelial layer (Fig. 4e). The pattern of laminin α4 localization, therefore, supports the hypothesis that MCAM-laminin α4 interactions facilitate MCAM + CD4 + T cell penetration of the CP endothelial layer and thereby entry into the CNS during neuro-inflammation.
The differential effects of blocking VLA-4 and/or MCAM in active EAE and in the spontaneous EAE model suggest that targeting VLA-4 restricts most encephalitogenic T cells from migrating into the CNS, while blockade of MCAM more specifically inhibits T H 17 cell migration into the CNS via the CP endothelium. Effects of MCAM blockade are particularly pronounced in murine models of MS that depend on CNS migration across the CP, such as in the case of mice lacking α4-integrin expression on T cell. This is in accordance with our previous observation of a specific accumulation of MCAM + CD4 + T cells in the CSF of MS patients with long-term Natalizumab therapy (VLA-4 blockade) [17], suggesting that MCAM + CD4 + lymphocytes preferentially migrate over the CP into the CNS. This may be further enhanced by the high expression of the chemokine receptor CCR6, a characteristic marker of the encephalitogenic T H 17 cells [10], which binds to CCL20 expressed on the choroid plexus epithelium in mice [45] and confirmed here for human tissue (see Additional file 4: Figure S4). Leukocytes that are recruited to the CP and thereby into the CNS have to migrate through the endothelium and underlying basement membrane before they can migrate through the epithelial barrier into the CSF. Here, we show that MCAM is required for migration of an encephalitogenic MCAM + CD4 + subpopulation of T cells into the CP, as cells without MCAM showed reduced migration across both endothelial and fibroblastic laminin 411 expressing layers in vitro and reduced CNS infiltration in vivo.
Accordingly, laminin α4 was localized to basement membranes underlying the choroidal endothelium, but not choroidal epithelium, suggesting that MCAM-laminin 411 interactions mediate the initial choroidal endothelium  crossing in vivo. The differential expression of laminin isoforms as a "doorway" for leukocyte migration across endothelial of post-capillary venules in the CNS parenchyma has previously been reported [34,46,47] and, therefore, a similar concept might also apply for the choroidal endothelium. Importantly, by interfering with migration over the CP endothelium, MCAM blockade might have a specific effect on disease initiation. Others have previously reported the CP as the site of T cell entry into the CNS during immune surveillance and at first stages of neuro-inflammation [10].
However, besides the CP, there are other entry routes into the CNS, such as through the circumventricular organs or through the leptomeninges, both of which are widely considered to be important in acute inflammation and perpetuation of CNS disease. In a rat EAE model, the leptomeninges was recently identified as an important checkpoint for T cell infiltration of the CNS, where effector T cells can enter the CSF and eventually invade the parenchyma [48]. These alternative entry routes into the CNS, which were not assessed in this study, might be the reason why a complete prevention of autoimmune CNS inflammation upon MCAM blockade in the absence of VLA-4 was not be observed.

Conclusions
Our results indicate that targeting MCAM results in an overall lower, but more specific clinical improvement compared to targeting VLA-4 and seems to affect mainly T H 17 cells. This suggests that therapeutic blocking of MCAM might be advantageous over VLA-4 blocking, in terms of the balance of blocking pathogenic cells versus immune surveillance mechanisms and may, therefore, have fewer adverse effects. Hence, in case of a clear T H 17 pathology, blocking of MCAM, instead or in addition, to VLA-4 or basic therapies such as anti-interferon treatment might represent a new therapeutic avenue.

Additional files
Additional file 1: Figure S1. MCAM expression on NK cells does not correlate with clinical score. Percentages of MCAM-expressing NK1.1 + cells (a) isolated from the spleen, blood, spinal cord, and brain of isotype control or anti-MCAM-treated mice were quantified by flow cytometry on day 22 post MOG   Immunofluorescence staining of choroid plexus explants on day 2 (mouse 1; a) and day 5 (mouse 2, 3; b, c) after adoptive transfer for CMFDA + CD4 + T cells (green), and nuclei (blue; DAPI). Scale bars: 10 μm. (TIF 913 kb) Additional file 3: Figure S3. Characterization of human brain-derived microvascular endothelial cells and fibroblasts derived from human choroid plexus cells. mRNA levels of laminin α4, cytokeratin 18 (CK18), PECAM1, VE-cadherin, and vimentin in primary human brain-derived microvascular endothelial cells (HBMEC) and fibroblasts originated from primary human choroid plexus epithelial cells (labeled as HCPEpiC) were quantified by real-time PCR, revealing the lack of epithelial markers by HCPEpiC and confirming their fibroblastic nature. n.d. = not detected (TIF 4578 kb) Additional file 4: Figure S4.