OUP user menu

Extracellular matrix metalloproteinase inducer shows active perivascular cuffs in multiple sclerosis

Smriti M. Agrawal , Jacqueline Williamson , Ritu Sharma , Hania Kebir , Kamala Patel , Alexandre Prat , V. Wee Yong
DOI: http://dx.doi.org/10.1093/brain/awt093 1760-1777 First published online: 17 May 2013

Summary

Inflammatory perivascular cuffs are comprised of leucocytes that accumulate in the perivascular space around post-capillary venules before their infiltration into the parenchyma of the central nervous system. Inflammatory perivascular cuffs are commonly found in the central nervous system of patients with multiple sclerosis and in the animal model experimental autoimmune encephalomyelitis. Leucocytes that accumulate in the perivascular space secrete matrix metalloproteinases that aid their transmigration into the neural parenchyma. We described previously that the upstream inducer of matrix metalloproteinase expression, extracellular matrix metalloproteinase inducer (CD147), was elevated in experimental autoimmune encephalomyelitis, and that its inhibition reduced leucocyte entry into the central nervous system. Here we investigated whether the expression of extracellular matrix metalloproteinase inducer varies with the temporal evolution of lesions in murine experimental autoimmune encephalomyelitis, whether it was uniformly upregulated across multiple sclerosis specimens, and whether it was a feature of inflammatory perivascular cuffs in multiple sclerosis lesions. In experimental autoimmune encephalomyelitis, elevation of extracellular matrix metalloproteinase inducer was correlated with the appearance and persistence of clinical signs of disease. In both murine and human samples, extracellular matrix metalloproteinase inducer was detected on endothelium in healthy and disease states but was dramatically increased in and around inflammatory perivascular cuffs on leucocytes, associated with matrix metalloproteinase expression, and on resident cells including microglia. Leucocyte populations that express extracellular matrix metalloproteinase inducer in multiple sclerosis lesions included CD4+ and CD8+ T lymphocytes, B lymphocytes and monocyte/macrophages. The extra-endothelial expression of extracellular matrix metalloproteinase inducer was a marker of the activity of lesions in multiple sclerosis, being present on leucocyte-containing perivascular cuffs but not in inactive lesions. By using a function-blocking antibody, we implicate extracellular matrix metalloproteinase inducer in the adhesion of leucocytes to endothelial cells and determined that its activity was more crucial on leucocytes than on endothelium in leucocyte-endothelial cell engagement in vitro. Extracellular matrix metalloproteinase inducer activity regulated the level of alpha 4 integrin on leucocytes through a mechanism associated with nuclear factor κB signalling. Blocking extracellular matrix metalloproteinase inducer attenuated the transmigration of monocytes and B lymphocytes across a model of the blood–brain barrier in culture. In summary, we describe the prominence of extracellular matrix metalloproteinase inducer in central nervous system inflammatory perivascular cuffs, emphasize its dual role in matrix metalloproteinase induction and leucocyte adhesion, and highlight the elevation of extracellular matrix metalloproteinase inducer as an orchestrator of the infiltration of leucocytes into the central nervous system parenchyma.

  • multiple sclerosis
  • perivascular cuff
  • EMMPRIN
  • metalloproteinases
  • neuroinflammation

Introduction

Multiple sclerosis is a chronic disease of the CNS in which post-capillary venules become inflamed and leucocytes cross the blood–brain barrier to enter the CNS parenchyma. Within the CNS, leucocytes contribute to myelin degradation, gliosis and axonal loss resulting in lesions or plaques. Multiple sclerosis plaques vary in location, number, size and shape across patients, and are found throughout the CNS with a predominance in optic nerves, subpial spinal cord, brainstem, cerebellum and periventricular white matter regions (Sobel, 1995; van der Valk and De Groot, 2000; Frohman et al., 2006). Based on the extent of inflammation and demyelination, lesions in multiple sclerosis are classified using several different criteria (Trapp et al., 1998; Lucchinetti et al., 2000; Howell et al., 2011). Although originally thought to be a white matter disease, several neuropathological and imaging studies have reported multiple sclerosis lesion pathology in the grey matter of the CNS (Peterson et al., 2001; Kutzelnigg et al., 2005; Calabrese et al., 2010; Lucchinetti et al., 2011). The acute inflammation and demyelination in multiple sclerosis can be modelled in animals afflicted with experimental autoimmune encephalomyelitis (EAE) (Raine et al., 1980; Wekerle et al., 1994; Nelson et al., 2004; Ransohoff, 2006; Constantinescu et al., 2011).

Several sites of leucocyte entry into the CNS are reported in multiple sclerosis and EAE. Leucocytes may enter the CNS through meningeal infiltration and this may be particularly important for cortical pathology in multiple sclerosis (Howell et al., 2011). Another route is through alpha 4 integrin-mediated leucocytes adhering to endothelial cells of post-capillary venules and transmigrating across the endothelial cell layer to enter the CNS (Engelhardt et al., 1994; Engelhardt, 2008). CNS samples from both EAE and multiple sclerosis have revealed the presence of two unique laminin-containing basement membranes surrounding the endothelial cells of venules within the CNS (Sixt et al., 2001; van Horssen et al., 2005); the membrane proximal to endothelial cells is known as the endothelial basement membrane and the membrane abutting the CNS parenchyma is the parenchymal basement membrane (Sixt et al., 2001; Agrawal et al., 2006; Toft-Hansen et al., 2006). The space in between these two basement membranes is known as the perivascular space (Sixt et al., 2001; Archambault et al., 2005; Agrawal et al., 2006; Toft-Hansen et al., 2006). Although the perivascular space is normally difficult to visualize, the accumulation of leucocytes in the perivascular space separates the two basement membranes and the resulting cluster may be referred to as an inflammatory perivascular cuff. The inflammatory perivascular cuffs in multiple sclerosis can be detected in post-mortem multiple sclerosis samples using haematoxylin and eosin histochemistry (Raine et al., 1980; Moore et al., 1985; Cuzner et al., 1988), or double immunofluorescence microscopy for CD45+ leucocytes and laminin-positive basement membranes (Sixt et al., 2001).

Matrix metalloproteinases are a zinc-containing family of proteases that are upregulated in the CNS and inflammatory cells of both multiple sclerosis and EAE (Pagenstecher et al., 1998; Cuzner and Opdenakker, 1999; Bar-Or et al., 2003; Toft-Hansen et al., 2004; Weaver et al., 2005; Agrawal et al., 2006). Previous studies have shown a crucial role for matrix metalloproteinases, specifically MMP2 and MMP9, in leucocyte transmigration from the perivascular space into the CNS parenchyma (Agrawal et al., 2006; Toft-Hansen et al., 2006), wherein mice genetically deficient for both MMP2 and MMP9 were resistant to EAE disease (Agrawal et al., 2006). However, since many matrix metalloproteinases are simultaneously upregulated in multiple sclerosis and EAE (Toft-Hansen et al., 2004; Weaver et al., 2005) and have compensatory functions for each other, we examined the extracellular matrix metalloproteinase inducer (EMMPRIN), an upstream ‘on switch’ for several matrix metalloproteinases in multiple sclerosis and EAE (Guo et al., 1997; Li et al., 2001; Nabeshima et al., 2004; Yoon et al., 2005). EMMPRIN is a transmembrane glycoprotein that is normally found in the adult CNS on endothelial cells where its function remains unclear. Although well described as an autocrine and paracrine regulator of matrix metalloproteinase expression to facilitate the metastasis of tumour cells (Tang et al., 2004; Nabeshima et al., 2006), EMMPRIN also interacts with molecules that impact on immune functions such as integrins, cyclophilins and osteopontin (Curtin et al., 2005; Man et al., 2007; Yurchenko et al., 2010). Indeed, lymphocytes from EMMPRIN null mice have defects in proliferative responses to several mitogens (Igakura et al., 1998), and function blocking antibodies to EMMPRIN reduce T cell proliferation (Renno et al., 2002; Staffler et al., 2003; Chen et al., 2008). Moreover, monocytes with reduced EMMPRIN level or activity are less able to adhere to endothelium (Schulz et al., 2011). Given these reports, we examined and found increased EMMPRIN expression in and around inflammatory perivascular cuffs in peak EAE CNS samples and in one case with multiple sclerosis, and we documented reduced EAE disease outcomes in animals treated with a function blocking anti-EMMPRIN antibody (Agrawal et al., 2011). Still unresolved is whether EMMPRIN is uniformly upregulated across multiple sclerosis specimens, whether EMMPRIN is a feature of perivascular inflammation and on particular lymphocyte subsets, whether the expression of EMMPRIN varies with the temporal evolution of lesions, and whether EMMPRIN affects leucocyte adhesion apart from MMP activity, and through what mechanisms.

In this study, we investigated EAE samples for the temporal expression of EMMPRIN in perivascular cuffs in the CNS in correlation with clinical disease activity, and characterized several multiple sclerosis samples for the spatial expression of EMMPRIN in relation to the nature of the plaques. Our results demonstrate that EMMPRIN activity is significantly upregulated in active EAE and multiple sclerosis lesions defined by inflammatory perivascular cuffs where matrix metalloproteinase activity is induced, but that EMMPRIN has additional roles in leucocyte adhesion by affecting alpha 4 integrin expression, which is associated with NFκB activity. Our results suggest that anti-EMMPRIN therapy may prevent the excessive entry of leucocytes into the CNS in conditions such as multiple sclerosis.

Materials and methods

Animals and experimental autoimmune encephalomyelitis induction

Eight to 10 week old female C57BL/6 mice were used for EAE immunization. All procedures are in accordance with guidelines of the Canadian Council of Animal Care and have received approval by local ethics committee. For immunization, 50 µg of (myelin oligodendrocyte glycoprotein) MOG35-55 peptide in complete Freund’s adjuvant containing 10 mg/ml of heat inactivated Mycobacterium tuberculosis H37RA (Difco) was injected subcutaneously, 50 µl on either side of the tail base (Agrawal et al., 2011). Animals were supplemented with 300 ng pertussis toxin injected intraperitoneally on Days 0 and 2 post-MOG immunization. The animals were monitored daily for weight loss and changes of EAE disease score using a scale of 1–15 described previously (Weaver et al., 2005).

Human central nervous system specimens

This study was performed on post-mortem brain tissue from four cases of multiple sclerosis obtained from the UK Multiple Sclerosis Tissue Bank at Imperial College, London; and from six cases of multiple sclerosis obtained from the Veteran’s Affairs, UCLA, USA (Table 1). Control tissue from these banks consisted of patients who died without neurological disorders. All multiple sclerosis tissues were obtained and used with ethics approval. One frozen unfixed tissue block (∼4 cm2 surface area) for each case was cut into three or four blocks, and each block was sectioned into 10 -μm thin sections using a cryostat before staining.

View this table:
Table 1

Multiple sclerosis and control cases used in study

Patient ID (coded)Age (years)GenderMultiple sclerosis duration (years)DiagnosisBrain weight (g)Post-mortem time (h)
MS 139F21MS type not specified101018
MS 282F45SPMS109315
MS 359F42RRMS8858
MS 456M32RRMS11408
C192M-No MS, cardiac failure132713
C284F-No MS, cardiac failure132024
C369F-No MS, lung cancer113033
C435M-No MS, carcinoma of the tongue167022
C578F-No MS, myeloid leukaemia125033
HSB 151F10SPMS112019.5
HSB 249F15SPMS??
HSB 370M16SPMS??
HSB 457F3SPMS125019.8
HSB 550M5PPMS128013.8
HSB 675M2SPMS129017.25
HSB 770M-No MS, renal failure115012
HSB 881F-No MS, COPD110014.5
HSB 980M-No MS, cardiac arrest125010.5
HSB 1073F-No MS, cardiac arrest106015
HSB 1177M-No MS, cardiac arrest119512.4
  • Samples coded as multiple sclerosis (MS) and control (C) cases were from the UK MS Tissue Bank, while samples with the HSB designation are from the Human Brain and Spinal Fluid Resource Centre, Los Angeles.

  • COPD = chronic obstructive pulmonary disease; RRMS = relapsing–remitting multiple sclerosis; PPMS = primary progressive multiple sclerosis; SPMS = secondary progressive multiple sclerosis.

Immunofluorescence and histochemical staining

For immunofluorescence staining, frozen unfixed CNS tissues (cerebellum, forebrain and spinal cord) from control or EAE mice at various time points post-immunization, or from the multiple sclerosis tissue banks, were used. Each tissue section was mounted on a glass slide, fixed in −20°C methanol before blocking with 1% bovine serum albumin in PBS for 30 min. Primary antibodies (Table 2) were applied to tissue sections followed by their respective species-specific secondary antibodies conjugated to Alexa Fluor® 488 or Alexa Fluor® 546 (Jackson Laboratories). Sections were examined using an Olympus BX51 fluorescence microscope or Olympus Fluoview FV10i and photographed.

View this table:
Table 2

Antibodies used in this study and the applications

Company (catalogue number)Clone, species, targetApplication and concentration used
Serotec (MCA2283)OX-114, rat anti-mouse EMMPRINImmunofluorescence (1:500)
Zymed (34-5600)Rabbit anti-human EMMPRINImmunofluorescence(1:20–50)
BD Pharmingen (550539)30-F11, Rat anti-mouse CD45Immunofluorescence (1:100) and flow cytometry (1:50)
BD Pharmingen (557694)UCHT1, anti-human CD3 AF488Flow cytometry (1:50); Immunofluorescence (1:50)
BD Pharmingen (560792)9F10, anti-human integrin alpha 4 PEFlow cytometry (1:10)
BD Pharmingen (555443)MAR4, anti-human integrin beta 1 PEFlow cytometry (1:10)
BD Pharmingen (555924)6.7, anti-human integrin beta 2 PEFlow cytometry (1:10)
BD Pharmingen (556059)V1/824, mouse anti-human CD68Immunofluorescence (1:100)
BD Pharmingen (550389)WM59, mouse anti-human CD31 (PECAM1)Immunofluorescence (1:100)
Chemicon (AB805)Rabbit anti- human MMP9Immunofluorescence (1:100)
Sigma (G4546)Rabbit anti-human GFAPImmunofluorescence (1:500)
BD Pharmingen (553940)R4-22, rat anti-mouse IgM, κ isotype controlFunction blocking (1:50)
Kind Gift from Dr. Lydia Sorokin, Muenster, GermanyRabbit anti-mouse pan lamininImmunofluorescence (1:1000)
Cell Signaling Technology (9242)Rabbit anti-IκbαWestern blot (1:1000)
Cell Signaling Technology (4060)Rabbit anti-phospho-AKTWestern blot (1:1000)
Cell Signaling Technology (4370)Rabbit anti- p42/44 MAPKWestern blot (1:1000)
Abcam (ab20272)Anti-actin HRPWestern blot (1:10 000)
Serotec (MCA1267)RPA-T4; anti-human CD4Immunofluorescence (1:50)
Serotec (MCA1226)RPA-T8; anti-human CD8Immunofluorescence (1:100)
Serotec (MCA1710)2H7; anti-human CD20Immunofluorescence (1:100)
BD Pharmingen (556634)9F10; anti-human integrin alpha 4Immunofluorescence (1:100)

For histochemical staining, tissue sections were mounted on glass slides and dehydrated by increasing concentrations of ethanol for 1 min each, before being incubated in Luxol Fast blue solution at 60°C for 1 h. Slides were then rehydrated in 95% and 70% ethanol before being dipped in 0.05% lithium carbonate for 10–15 s and washed in distilled water. All sections were counterstained with haematoxylin and eosin, and coverslips were applied using mounting media (Acrytol). Sections were viewed under an Olympus bright field microscope (BH2) equipped with a digital camera (Olympus Q Color 3). Images were acquired using Imagepro software (MediaCybernetics).

In situ zymography

Gelatinolyic activity was localized in CNS sections by in situ zymography as previously described (Oh et al., 1999; Agrawal et al., 2006). Briefly, 10 μg/ml DQ-gelatin-FITC (EnzCheck; Invitrogen) in 50 mM Tris-HCl, pH 7.4 and 1 mM CaCl2 was applied to unfixed, frozen cerebellar cryosections. These were incubated for 4 h in a humid chamber at 37°C. Digestion of the gelatin-FITC substrate results in unquenching of fluorescein and the localization of sites of proteolytic activity in the cerebellar samples. Each section was then washed in PBS and fixed in −20°C methanol before processing for immunofluorescence staining, as described above.

Isolation and treatment of human T Cells

Human peripheral blood mononuclear cells were isolated from the blood of healthy adult volunteers by Ficoll-Hypaque centrifugation (GE Healthcare Life Sciences) as previously described (Chabot et al., 1997; Giuliani et al., 2003). The peripheral blood mononuclear cells were washed once with PBS and suspended in serum-free AIM-V medium (Invitrogen Life Technologies). To activate T cells in the peripheral blood mononuclear cell populations, 96 well round-bottomed plates were coated with 100 ng/ml purified mouse anti-human CD3 (BD Pharmingen) for a period of 3 h. Then 10 ng/ml anti-CD28 (BD Pharmingen) was added as a suspension to anti-CD3 wells, and human peripheral blood mononuclear cells were plated at a density of 1 000 000 cells/ml. Cells were left for 3 days at 37°C in a 5% humidified CO2 incubator and referred to as activated T cells thereafter as the CD3+ population constituted >90% of cells. In some experiments, IgM isotype control (10 μg/ml; BD Bioscience) or clone 10, an in-house generated anti-EMMPRIN antibody (10 μg/ml) (Agrawal et al., 2012) was added. Certain peripheral blood mononuclear cell preparations did not receive anti-CD3 or anti-CD28, and the floating cells were collected 3 days thereafter and referred to as non-activated T cells.

Flow cytometry

Flow cytometry was performed using fluorescence-conjugated antibodies against CD45-PercP (leucocytes; BD Bioscience), CD3-PE (T cells; BD Bioscience), CD49D-PE (alpha 4 chain integrin receptor; BD Bioscience), CD29-PE (beta 1 chain integrin receptor; BD Bioscience), CD18-PE (beta 2 chain integrin receptor; BD Bioscience) (Table 2). Briefly, cells were suspended in fluorescence-activated cell sorting (FACS) buffer (PBS + 2% foetal bovine serum), and blocked with an Fc blocker CD16/CD32 (1:100; BD Bioscience) for 20 min at 4°C. Cells were washed in FACS buffer and incubated in diluted antibodies of choice for 30 min at 4°C in the dark. Cells were washed in FACS buffer and fixed in 1% formalin before being suspended in FACS buffer and analysed using an LSRII FACS sorter.

Isolation of human umbilical vein endothelial cells

Human umbilical vein endothelial cells (HUVEC) were isolated from freshly isolated umbilical cords post-parturition by treatment with collagenase (280 U/ml) for 10 min and collected in 50 ml falcon tube. After being centrifuged at 1250 rpm for 5 min, the cell pellet was dissolved in 10 ml endothelial cell medium containing M199 (Gibco®) with NaHCO3 (2.2 mg/ml), HEPES (5.9 mg/ml) and 20% human serum and 1× PSG (penicillin–streptomycin–glutamine) before being cultured in T75 flasks. On the following day the cells were washed and fed with fresh endothelial cell medium. After 2 days, cells were plated in 24-well culture plates coated with 0.2% gelatin until they formed a confluent monolayer. One group of HUVEC cells was pre-activated with TNFα (10 ng/ml) for 4 h before use in static adhesion experiments described below (Ganguly et al., 2012).

Static adhesion with T cells and human umbilical vein endothelial cells

In one experiment, activated or non-activated T cells were treated with 10 μg/ml clone 10 (anti-EMMPRIN antibody) (Agrawal et al., 2012), or 10 μg/ml IgM isotype control before being suspended (1 × 10−6/ml) in serum-free AIM V® medium (Gibco) and 0.5 × 106 total cells were applied to the top of activated or non-activated HUVEC cells and incubated at 37°C for 10–30 min to allow adhesion to HUVECs. In another experiment, activated or non-activated HUVECs were treated with 10 μg/ml clone 10 (Agrawal et al., 2012) or IgM isotype control for 14–16 h before activation for 4 h with TNFα (10 ng/ml). Then 0.5 × 106 total non-activated or activated T cells were added to these HUVECs and incubated at 37°C for 30 min. Unattached T cells were washed off with AIM V® and fresh AIM V® was applied to each well. Six different fields of view per well were imaged using a Leica phase contrast microscope and recorded with a CCD camera (KP-MIV: Hitachi Denshi, Ltd). The cells were finally stained with Giemsa and washed before being imaged with a Lieca DMI 3000B microscope, with a Leica DFC 320 camera using Leica application suite software. Peripheral blood mononuclear cell adhesion to HUVECs was quantified from the Giemsa stained images with the ImageJ software, where HUVECs only were used to set the lower threshold, and activated HUVECs + activated T cells were used to set the upper threshold in each experiment. Experiments were repeated three times with different donors for HUVECs and T cells each time.

Human blood–brain barrier endothelial cell isolation and culture

Blood–brain barrier endothelial cells were isolated from CNS tissue specimens of temporal lobe resections from young adults undergoing surgery for the treatment of intractable epilepsy, as previously described (Biernacki et al., 2001). Informed consent was obtained before surgery and local institutional ethics committee has approved the use of cells. Cultures expressed the endothelial markers factor VIII, Ulex Agglutenens Europaensis-1 and antigen HT-7 until passage 7–8. Cells in the experiments were used before passage 4. No immune reactivity with β-tubulin, α-myosin or GFAP was detected, confirming the absence of contaminating neurons, smooth muscle cells or astrocytes, respectively.

Human leucocyte transmigration assay

Primary cultures of blood–brain barrier endothelial cells were used to generate an in vitro model of the human blood–brain barrier, as previously described (Biernacki et al., 2001; Kebir et al., 2009). In brief, blood–brain barrier endothelial cells were seeded on gelatin-coated 3 -μm pore size Boyden chambers (BD Biosciences) at a density of 3 × 104 cells per well in endothelial cell culture media supplemented with 40% (v/v) astrocyte-conditioned media, which has been shown to induce and maintain blood–brain barrier characteristics in vitro. Brain endothelial cells were allowed to grow for 3 days to form a confluent monolayer. A suspension of 1 × 106 per ml human peripheral blood CD14+ monocytes or CD19+ B lymphocytes obtained from healthy volunteers and purified by magnetic bead sorting (Miltenyi Biotec) was pretreated for 24 h with 10 μg/ml clone 10 anti-EMMPRIN antibody (Agrawal et al., 2012), or 10 μg/ml IgM isotype control and loaded in the upper chamber. After 18 h, the absolute number of cells that had transmigrated to the lower chamber was counted.

Western blots

Human peripheral blood mononuclear cells were activated with anti-CD3/CD28 as described above, in the absence or presence of anti-EMMPRIN antibody treatment. Two hours later, cells were pelleted and sonicated in protein lysis buffer containing 1% Triton™ X-100, phosphatase inhibitor cocktail tablet (PhosSTOP; Roche Diagnostics) and protease inhibitor tablet (Roche Diagnostics). Lysates were separated by SDS-PAGE on 10% gels, transferred to polyvinylidene fluoride membranes, and probed using primary antibodies listed in Table 2. Secondary anti-mouse antibodies conjugated with horseradish peroxidase were added and detected using an ECL chemiluminesence kit (GE Lifesciences). Membranes were imaged and analysed using a gel documentation system (Syngene).

Statistical analysis

Treatment effects of leucocytes and endothelial cell co-cultures were analysed with one-way ANOVA followed by Tukey’s test for multiple comparisons. For comparisons between two groups, Student’s t-test was applied. P < 0.05 was considered statistically significant.

Results

In experimental autoimmune encephalomyelitis, inflammatory perivascular cuffs evident after onset of clinical signs show EMMPRIN expression

Inflammatory perivascular cuffs, defined herein as the accumulation of leucocytes in the perivascular space between the endothelial and the parenchymal basement membranes, can be detected in the CNS by staining for CD45+ leucocytes, and by pan-laminin to demarcate both basement membranes (Fig. 1A). Inflammatory perivascular cuffs were not detected in the CNS of control naïve animals or presymptomatic mice (before Day 10) that have been immunized with MOG (Fig. 1B). However, at the onset of clinical signs of EAE (∼Day 10 post-MOG immunization), cuffs became visible in the white matter of the cerebellum. With evolution of disease to peak clinical severity (∼Day 15), inflammatory perivascular cuffs became more numerous and the infiltration of leucocytes into the parenchyma of cerebellar white matter became readily apparent (Fig. 1B). At post-peak EAE (Day 20) (Fig. 1B), large perivascular cuffs were less prominent while diffuse parenchymal inflammation was extensive. Inflammatory perivascular cuffs were also evident in other CNS white matter areas sampled at peak clinical severity (Fig. 1C).

Figure 1

Inflammatory perivascular cuffs are evident in various CNS regions of mice only after onset of clinical signs of EAE. (A) A schematic representation and the corresponding CNS tissue in cross-section from an EAE mouse stained with antibodies to pan-laminin and CD45 to depict an inflammatory perivascular cuff. The scheme displays leucocytes transmigrating from the lumen through the endothelial basement membrane, accumulating in the perivascular space, and then transmigrating through the parenchymal basement membrane into the CNS parenchyma. (B) Using haematoxylin and eosin and Luxol Fast blue staining (H&E/LFB) or immunofluorescence for pan-laminin and CD45, the presence of inflammatory perivascular cuffs (arrowheads in bottom panels) or parenchymal inflammation (arrows) in both control and EAE mice at various days post-MOG immunization is evaluated. Day 0–10: pre-onset of clinical signs; Day 10: onset of clinical signs; ∼Day 15: peak severity of clinical signs; and >Day 15: post-peak. (C) Haematoxylin and eosin and Luxol Fast blue or immunofluorescence for pan-laminin and CD45 shows a widespread spatial distribution of inflammatory perivascular cuffs in Day 15 peak EAE mouse CNS, at low and high magnification. Arrowheads show some inflammatory perivascular cuffs. Pictures are representative of stainings from more than five mice per group, and magnification bars are indicated as 50 or 100 μm within immunofluorescence micrographs.

We examined EMMPRIN expression in CNS tissue from control and MOG immunized mice. EMMPRIN was associated with pan-laminin immunoreactivity in control and presymptomatic CNS but not in the CNS parenchyma (Fig. 2A). On onset of clinical signs, EMMPRIN expression markedly upregulated within and around inflammatory perivascular cuffs (Fig. 2A).

Figure 2

EMMPRIN expression is upregulated and coincident with inflammatory perivascular cuffs from onset of clinical signs of EAE. (A) Cerebellar sections from control or EAE mice at various stages of disease activity. The panels (bottom row is higher magnification than the upper row) are of merged immunofluorescence for EMMPRIN and pan-laminin, while the inserts depict single stains. (B) The percentage (mean ± SD) of inflammatory perivascular cuffs positive for EMMPRIN is quantified in control and in EAE CNS sections in more than five mice at each time point.

We counted the number of inflammatory perivascular cuffs in the cerebellum of mice to address their frequency of EMMPRIN expression. No inflammatory perivascular cuffs were found in control or pre-symptomatic mice. Inflammatory perivascular cuffs were detected in mice from onset of clinical signs, and all of these were immunoreactive for EMMPRIN (Fig. 2B). Similarly, in mice at peak or post-peak clinical severity, the elevated number of inflammatory perivascular cuffs from clinical onset all had EMMPRIN positivity (Fig. 2B). Thus, EMMPRIN induction is a signature for active inflammatory lesions in EAE.

EMMPRIN upregulation is a hallmark of inflammatory perivascular cuffs in multiple sclerosis

We first examined CNS specimens from non-multiple sclerosis control cases and did not find evidence of inflammatory perivascular cuffs using haematoxylin and eosin and Luxol Fast blue staining or CD45/laminin immunoreactivity (Fig. 3A and B). In these normal samples, an antibody to human EMMPRIN (from Zymed; Table 2) revealed striated stainings on endothelial-like structures (Fig. 3C) more distinctly than was possible with the Serotec antibody to mouse EMMPRIN in normal mouse brain (Fig. 2A). We confirmed that the striated EMMPRIN stainings were intercellular on endothelium, in location of tight junctions between endothelial cells, by using an antibody to PECAM1 to label endothelial cells (Fig. 3D).

Figure 3

EMMPRIN upregulation in multiple sclerosis is observed only in inflammatory perivascular cuffs. CNS sections were subjected to haematoxylin and eosin and Luxol Fast blue (A) or immunofluorescence for pan-laminin/CD45 (B) to show the absence [in control or in multiple sclerosis (MS) tissue with no evidence of inflammatory perivascular cuffs] or presence (right column) of inflammation and perivascular cuffs. (C) EMMPRIN is normally found on endothelial structures but is significantly elevated in areas with inflammation (CD45+) in and around perivascular cuffs (yellowish stainings from the red and green overlap). Data are representative of more than three CNS samples from individual patients, and magnification bars are indicated within micrographs. (D) Staining for PECAM1/EMMPRIN reveals that EMMPRIN expression is intercellular between PECAM1-positive endothelial cells.

Next, we examined several CNS specimens from a number of subjects with multiple sclerosis (Table 1) and found inflammatory perivascular cuffs in the white matter associated with demyelination (loss of Luxol Fast blue staining) in some multiple sclerosis cases (Fig. 3A and B). These inflammatory perivascular cuffs were associated with an upregulation of EMMPRIN (Fig. 3C) in areas with inflammation, and always in active cases (Table 3). Specifically, beyond the endothelial expression that was also found in control samples, EMMPRIN upregulation in multiple sclerosis white matter was observed on leucocytes in the perivascular space and on those that were in the CNS parenchyma. Some cells in the CNS parenchyma around perivascular cuffs that were presumably of neural origin were also immunoreactive for EMMPRIN (Fig. 3C).

View this table:
Table 3

EMMPRIN is uniformly upregulated in inflammatory perivascular cuffs in active multiple sclerosis lesions

Patient ID (coded)Pathological activityNumber of blocks sampledEMMPRIN expression in inflammatory perivascular cuffs
MS 1Active/chronic active812/12
MS 2Active/chronic active1518/18
MS 3Acute active518/18
MS 4Active83/3
C1-1-
C2-1-
C3-1-
C4-1-
C5-1-
HSB 1Active42/2
HSB 2Active32/2
HSB 3Inactive3-
HSB 4Inactive3-
HSB 5Inactive3-
HSB 6Inactive3-
HSB 7-1-
HSB 8-1-
HSB 9-1-
HSB 10-1-
HSB 11-1-
  • Pathological activity reported here is that documented in the case report for each patient from the respective tissue bank. Patient IDs are coded for confidentiality. In the right column, the denominator is the number of lesions with active inflammatory perivascular cuffs examined in the sampled blocks, while the numerator is the number of EMMPRIN-positive lesions.

In grey matter areas of multiple sclerosis specimens, we found rare inflammatory perivascular cuffs. Given their low frequency and thus requiring cautious interpretation, the few inflammatory perivascular cuffs in grey matter areas displayed EMMPRIN upregulation on leucocytes (Supplementary Fig. 1). We did not have meningeal lesions in our collection of cases with multiple sclerosis and it remains unresolved if EMMPRIN is expressed in these lesions.

Two cases with multiple sclerosis examined contained several large inflammatory perivascular cuffs in the white matter associated with areas of demyelination (Fig. 4A and B), permitting us to examine further the expression of EMMPRIN with lesion activity. EMMPRIN was present on leucocytes in the perivascular space (Fig. 4C and D; arrow), as well as on those that have entered the CNS parenchyma (Fig. 4C; arrow head). At high magnification, it is evident that EMMPRIN is co-localized to CD45+ leucocytes (Fig. 4D); whether EMMPRIN is expressed in the extracellular matrix surrounding leucocytes is indiscernible, given the tightly packed nature of the cells.

Figure 4

EMMPRIN is highly expressed in cases with multiple sclerosis with large inflammatory perivascular cuffs. (A) Haematoxylin and eosin with Luxol Fast blue (H&E/LFB), and (B) immunofluorescence for pan-laminin and CD45 identified two multiple sclerosis cases (single staining depicted in vertical columns) with several large inflammatory perivascular cuffs. (C) Immunofluorescence staining shows increase in EMMPRIN expression in areas with inflammation and in (arrow) and around (arrowhead) perivascular cuffs. (D) Higher magnification images and 3D z-stacks show co-localization of EMMPRIN with CD45. Magnification bars are indicated within micrographs.

We sought to determine the immune cell subset(s) expressing EMMPRIN in inflammatory perivascular cuffs. Figure 5 shows that several CD3+ T cells express EMMPRIN, and these were of both the CD4+ or CD8+ subsets (Fig. 6). Moreover, several CD68+ macrophages/microglia as well as CD20+ B lymphocytes (Fig. 6) were immunoreactive for EMMPRIN within the perivascular cuffs as well in the CNS parenchyma. We note that not all leucocytes were EMMPRIN-positive in inflammatory perivascular cuffs. Indeed, as depicted in Fig. 5F, some CD3+ T cells do not express EMMPRIN (arrow head), while some CD3− cells (likely macrophages/microglia as detected by CD68 in Fig. 6) within the CNS parenchyma were found to express EMMPRIN (Fig. 5F; short arrow). In inflammatory perivascular cuffs with significant EMMPRIN upregulation, fibrinogen leakage into the CNS parenchyma was evident, a manifestation of the breakdown of the blood–brain barrier (Supplementary Fig. 2).

Figure 5

EMMPRIN is upregulated on T cells in inflammatory perivascular cuffs. (A) Histochemical staining of multiple sclerosis CNS using haematoxylin and eosin coupled with Luxol Fast blue (H&E/LFB) shows inflamed vessels and demyelination. One particular inflamed vessel (white box) is shown at a higher magnification in (B), and this was characterized further for EMMPRIN and leucocyte subsets (Figs 5 and 6). (C) Immunofluorescence staining for CD45 (red), pan-laminin (green) and a nuclear dye (nuclear yellow, NY, blue) reveal an inflammatory perivascular cuff with CD45+ cells in the perivascular space and invading into the CNS parenchyma. The lumen of the vessel is indicated for orientation and the asterisk marks where the vessel has become separated from the CNS parenchyma (frozen tissues are susceptible to such types of tears). Immunofluorescence staining for CD3 (red) with nuclear yellow (NY) (D) shows that some T cells express EMMPRIN (green) (E). The merged image in F shows EMMPRIN expression on T cells within the perivascular cuff (long arrow); some CD3+ T cells do not express EMMPRIN (arrow head) while some cells within the CNS parenchyma that express EMMPRIN are not CD3+ (short arrow). Although the image in F implies that CD3+ T cells that have entered the CNS parenchyma are EMMPRIN-negative, this is not the case in other sections analysed (data not shown). Magnification bars are indicated within micrographs.

Figure 6

EMMPRIN is upregulated on T lymphocyte subsets, macrophages/microglia and B lymphocytes in and around inflammatory perivascular cuffs. Double immunofluorescence staining shows EMMPRIN expression on T (CD4 and CD8) cell subsets, macrophage/microglia (CD68), and B cells (CD20). Left: Cellular subset markers (CD4, CD8, CD68 or CD20) (red) co-labelled for Hoechst (blue, in nucleus) to allow the visualization of markers on the surface of cells, while the middle panels depict the corresponding EMMPRIN (green) immunoreactivity. Right: Merged images of EMMPRIN on markers of cellular subsets where the yellowish tinge results from the overlap of signals from the red and green channels. The insert in the merged image highlights a single cell. Note that the tissue sections were from frozen, unfixed blocks, and that following sectioning, the sections were then fixed before antibody was applied; while this protocol preserved many markers for immunohistochemistry, there is some loss of cellular integrity.

EMMPRIN upregulation in situ is associated with increased matrix metalloproteinase activity

Since EMMPRIN induces the cellular expression of several matrix metalloproteinases (reviewed in Li et al., 2001; Nabeshima et al., 2006) and as MMP9 is implicated in multiple sclerosis (reviewed in Chandler et al., 1997; Boz et al., 2006; Yong et al., 2007), we investigated MMP9 protein expression and enzymatic activity at sites of EMMPRIN upregulation. We first used mouse cerebellar sections and found that MMP9 and EMMPRIN proteins were coincident in inflammatory perivascular cuffs (Fig. 7A). Moreover, when MMP9 activity was examined using in situ zymography, we found that this was absent in control or presymptomatic CNS but prominently expressed in inflammatory perivascular cuffs found in cerebellar white matter of mice at onset and at peak EAE clinical severity (Fig. 7B). Zymographic activity was located within the perivascular space and immediately outside of the parenchymal basement membrane.

Figure 7

Expression of EMMPRIN and matrix metalloproteinase is coincident in inflammatory perivascular cuffs in EAE and multiple sclerosis. (A) Immunofluorescence staining for EMMPRIN/MMP9 and (B) pan-laminin coupled with in situ zymography to detect matrix metalloproteinase activity in control or EAE CNS tissue. The results show EMMPRIN to be co-localized with MMP9 staining and activity. (C) Immunofluorescence staining for CD45 and MMP9 in human multiple sclerosis (MS) CNS sections. Data are representative of CNS samples from more than three mice or individual patients, and magnification bars are indicated within micrographs.

For multiple sclerosis specimens, we examined several available commercial anti-EMMPRIN antibodies in immunofluorescence staining, and used one antibody (rabbit anti-human EMMPRIN; Invitrogen, previously sold by Zymed) (Table 2) that bound EMMPRIN most efficiently. As this antibody was raised in rabbit, we were unable to use it simultaneously with the available anti-MMP9 (rabbit anti-human MMP9; Chemicon; Table 2) or anti-pan-laminin (rabbit anti-pan-laminin; Table 2) antibodies so this precluded the simultaneous immunofluorescence detection of human EMMPRIN with MMP9 or laminin. In addition, in situ zymography that works best on fresh frozen EAE tissue yielded very high background staining in the human samples, preventing the use of this technique to accurately determine matrix metalloproteinase activity in human multiple sclerosis sections. Therefore, EMMPRIN and MMP9 antibodies were applied to adjacent sections with CD45 immunoreactivity as a commonality between the sections. Significant immunoreactivity for MMP9 protein could be found in CD45+ areas, which were also immunoreactive for EMMPRIN (Fig. 7C). Thus, as in EAE sections, EMMPRIN and one of its downstream proteases, MMP9, were co-expressed surrounding inflammatory perivascular cuffs in multiple sclerosis.

EMMPRIN function on leucocytes is crucial for their adhesion to and migration across endothelium

In previous work with EAE mice (Agrawal et al., 2011), we reported that a function-blocking antibody to EMMPRIN reduced downstream matrix metalloproteinase activity and prevented the infiltration of leucocytes into the CNS parenchyma. That study also showed that EMMPRIN blockade decreased the number of inflammatory perivascular cuffs, suggesting that EMMPRIN was also involved in the initial adhesion of leucocytes onto endothelial cells. The latter possibility is supported by the findings that EMMPRIN is expressed normally on endothelial cells, and significantly elevated in infiltrating leucocytes (Figs 2–4). To determine whether the EMMPRIN on infiltrating cells or the endothelium is crucial for leucocyte migration from blood vessels into the CNS parenchyma, we used an in vitro static adhesion assay whereby a function-blocking antibody to EMMPRIN (clone 10, generated in-house) (Agrawal et al., 2012) was applied to either T cells or HUVECs. Activated human T cells adhered to HUVECs more efficiently than non-activated T cells. When EMMPRIN function was blocked on T cells before their application to endothelial cells, fewer T cells adhered compared with those treated with isotype control (Fig. 8A and B). Conversely, when the EMMPRIN antibody was applied to HUVEC cells, but not to T cells, the adhesion of T cells was not altered (Fig. 8C). These results highlight an important role for EMMPRIN on T cells in their adhesion to endothelium before their migration into the perivascular cuff and eventually into the CNS parenchyma. On the other hand, the endothelial expression of EMMPRIN does not appear to regulate T cell adhesion.

Figure 8

EMMPRIN function on leucocytes is crucial for their adhesion to and migration across endothelial cells (ECs). (A) Giemsa staining and phase contrast microscopy were used to image the static adhesion of non-activated T cells (NON ACT), activated T cells (ACT), T cells activated in the presence of anti-EMMPRIN antibody (EC + ACT + ANTI-EMMPRIN) or isotype control (EC + ACT + isotype control) onto activated HUVEC cells for 30 min at 37°C. Giemsa staining of fixed cells or phase contrast microscopy of living cultures are displayed along with the original magnification. (B) Quantification by ImageJ of total adherent T cells to endothelial cells when T cells were pretreated with anti-EMMPRIN antibody or isotype control shows that the increased adhesion of activated T cells was reduced by anti-EMMPRIN (one-way ANOVA, Tukey’s multiple comparison post hoc test, ***P < 0.001). (C) When endothelial cells were pretreated with anti-EMMPRIN antibody before the addition of T cells, there was no difference in the capacity of activated T cells to adhere (NS = not significant). Human monocyte (D) or B cell (E) migration across blood–brain barrier (BBB) endothelial cells was significantly reduced by treatment with anti-EMMPRIN antibody compared to isotype control (***P < 0.001). Results of B and C are mean ± SEM of data from three experiments (three different HUVEC and peripheral blood mononuclear cell preparations) where in each experiment, replicate cultures were tested and where six fields of view were imaged per replicate. The monocyte migration data of (D) represent two independent experiments performed in triplicate, while the B cell migration data of E represent an independent experiment performed in triplicate.

To evaluate the function of EMMPRIN on leucocytes other than T cells, and to clearly examine the effect of EMMPRIN on the blood–brain barrier, we investigated the transmigration of human monocytes or B cells across a model of the blood–brain barrier constituted by human brain endothelial cells (Biernacki et al., 2001; Kebir et al., 2009). Relative to the IgM isotype antibody control, we determined that anti-EMMPRIN reduced the transmigration of human monocytes (Fig. 8D) and B cells (Fig. 8E) by >90% and 50%, respectively.

Finally, we sought to examine the mechanisms by which reducing EMMPRIN activity in T cells with clone 10 decreased their ability to adhere to endothelium. Specifically, we addressed levels of integrin receptors on T cells that facilitate their adhesion to endothelial cells. Using flow cytometry we found that T cells treated with anti-EMMPRIN versus isotype controls expressed less alpha 4 integrin subunit on their surface (Fig. 9A–F), providing a mechanism for their reduced adhesion to endothelial cells. Levels of beta 1 and beta 2 integrin subunits were unaffected by anti-EMMPRIN exposure. By immunohistochemistry, we found that in multiple sclerosis brain sections with inflammation, alpha 4 integrin was co-localized with EMMPRIN on infiltrating leucocytes (Fig. 9G).

Figure 9

Treatment with anti-EMMPRIN antibody reduces alpha 4 integrin on activated human T cells. Human T cells treated with anti-EMMPRIN antibody or isotype control were stained and gated in flow cytometry as (A) all leucocytes (CD45+), from which (B) all CD3+ T cells were gated; from these the cells expressing (C) either α4, β1 or β2 integrin were examined. Differences in staining intensities of various integrin subunits with or without treatment with anti-EMMPRIN or isotype are shown using (D) histogram overlays, (E) percentage of CD3+ cells and (F) mean fluorescent intensity (MFI). Data are representative of cells derived from three healthy donors. (G) Immunofluoresence staining for EMMPRIN and alpha 4 integrin in human multiple sclerosis CNS sections. (H) Western blot and densitomentry analysis for IκBα protein levels in peripheral blood mononuclear cells treated with anti-EMMPRIN (clone 10, 10µg/ml) versus IgM control (10 µg/ml). Data are mean ± SEM from three individual samples.

Next, we sought to examine the mechanisms that may account for the reduction of alpha 4 integrin by anti-EMMPRIN treatment. When peripheral blood mononuclear cells were harvested 2 h after anti-CD3/CD28 activation, treatment with anti-EMMPRIN inhibited the NFκB pathway evident by the retention of IκBα protein levels (Fig. 9H) in cells from three individual donors. At the 2 h time point, we did not observe changes to the phosphorylated (active) form of Akt (protein kinase B) and p42/44 mitogen activated protein (MAP) kinase (data not shown). These data suggest that the reduced alpha 4 integrin expression on T cells could be due to anti-EMMPRIN blockage of the NFκB translocation to the nucleus for gene transcription.

Discussion

Multiple sclerosis lesions have been classified into several patterns on the basis of myelin loss, plaque extent and topography, oligodendrocyte death, and the nature and persistence of an inflammatory response (Trapp et al., 1998; Lucchinetti et al., 2000; Howell et al., 2011). Despite the heterogeneity, a common factor in these lesions is the abundance of immune cells that infiltrate into the CNS (Trapp and Nave, 2008; Popescu and Lucchinetti, 2012). It is thought that immune cell infiltration into the CNS marks the onset of a multiple sclerosis relapse.

The process of leucocyte transmigration across a post-capillary venule into the CNS involves several important steps (Engelhardt and Ransohoff, 2005; Larochelle et al., 2011). The first step is a slowing of leucocytes within the blood through the interaction and binding of integrin alpha 4 beta 1 receptors present on leucocytes with several cell adhesion molecules on the endothelium including VCAM1, ICAM1 and Leukocyte function antigen-1 (LFA-1) (Yednock et al., 1992; Engelhardt et al., 1994; Engelhardt, 2008). This is followed by leucocyte arrest to endothelial cells possibly due to a response to chemokines secreted by endothelial cells (Alt et al., 2002). Once leucocytes adhere, they migrate across the endothelial cell barrier and endothelial basement membrane to accumulate in the perivascular space and form an inflammatory perivascular cuff.

The perivascular space have been defined as the space where leucocytes accumulate before entering the CNS parenchyma (Raine et al., 1980; Man et al., 2007), and the inflammatory perivascular cuff has been classically detected using histological stains to show increased accumulation of immune cells around a blood vessel lumen (Raine et al., 1980; Moore et al., 1985; Cuzner et al., 1988). More recently, the perivascular space has been better delineated as the space between the laminin-immunoreactive endothelial basement membrane and the parenchymal basement membrane (Sixt et al., 2001; van Horssen et al., 2005; Agrawal et al., 2006; Wu et al., 2009).

The family of matrix metalloproteinases helps leucocytes transit into the CNS. Matrix metalloproteinases do not seem to be required in leucocyte migration across the endothelial basement membrane (Wu et al., 2009), but they are secreted and needed by leucocytes when transmigrating the parenchymal border (Agrawal et al., 2006; Toft-Hansen et al., 2006). Specifically, MMP2 and MMP9 have been shown to cleave β-dystroglycan, a receptor on astrocyte end feet that abuts the parenchymal basement membrane, allowing cells to enter the CNS parenchyma (Agrawal et al., 2006). These findings have emphasized the dependence of leucocytes on matrix metalloproteinase activity to transit the perivascular space and infiltrate the CNS parenchyma. Aside from MMP2 and MMP9, several matrix metalloproteinases have been suggested to have important roles in multiple sclerosis and EAE, including MMP7, MMP8, MMP14 and others (Cossins et al., 1997; Adair-Kirk et al., 2003; Bar-Or et al., 2003; Toft-Hansen et al., 2004; Weaver et al., 2005; Folgueras et al., 2008; Buhler et al., 2009).

Despite the myriad of evidence supporting a role for matrix metalloproteinases in multiple sclerosis and EAE, treatment with matrix metalloproteinase inhibitors has been challenging, mostly due to the lack of specific inhibitors, and because of the multiple matrix metalloproteinase members that are upregulated simultaneously. In contrast, targeting a single molecule that regulates the matrix metalloproteinases would be advantageous. Our recent work has shown that EMMPRIN, an upstream regulator of several matrix metalloproteinases, is upregulated in EAE with disease progression and in one case with multiple sclerosis in areas of inflammation within the CNS. We proposed an important role for EMMPRIN as an EMMPRIN function-blocking antibody reduced matrix metalloproteinase activity around perivascular cuffs and reduced EAE disease symptoms (Agrawal et al., 2011). Critical information that is still lacking includes whether EMMPRIN is uniformly expressed across lesions in multiple sclerosis, and the temporal and spatial expression of EMMPRIN in relation to the inflammatory perivascular cuff.

In this study, upon examining EAE CNS samples spatially and temporally, we determined that inflammatory perivascular cuffs are absent in the CNS of control mice and only appear at disease onset, persisting with disease progression. In EAE, perivascular cuffs occur randomly within the white matter tracks in various regions of the CNS including spinal cord, cerebellum and in forebrain. We show that EMMPRIN expression is specifically upregulated in and around these inflammatory perivascular cuffs in EAE, correlating with matrix metalloproteinase activity. Extending to human samples, we find that EMMPRIN expression is at the intercellular border of PECAM-1 expressing endothelial cells in the normal CNS. In multiple sclerosis, however, EMMPRIN is significantly upregulated on many cell types, but only in association with inflammatory perivascular cuffs. Indeed, EMMPRIN expression is a hallmark of these active lesions, correlating with MMP9 expression at the site of entry of leucocytes into the CNS parenchyma. These findings strongly suggest EMMPRIN as a regulator of the trafficking of leucocytes into the CNS in active multiple sclerosis lesions.

In addition to a role in leucocyte migration across the parenchymal basement membrane border, our data also suggest a role for EMMPRIN in adhesion of leucocytes onto endothelial cells in the first place. In previous work, we found that EMMPRIN levels were elevated on blood leucocytes before the initiation of clinical signs of EAE, and that blocking EMMPRIN not only reduced the infiltration of leucocytes into the CNS parenchyma but lowered the number of inflammatory perivascular cuffs in the CNS in the first place (Agrawal et al., 2011). This, together with EMMPRIN expression on normal brain endothelium (Seulberger et al., 1990; Mutin et al., 1997; Fan et al., 1998), prompted an investigation of the potential role of EMMPRIN in cellular adhesion. We found that when EMMPRIN function was blocked on activated T cells, it reduced their capacity to adhere to or migrate across endothelial cells, which correlated with a reduction in alpha 4 integrin on T cells. We also detected an accumulation of IκB in these cells suggesting a reduction in signalling through the NFκB pathway. On the other hand, blocking EMMPRIN on endothelial cells did not alter the adhesion of T cells; the function of EMMPRIN on endothelial cells remains unknown.

Signalling pathways including the NFκB pathway are implicated in alpha 4 integrin mediated leucocyte recruitment by endothelial cells (Kaur et al., 2003). Typically, IκB that is bound to the p65/p60 units of NFκb is phosphorylated and then degraded upon cell activation resulting in the translocation of p60/p65 NFκb into the nucleus to induce gene transcription. Upon cellular inactivity, early accumulation of IκB is indicative of inhibition of the NFκB pathway (de Winther et al., 2005). Treatment of T cells with recombinant EMMPRIN has been shown to activate cells and the NFκB pathway, evident by reduced IκBα protein in the cell cytoplasm (Schmidt et al., 2008). Our results show that treatment with anti-EMMPRIN reduces alpha 4 integrin and leucocyte adhesion onto endothelial cells possibly due to inhibition of the NFκB pathway.

Although EMMPRIN is an upstream regulator of matrix metalloproteinase expression (Biswas et al., 1995; Guo et al., 2000; Li et al., 2001; Nabeshima et al., 2004), there are also functions of EMMPRIN that are matrix metalloproteinase-independent (Pushkarsky et al., 2005; Ge et al., 2007; Agrawal and Yong, 2011). For instance, EMMPRIN regulates T cell proliferation through a mechanism that is not blocked by potent matrix metalloproteinase inhibitors (Agrawal et al., 2011). Thus, the widespread upregulation of EMMPRIN in active multiple sclerosis lesions in this study could have multiple targets and separate roles. The elevated expression of EMMPRIN at the parenchymal border of inflammatory perivascular cuffs, and its close association with MMP9 activity (Fig. 7) is consistent with our previous report that the MMP9 cleavage of β-dystroglycan at the parenchymal basement membrane/astrocyte end feet (glial limitans) regulates leucocyte trafficking into the CNS parenchyma. Our data also show a role for EMMPRIN in the adhesion of leucocytes onto endothelial cells likely through an alpha 4 integrin-EMMPRIN mediated mechanism. Thus, in our study whereby anti-EMMPRIN reduced the transmigration of monocytes and B cells across a model of the blood–brain barrier (Fig. 8), it is likely that the effect of anti-EMMPRIN is contributed by both reduced adhesion and decreased proteolysis.

In summary, we have found several novel aspects of EMMPRIN in multiple sclerosis. EMMPRIN is not only involved in the matrix metalloproteinase-mediated cell migration across the parenchymal basement membrane into the CNS parenchyma, but it has a crucial role in leucocyte adhesion to endothelial cells, the first step in immune cell migration into the CNS. An additional novelty is that it is the EMMPRIN on leucocytes that influenced adhesion to endothelial cells, while blocking endothelial EMMPRIN expression did not significantly impair cellular adhesion. EMMPRIN upregulation is prominent in multiple sclerosis specimens only in areas of inflammatory perivascular cuffs; we determined that EMMPRIN is elevated on CD4 and CD8 T cell subsets, B lymphocytes and macrophages/microglia around these cuffs. We describe for the first time that B cell transmigration is also affected by EMMPRIN. Finally, we describe a new relationship between EMMPRIN and alpha 4 integrin, through a mechanism associated with NF-κB activation. In conclusion, our findings suggest that EMMPRIN not only acts at multiple levels in multiple sclerosis and EAE, but that EMMPRIN may be predictive of CNS inflammation and may serve as a biomarker for multiple sclerosis disease activity. These results reinforce the importance of blocking EMMPRIN function in EAE and multiple sclerosis.

Funding

This study was funded by an operating grant to Dr V.W. Yong from the Canadian Institutes of Health Research.

Supplementary material

Supplementary material is available at Brain online.

Acknowledgements

We thank Dr. Richard Reynolds and the United Kingdom MS Tissue Bank (www.ukmstissuebank.imperial.ac.uk), and Dr. Wallace Tourtellotte and the Human Brain and Spinal Fluid Resource Centre (VA West Los Angeles Healthcare Centre, USA) for providing the post-mortem control and multiple sclerosis samples analysed in the study. We thank Dr Lydia Sorokin for the provision of antibody to pan-laminin and for critically reviewing this manuscript. We are grateful to Dr Yan Fan for her technical support with immunohistochemical staining and Ms Claudia Silva for technical support in the generation of the monoclonal anti-EMMPRIN clone 10 antibody. We thank Unit 51 at the Foothills Hospital, Calgary, for providing umbilical cords. The authors do not have any conflicting financial interests.

Abbreviations
EAE
experimental autoimmune encephalomyelitis
EMMPRIN
extracellular matrix metalloproteinase inducer
HUVEC
human umbilical vein endothelial cells

References

View Abstract