Metalloproteinases Control Brain Inflammation Induced by Pertussis Toxin in Mice Overexpressing the Chemokine CCL2 in the Central Nervous System¹

Multiple sclerosis (MS)³ is a common neurological disease that is characterized by demyelination and inflammation of the CNS. The cause of MS is unknown, but may be the result of an environmental stimulus in genetically predisposed individuals (1). There is an established link between genotype and disease susceptibility to MS, and microbial infections can precipitate attacks in MS patients (2, 3, 4). Transgenic (Tg) mice in which all TCRs recognize a myelin basic protein (MBP) epitope only developed inflammatory demyelinating disease whether given pertussis toxin (PTx) or kept in a nonsterile environment (5). PTx is derived from Bordetella pertussis, the causative agent of whooping cough, and is routinely used as adjuvant to induce experimental autoimmunity, possibly substituting for critical environmental cues in the pathogenesis of autoimmune disease.

Chemokines are small chemoattractant cytokines with the ability to direct cellular migration. CCL2, previously named MCP-1, is critical for cellular entry to the CNS in MS and experimental autoimmune encephalomyelitis (EAE), an animal model of MS (6). Blockade of CCL2 function in systems of nonimmune-mediated CNS infiltration curbs migration of leukocytes to sites of damage (7, 8, 9). Tg expression of CCL2 targeted to oligodendrocytes leads to perivascular accumulation of leukocytes in the brain without clinical symptoms (10). Tg mice overexpressing CCL2 in astrocytes are also unaffected until 6 mo of age, after which they develop, delayed encephalopathy (11). If PTx is injected into astrocyte-directed CCL2 Tg mice at an early age (8–10 wk old), it induces a condition described as PTx-induced reversible encephalopathy dependent on MCP-1/CCL2 overexpression (PREMO) (12). Both delayed encephalopathy and PREMO are dependent on CCR2.

Various structures associated with CNS vasculature restrict the entrance of macromolecules and cells into the CNS. Endothelial tight junctions, constituting the classically defined blood-brain barrier (BBB), are responsible for limiting passage of macromolecules such as dyes (13). The mechanisms behind cell trafficking to the CNS are likely more complicated than the transport of macromolecules. In addition to endothelial tight junctions, leukocytes that infiltrate the CNS must cross the endothelial basement membrane, a separate astroglial basement membrane, the latter associated with the glia limitans (14), and the perivascular space between them. In the context of cellular infiltration, the term BBB is frequently used to describe the operational impediment to cellular entry, which can include combinations of all of these entities. The mechanisms behind CNS infiltration and, in particular, the precise nature of the actual barrier, remain to be fully explained (15), but are likely to involve degradation of extracellular matrix proteins by metalloproteinases (MPs). MPs have been implicated in virtually all CNS inflammatory diseases (16, 17).

Twenty-two matrix metalloproteinases (MMPs) have been identified (18, 19) in mice, and collectively they can degrade all components of the extracellular matrix. Six MMPs are membrane-bound. Metalloproteinase-disintegrins (a disintegrin and MP (ADAM)) are another family of MPs, which comprise >30 members (20). There are four tissue inhibitors of metalloproteinases (TIMPs). A series of articles have demonstrated that administration of synthetic broad-spectrum MP inhibitors alleviates symptoms of EAE (21, 22, 23).

In this study, we examined the role of MMPs in PTx-induced encephalopathy in CNS-specific CCL2 Tg mice. We show that leukocyte infiltration across the astroglial basement membrane, not the endothelial basement membrane, was the critical step inducing clinical symptoms. The parenchymal brain infiltration induced by PTx was associated with changes in expression of MP genes and could be alleviated by treatment with a broad-spectrum MP inhibitor.

Tg mice expressing the chemokine CCL2 in the CNS under control of a MBP promoter (10) were obtained from Bristol-Myers Squibb and maintained as a colony at the Montreal Neurological Institute. Male and female 8- to 12-wk-old Tg and wild-type (WT) C57BL/6 (The Jackson Laboratory) mice were used for all experiments. Mice were bred and maintained in a specific pathogen-free environment. Animal breeding, maintenance, and all experimental protocols were performed in accordance with the Canadian Council for Animal Care guidelines as approved by the McGill University Animal Care Committee.

PTx (10 μg/kg; Sigma-Aldrich) or HBSS (Invitrogen Life Technologies) were injected i.p. at day 0. Mice were weighed and monitored daily. BB-94 (Vernalis, a gift from Dr. V. W. Yong, University of Calgary, Alberta, Canada) was injected i.p. as a suspension of 3 mg/ml in PBS containing 0.01% Tween 80. To increase solubility, the suspension was sonicated three times for 10 s on ice. Mice were treated with BB-94 (50 mg/kg per dose) or vehicle at 1 and 4 h after injection with PTx, and once per day thereafter.

Mice were anesthetized with Somnotol (MTC Pharmaceuticals) and intracardially perfused with 20 ml of ice-cold PBS. Brain and spinal cord were collected and a single-cell suspension was generated by passing through a 70-μm cell strainer (BD Biosciences). After centrifugation on 37% Percoll (Amersham Biosciences), the myelin was removed, and cells were washed in HBSS and incubated in supernatant from the anti-FcR 24G2 (anti-CD16/32) hybridoma containing an additional 2% FCS (Sigma-Aldrich) and hamster IgG (50 μg/ml; Cedarlane Laboratories) on ice for 20 min to block nonspecific Ab binding. The cells were stained on ice for 30 min with Abs as indicated in the figure legends. Cy5-conjugated streptavidin and the Abs PE-conjugated anti-CD45, biotin-conjugated anti-CD3, FITC-conjugated anti-CD11b, and anti-Gr-1 were purchased from BD Biosciences Pharmingen. The cells were washed in HBSS with 2% FCS and sorted using a BD FACSVantage cell sorter (BD Biosciences). Flow cytometry data were analyzed using CellQuest software.

Total RNA was purified using TRIzol RNA isolation reagent (Invitrogen Life Technologies) according to the manufacturer’s protocol for whole tissue RNA extraction. For sorted cells, TRIzol was used according to the manufacturer’s protocol for RNA extraction for low amounts of RNA.

RNA (3 μg) from each brain sample was incubated with Moloney murine leukemia virus RT (Invitrogen Life Technologies) according to the manufacturer’s protocol using random hexamer primers. RNA from sorted cells was incubated with SuperScript II RT (Invitrogen Life Technologies) according to the manufacturer’s protocol using random hexamer primers and glycogen as carrier.

qPCR was done using the ABI Prism 7000 Sequence Detection System according to our previously described method and probe and primer sequences (24). Expression of 18S rRNA (primers and probes from Applied Biosystems) in cDNA samples diluted 1/1000 was used to control for differences in the extraction and reverse transcription of total RNA. All other primers were synthesized by Sigma-Aldrich. Corresponding probes were synthesized at Applied Biosystems. Each reaction was performed in 25 μl with 50% TaqMan 2× PCR Master Mix (Applied Biosystems), 100 nM each of the forward and reverse primer, and 200 nM of probe. Conditions for the PCR were 2 min at 50°C, 10 min at 95°C, and then 40 cycles, each consisting of 15 s at 95°C, and 1 min at 60°C. To determine the relative RNA levels within the samples, standard curves for the PCR were prepared by making 4-fold serial dilutions (8-fold dilutions for 18S rRNA) of cDNA. Standard curves for cycle threshold (the cycle at which the detected signal became significantly different from background signal) vs arbitrary levels of input cDNA were prepared, and the expression level of mRNA in each sample was determined. Cycle threshold values were verified to be in the linear amplification range on the appropriate standard curves. The relative expression level for each sample was calculated by dividing the expression level of the target gene by the expression level of 18S rRNA.

Mice were anesthetized with Somnotol (MTC Pharmaceuticals) and intracardially perfused with 5 ml of ice-cold PBS followed by 20 ml of 4% PFA (Fisher Scientific) in PBS. Brains were dissected, fixed for 1 h in 4% PFA in PBS at 4°C, and incubated overnight in PBS with 20% sucrose (EMD Chemicals) at 4°C. Brains were freeze-embedded in OCT (Cedarlane Laboratories) and cut in 12-μm cryostat sections. The following primary Abs were used: rabbit anti-mouse Laminin-1 (Cedarlane Laboratories) and rat anti-mouse CD45 (Serotec). Cryostat sections were fixed with 4% PFA followed by 1% Triton X-100 treatment for 20 min. They were then blocked with 20% goat serum in PBS for 1 h. Endogenous biotin was blocked with a biotin block kit (Vector Laboratories) according to the manufacturer’s instructions. Sections were incubated with primary Abs 1 h at room temperature. Biotinylated goat anti-rabbit (Vector Laboratories) and goat anti-rat Alexa555 (Molecular Probes) were added for 45 min, followed by streptavidin-Alexa488 (Molecular Probes). Nuclei were stained with Hoechst staining reagent (Molecular Probes) for 10 min. As negative controls for primary Abs, rat γ-globulin (Jackson ImmunoResearch Laboratories) and rabbit Ig (DakoCytomation) were used at equivalent concentrations. All Abs were used at empirically determined optimal dilutions. Sections were photographed using a Leica DMIRE2 fluorescent microscope (Leica Microsystems).

Mice were anesthetized with Somnotol (MTC Pharmaceuticals) and intracardially perfused with 5 ml of ice-cold PBS followed by 20 ml of 4% PFA (Fisher Scientific) in PBS. Brains were dissected, fixed for 1 h in 4% PFA in PBS at 4°C, and then embedded in paraffin. Five-micrometer sections of paraffin-embedded brains were cut on a microtome. The PB stain was performed to detect ferric iron. Briefly, brain sections were incubated at room temperature for 30 min in a solution of 5% potassium ferrocyanide (American Chemicals) and 5% hydrochloric acid (Anachemia), followed by counterstaining for 5 min in Nuclear Fast Red (Sigma-Aldrich). For H&E staining, sections were incubated for 10 min in Harris hematoxylin (Surgipath), followed by 1 min in eosin (Surgipath).

T2-weighted images were acquired using a Bruker Biospec 7T/21cm spectrometer with a 2-cm diameter quadrature volume coil (National Research Council, Institute for Biodiagnostics) using a spin-echo multislice multiecho sequence with 12 slices spanning the brain (echo time, 26.8 ms; repetition time, 2540 ms; FOV, 2.5 × 2.5 cm²; slice thickness, 0.75 mm; matrix, 256 × 256; and signal averages, 4). Anesthesia was induced using 5% halothane in 70% nitrous oxide/30% oxygen and maintained at 1.5–2% halothane during the imaging session. PTx (10 or 20 μg/kg) was injected i.p. on day 0 followed by a tail vein injection of 15 mg/kg of the USPIO ferumoxytol (Advanced Magnetics) that was readministered every 24 h. No differences were observed between mice injected with different doses of PTx. The first imaging session took place at least 1 wk before the injection of PTx and subsequent imaging was performed daily after PTx treatment. Control experiments were performed either on Tg mice injected daily with USPIO but not PTx, or on WT mice injected with both PTx and USPIO.

The CCL2 Tg mice used in this study overexpress the chemokine CCL2 specifically in the CNS under control of a truncated MBP promoter (10). CCL2 was produced throughout the white matter and leukocytes accumulated spontaneously in the perivascular space around CNS vessels in myelin-rich areas without parenchymal infiltration (10). No accompanying clinical symptoms were observed.

We used PTx to induce encephalopathy in CCL2 Tg mice. PTx caused Tg mice to lose weight compared to Tg mice injected with HBSS only. PTx-injected Tg mice showed a statistically significant (p < 0.05) weight loss starting at day 2 and lasting throughout a 5-day period with maximum weight loss at day 4. The weight loss range was between 5 and 15%. Weight loss was not observed in WT control mice injected with PTx. Clinical symptoms induced by PTx in Tg mice included tremor, inactivity, limb clasping, and/or death. Clinical symptoms were observed in 22% of PTx-treated Tg mice, whereas all PTx-treated Tg mice showed histopathology in the form of parenchymal infiltration. Clinical symptoms were not observed in WT mice injected with PTx or in Tg mice injected with HBSS (a total of 12 WT and 21 Tg controls).

Flow cytometry revealed a large number of leukocytes (CD45^high) in the CNS of CCL2 Tg mice (Fig. 1,A, left panel). The combination of CD45 and CD11b staining makes it possible to distinguish CNS-resident microglia (CD45^dimCD11b⁺) and a population consisting of blood-derived macrophages and granulocytes (CD45^highCD11b⁺) (25). These flow cytometry profiles did not change dramatically following administration of PTx (Fig. 1,A, right panel). However, the total number of cells was significantly higher in the PTx-injected mice (3.85 ± 0.77 × 10⁶ compared to 2.38 ± 0.39 × 10⁶ in the HBSS-injected mice, an increase of 62%, p < 0.05). When numbers of individual cell types were compared (Fig. 1 B), it was clear that the increase in total cell number in PTx-treated mice was mainly due to an increase in CD45^highCD11b⁺ cells, which doubled in number. The vast majority of these cells were macrophages, because we observed <1% granulocytes, identified by Gr-1 staining (data not shown). T cells were isolated as CD45^highCD3⁺ cells (data not shown). There was a 3-fold increase in T cells, yet T cells remained a relatively small proportion of the infiltrate. Microglial numbers were unchanged.

Infiltration into the brain was followed over the course of 1 wk using USPIO-enhanced MRI. The dextran-covered USPIO particles were injected into the blood and engulfed by actively phagocytosing cells. Cellular infiltration of USPIO-loaded cells can be followed on T2-weighted MRI, where the iron in the USPIO particles results in an area with low signal intensity on the scan (26, 27). USPIO was injected on day 0 and every 24 h thereafter. The mice were imaged every day and results for one mouse are shown in Fig. 2,A. There was a clear development of lesion areas on the MRI scans with low signal intensities due to magnetic susceptibility effects representing infiltration of USPIO-loaded cells (Fig. 2,A, arrowheads). These areas became noticeable starting on day 2 or 3 and persisted to day 7. The development of lesions was dynamic, with some lesion areas eventually decreasing in size. No lesions were observed in any WT control mice injected with PTx and USPIO (Fig. 2,B) or in any CCL2 Tg mice receiving USPIO, but HBSS instead of PTx (Fig. 2,C). Infiltrating USPIO-loaded cells were visualized in situ by PB staining, which stains iron (and, therefore, USPIO) blue in tissue sections. In Tg mice receiving PTx and USPIO, blue-stained cells were seen around vessels and in the surrounding parenchyma (Fig. 2,D, arrows), corresponding to areas of low signal intensity on MRI scans. The lesion area on the scan appears larger than the actual infiltrated area because of blooming of the MRI signal (28). No PB staining was observed in Tg mice receiving USPIO but no PTx (Fig. 2,E) or in WT mice receiving both PTx and USPIO (Fig. 2 F), indicating that parenchymal infiltration of cells that actively phagocytosed USPIO requires signals from both PTx and CCL2. Based on the observation that mice showed maximal weight loss at day 4, and that USPIO-enhanced MRI showed dynamic lesions increasing up to day 7, we decided to sacrifice animals at day 5 for further analysis. At this time, weight loss and the size of lesions on MRI scans were both prominent.

We analyzed expression of two proinflammatory cytokines and six MP genes using real-time PCR (Fig. 3). The Tg expression of CCL2 alone induced IL-1β and TNF-α expression compared to WT mice. Both were further up-regulated in the PTx-treated Tg mice. The six MP genes included ADAM-12 and three secreted MMPs: MMP-8, -10, and -12, the membrane-bound MMP-15, as well as the physiological MP inhibitor TIMP-1. These six genes were chosen on the basis of their selective involvement in EAE, where expression of MMP-8, -10, -12, ADAM-12, and TIMP-1 were markedly induced, whereas MMP-15 was down-regulated (24). Expression of MMP-10, MMP-12, and TIMP-1 were inherently significantly higher in the Tg mice compared to WT mice. MMP-8, MMP-10, ADAM-12, and TIMP-1 were significantly up-regulated in Tg mice by PTx. MMP-12 was not significantly up-regulated by PTx in Tg mice (p < 0.07). Expression of the membrane-bound MMP-15 was unaffected in all cases.

We have previously shown (24) that MMP-8 is primarily expressed by infiltrating CD45^highCD11b⁺ cells, that MMP-15 is primarily expressed by microglia, and that ADAM-12 is exclusively expressed by T cells among the infiltrating cells in EAE. We analyzed expression of these three MPs in cells sorted from CCL2 Tg mice (Fig. 4). We found up-regulated expression of MMP-8 in the CD45^highCD11b⁺ population. MMP-15 was expressed almost exclusively by microglia and levels were unaffected by PTx injection. ADAM-12 was specifically expressed by T cells among the populations that were sorted and did not show significant change in expression after PTx injection.

BB-94 is a synthetic broad-spectrum MP inhibitor, which acts by competitive reversible binding to the catalytic site of MPs (29). Treatment of PTx-injected CCL2 Tg mice with BB-94 curbed weight loss, with weights of treated animals being significantly higher than the weights of untreated mice starting on day 3 and lasting to day 5 (Fig. 5). Administration of BB-94 alone did not affect body weight (data not shown).

Given the findings that PTx induced CNS infiltration, MP up-regulation, and weight loss, and that the broad-spectrum MP inhibitor BB-94 curbed weight loss, we investigated whether daily treatment with BB-94 would inhibit CNS infiltration. CNS infiltration occurs by at least two steps: 1) by transendothelial migration and 2) migration across the glia limitans. By using flow cytometry on whole CNS samples from perfused mice, we could measure the migration of CD45^high cells across the endothelium into the nonperfusable space, which includes both the perivascular space and the parenchyma. As expected, PTx caused a dramatic increase of CD45^high cells in the CNS of CCL2 Tg mice 5 days after administration (Fig. 6). Interestingly, daily treatment with BB-94 after injection of PTx did not reduce this infiltration. This points to a role for PTx in potentiating accumulation of cells in the perivascular space of Tg mice, which was not affected by the MP inhibitor. BB-94 alone without PTx did not diminish total CNS infiltration (Fig. 6). Indeed the BB-94-treated mice showed a trend toward higher levels of infiltration, although not statistically significant, possibly reflecting an effect on turnover of cells in the perivascular space. We confirmed by histology that treatment with BB-94 alone for 5 days did not reduce the level of infiltration in Tg mice (results not shown).

To assess the effect of MP inhibition on the PTx-induced parenchymal infiltration, distinct from perivascular accumulation, we stained for CD45 and laminin-1 in sections of brains from PTx-injected Tg mice treated with BB-94 or PBS, as well as control Tg mice receiving only PBS (Fig. 7,A). Laminin-1 stains both the endothelial basement membrane (Fig. 7,A, left panel, thin arrow) and the astroglial basement membrane (Fig. 7,A, left panel, thick arrow) of the BBB. CD45 is expressed at high levels on infiltrating leukocytes (Fig. 7,A, middle panel, arrowheads, and Fig. 1,A). Consequently, this staining allows us to identify the location of leukocytes in relation to CNS vessels. We counted infiltrating cells within a 100-μm distance of the astroglial basement membrane around distinct infiltrated vessels. In control Tg mice, numbers of infiltrating cells were at a basal level (Fig. 7,A, left panel). Injection of PTx caused striking parenchymal infiltration associated with infiltrated blood vessels (Fig. 7,A, middle panel). Treatment with BB-94 abolished this parenchymal infiltration (Fig. 7,A, right panel). Counting CD45-positive cells within a distance of 100 μm from the astroglial basement membrane of infiltrated vessels on each section demonstrated that BB-94 treatment reduced PTx-induced brain infiltration to a level close to basal level (Fig. 7 B).

The process of cellular entry to the inflamed CNS parenchyma involves leukocyte migration across the endothelium, the glia limitans, and the perivascular space between them, driven by multiple stimuli that include chemokines and action of proteinases. In the system we have studied, unmanipulated CCL2 Tg mice contain large numbers of CD45^high leukocytes in the perivascular space surrounding CNS vessels. The bacterial toxin PTx induced leukocytes to cross the astroglial basement membrane in Tg mice, and this led to weight loss and clinical symptoms. Therefore, the combination of a bacterial product and chemokine overexpression resulted in parenchymal infiltration. This demonstrates that exposure of a primed CNS milieu to an environmental stimulus can result in disease; in this case, encephalopathy. Gene expression of two MPs was induced by the chemokine CCL2 and one of these and two others were further induced by PTx. Because parenchymal CNS infiltration could be curbed by treatment with a MP inhibitor, the PTx-induced expression of effector molecules such as MPs likely promotes brain infiltration. The molecular mechanisms behind this action of PTx remain to be established, but the involvement of innate immune receptors such as TLRs is an interesting possibility (30, 31, 32).

The symptoms we observed following PTx injection in CCL2 Tg mice were similar to, but less severe than, the PREMO condition described by Huang et al. (12) to occur in astrocyte-directed CCL2 Tg mice treated with s.c. CFA and i.v. PTx. Huang et al. (12) demonstrated that a milder version of PREMO could be induced with PTx alone, but not CFA, LPS, or staphylococcal enterotoxin B alone, illustrating the unique ability of PTx to induce symptoms. In the present study, we used a lower dose of PTx and no CFA. Consequently, we observed lower incidence, less severe symptoms, and a lower mortality. The symptoms of encephalopathy and deaths in our experiment occurred in an unpredictable pattern with some mice dying without showing previous symptoms and some mice recovering from symptoms.

We previously demonstrated (24) that expression of a number of MP genes was altered in spinal cord of mice with adoptive transfer EAE. Although EAE and the encephalopathy induced by PTx in CCL2 Tg mice are different pathologies, they share the process of parenchymal infiltration and may share fundamental mechanisms of leukocyte migration across the BBB. In the spinal cord of mice with EAE, MMP-8, MMP-10, MMP-12, ADAM-12, and TIMP-1 are all up-regulated >3-fold, whereas MMP-15 is down-regulated >3-fold. Comparison of the findings in EAE and in infiltration driven by the synergistic effects of PTx and a chemokine may be instructive for understanding mechanisms of CNS inflammation.

MMP-8 is strongly expressed by granulocytes in EAE (24). In CCL2 Tg mice, <1% of CNS cells analyzed by flow cytometry are granulocytes (CD45^highGr1⁺). It is more likely that CD45^highCD11b⁺ macrophages are responsible for the up-regulation of MMP-8 mRNA following PTx administration. It has elsewhere been demonstrated (33) that monocytes/macrophages can express MMP-8. Given the 8.6-fold increase in expression level per CD45^highCD11b⁺ cells (Fig. 4,A) and the 2.5-fold increase in numbers of these cells (Fig. 1,B), one could expect an increase of MMP-8 expression in the whole brain tissue of ∼21.5-fold in PTx-injected CCL2 Tg mice if CD45^highCD11b⁺ cells were the only source of MMP-8 message. The increase in MMP-8 expression in Tg mice injected with PTx was 21.7-fold (Fig. 3,B). Therefore, the increase in MMP-8 at the whole brain level can be accounted for entirely by influx of CD45^highCD11b⁺ cells. We found ADAM-12 to be exclusively expressed by T cells among the cell types analyzed (Fig. 4,C). T cells did not up-regulate ADAM-12 expression significantly following PTx, but the total number of T cells in the CNS of PTx-treated Tg mice rose 3.2-fold (Fig. 1,B). The overall brain ADAM-12 expression level in Tg mice increased 3.4-fold following PTx (Fig. 3,F); therefore, influx of ADAM-12-expressing T cells could account for this increase. ADAM-12 has also been reported (34) to be expressed by oligodendrocytes, which could explain the ADAM-12 expression seen in WT control brain (Fig. 3 F). Based on our study, a contribution of MMP-8 and ADAM-12 message from cell types other than T cells, microglia, or CD45^highCD11b⁺ cells cannot be excluded. However, the size of PTx (117 kDa) should normally prevent it from readily crossing the BBB to enter the parenchyma and affect glial cells directly. This is supported by the finding that expression of microglial membrane-bound MMP-15, which we previously found to be down-regulated by microglia in EAE (23), was not altered in the Tg mice or by PTx administration. Also, we observed no change in microglial cell numbers in PTx-treated Tg mice.

The up-regulation of MMP-10 following PTx administration is consistent with EAE studies, reflecting the recruitment of macrophages and T cells (24). The expression level of the macrophage-specific MMP-12 was inherently higher in Tg mice compared to WT, but neither PTx nor the leukocyte infiltration that it induced, led to significantly increased MMP-12 expression. This may reflect differential regulation of infiltrating macrophages compared to EAE, where MMP-12 is the most highly up-regulated MMP (24, 35). Interestingly, Weaver et al. (35) found that MMP-12 null mice developed more severe EAE than WT mice, indicating that the role of MMP-12 in CNS infiltration is complex. TIMP-1 is an inhibitor of most MPs (36) and is expressed by macrophages (24) and astrocytes (37) in EAE. The higher level of TIMP-1 in Tg mice compared to control, and the PTx-induced up-regulation, probably reflects CCL2- and PTx-driven influx of monocytes/macrophages in the Tg mice and/or a response by astrocytes, which were unavailable to us by flow cytometric analysis.

PTx up-regulated expression of proinflammatory cytokines TNF-α and IL-1β in Tg mice. Both IL-1β and TNF-α are associated with weight loss (38) and the increased expression level of these cytokines might contribute to the observed PTx-induced weight loss in Tg mice. In addition, proinflammatory cytokines could increase leukocyte migration by a direct effect on the CNS vessels (39, 40).

PTx is classically considered to open the BBB (41). However, EAE can be induced in some mouse strains independently of PTx and its role in EAE is likely more complex than increasing vascular permeability (30). This is supported by findings in EAE, where opening of the BBB in mice was found to be independent of PTx, although PTx did enhance severity of disease (42). PTx is used to promote development of experimental autoimmunity, possibly by its potentiating the effects of histamine in blood vessels. The gene controlling susceptibility to PTx-induced hypersensitivity to histamine (Bphs) was found to be identical to the histamine 1 receptor (H1R) gene, and H1R-deficient mice had delayed onset and decreased severity of EAE (43).

In our study, administration of PTx did not increase expression of any of the investigated genes in the brains of WT mice. Nor did it induce weight loss or clinical symptoms in WT mice. Only in the case of overexpression of CCL2 in brains of Tg mice did PTx have an effect, presumably either by acting on the already accumulated leukocytes, or on leukocytes in blood, which were then attracted to the brain by the CCL2 chemokine. The synergizing effects of a chemokine-rich brain environment and exposure to the environmental stimulus PTx result in dramatic up-regulation of proinflammatory and MP genes, accompanied by clinical symptoms.

CCL2 Tg mice have large numbers of leukocytes accumulated in the perivascular space, and the fact that perivascular accumulation of leukocytes does not cause disease is of interest. Similar observations were made when EAE was induced in TNF-α-deficient mice (44) and in mice in which peripheral macrophages have been depleted (45). Perivascular accumulation of leukocytes is therefore not in itself enough to cause pathology. In this context, it is important to take into consideration that cellular entry to the CNS involves crossing two separate structures: the endothelium and the glia limitans (14). The endothelium of CNS vessels allows leukocyte transit in states of perivascular accumulation without accompanying clinical symptoms. In our study, migration across the endothelial basement membrane alone was not sufficient to initiate clinical symptoms. The critical event for disease was the PTx-induced migration across the astroglial basement membrane leading to parenchymal infiltration. Our results show that leukocyte migration across the astroglial basement membrane is dependent on action of MPs, since it could be blocked by treatment with a MP inhibitor. Conversely, endothelial basement membrane transmigration induced by CCL2 and potentiated by PTx was not blocked with BB-94 treatment. Treatment of Tg mice with BB-94 alone for 5 days did not lead to diminished perivascular accumulation. Therefore, the beneficial effect of BB-94 treatment in our study is not due to blockade of the transendothelial migration, but lies in blocking the parenchymal infiltration across the glia limitans. It should be noted, however, that in vitro studies (24, 35) of human leukocyte migration across endothelial basement-like structures suggest that MPs do play a role in this. Our study emphasizes the need to distinguish between the two basement membranes of the BBB in the context of CNS infiltration. Recently, Agrawal et al. (49) showed that MMP-2- and MMP-9-mediated cleavage of dystroglycan, which anchors astrocyte endfeet to the astroglial basement membrane, is critical for leukocyte trafficking across this membrane. The particular MP profiles needed to cross the endothelial basement membrane and the astroglial basement membrane may not be the same, as the two membranes are structurally different and distinct in their composition of extracellular matrix proteins, especially laminin isoforms (14, 50).

Our results demonstrate that exposure to the environmental agent PTx in concert with CNS chemokine expression can promote parenchymal brain infiltration through an effect on MPs. This illustrates potential interplay between infection and inflammation in promoting CNS autoimmune disease. We also demonstrate that it is MP-dependent leukocyte migration across the astroglial basement membrane of the BBB, not across the endothelial basement membrane, which is the critical event in inducing disease in our system.

We thank Dina Dræby, Pia Nyborg Nielsen, Lyne Bourbonnière, and Maria Caruso for excellent technical assistance and Alicia Babcock for input on the manuscript. We also thank Dr. V. Wee Yong (University of Calgary, Alberta, Canada) for helpful discussions, and we thank Dr. V. Wee Yong and Vernalis (R&D Centre) for provision of BB-94.

The authors have no financial conflict of interest.