Crawling Phagocytes Recruited in the Brain Vasculature after Pertussis Toxin Exposure through IL6, ICAM1 and ITGαM

Abstract

The cerebral vasculature is constantly patrolled by rod‐shaped leukocytes crawling on the luminal endothelial surface. These cells are recruited in greater numbers after exposure to bacterial lipopolysaccharide (LPS) by a mechanism involving tumor necrosis factor (TNF), interleukin‐1β (IL1β) and angiopoietin‐2 (Angpt2). Here, we report that the population of crawling leukocytes, consisting mainly of granulocytes, is also increased in the brains of mice suffering from experimental autoimmune encephalomyelitis (EAE) or injected with pertussis toxin (PTX), which is commonly used to induce EAE. However, this recruitment occurs through an alternative mechanism, independent of Angpt2. In a series of experiments using DNA microarrays, knockout mice and neutralizing antibodies, we found that PTX acts indirectly on the endothelium in part through IL6, which is essential for the post‐transcriptional upregulation of intercellular adhesion molecule 1 (ICAM1) in response to PTX but not to LPS. We also found that phagocytes adhere to brain capillaries through the interaction of integrin αM (ITGαM) with ICAM1 and an unidentified ligand. In conclusion, this study supports the concept that PTX promotes EAE, at least in part, by inducing vascular changes necessary for the recruitment of patrolling leukocytes.

INTRODUCTION

Multiple sclerosis (MS) is an autoimmune demyelinating disease often characterized by unpredictable episodes of exacerbation and remission. While its etiology is still unknown, epidemiological studies suggest that environmental factors influence its course. For example, the risk of exacerbation is significantly increased during or after common infections of the upper respiratory tract 1, 6, 10, 21, 26, 32, 34. Several non‐exclusive mechanisms have been proposed to explain how infections may influence MS, including adjuvant actions, molecular mimicry, bystander activation and epitope spreading (23).

To induce encephalomyelitis (EAE), the most widely used model of MS, mice immunized with myelin antigens or transplanted with myelin‐reactive T lymphocytes are commonly injected with pertussis toxin (PTX), which is used as a surrogate for infection to increase the severity of the disease. PTX is a hexameric protein produced by Bordetella pertussis, the bacterium that causes whooping cough. The subunit S1 forms the A protomer responsible for the enzymatic activity, while the four other subunits (including two copies of S4) form the B oligomer responsible for binding to the cell surface (33). The best‐known action of PTX involves its endocytosis and transport to the endoplasmic reticulum (28). There, the A protomer catalyzes the ADP‐ribosylation of the α subunits of G proteins 5, 8, thereby preventing their interaction with G protein‐coupled receptors. A less understood mechanism relies on the ability of the B oligomer to interact with transmembrane receptors and modulate intracellular signaling pathways. For example, it has been shown that PTX stimulates P‐selectin expression in the cerebral endothelium by acting through toll‐like receptor 4 (TLR4) (17), better known as the signaling receptor for lipopolysaccharide (LPS). Through such actions, PTX induces responses that presumably promote EAE, such as leukocyte adhesion to the cerebral vasculature, activation of antigen‐presenting cells and differentiation of CD4⁺ T lymphocytes (29).

We have recently uncovered a mixed population of phagocytes that constantly patrol the cerebral vasculature by crawling on the luminal endothelial surface 2, 35. Our knowledge about these cells can be summarized as follows: (i) they include granulocytes and monocytes, whose proportions remain uncertain; (ii) they adopt a rod‐shaped morphology when crawling in the capillary network, facilitating their discrimination from parenchymal phagocytes; (iii) they are recruited upon stimulation with LPS by a mechanism involving the pro‐inflammatory cytokines tumor necrosis factor (TNF) and interleukin‐1β (IL1β), and the vascular destabilization factor angiopoietin‐2 (Angpt2); and (iv) neutralization of the latter using a peptide‐Fc fusion protein called L1‐10 blocks the recruitment of crawling phagocytes during endotoxemia. An important question to be addressed is whether crawling phagocytes are recruited in other conditions through the same mechanism as the one induced by LPS. Considering their privileged position at the blood–brain interface and the roles of myeloid cells in immunity, it is plausible that crawling phagocytes are involved in protective neuroinflammatory responses against infection, injury and degenerative diseases, as well as in detrimental neuroinflammatory disorders, such as MS.

The initial goals of the present study were (i) to examine whether the population of crawling leukocytes increases during EAE, a model of MS, and in response to PTX, which is commonly used to induce EAE without clear understanding of its mechanism of action; and (ii) to explore the possibility of using the Angpt2 inhibitor L1‐10 to block such recruitment. The results confirmed that crawling leukocytes are attracted in greater numbers in response to PTX and EAE, but through an Angpt2‐independent mechanism. To clarify this alternative mechanism, our ultimate goal was to compare the effects of PTX and LPS using a combination of in vitro and in vivo approaches.

METHODS

Animals

C57BL/6 mice were purchased from Charles River Laboratories (Montreal, Quebec, Canada). Mice deficient in IL6 or intercellular adhesion molecule 1 (ICAM1) and their wild‐type controls (C57BL/6 background) were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). The animals were adapted to standard laboratory conditions for at least 1 week before any manipulation. The experiments were done on 8–10‐week‐old males in accordance with the current guidelines of the Canadian Council on Animal Care.

Toxins

PTX (List Biological Laboratories, Campbell, CA, USA) was dissolved in phosphate‐buffered saline (PBS) and injected intraperitoneally at a dose of 20 µg/kg. LPS from Escherichia coli O55:B5 (Sigma‐Aldrich, Oakville, Ontario, Canada) was dissolved in PBS and injected intraperitoneally at a dose of 2 µg/kg or 1 mg/kg. Control mice were injected with PBS.

Neutralizing antibodies

One hour before PTX injection, mice were injected via a tail vein with either one of the following antibodies at a concentration of 4 mg/kg and diluted in saline: anti‐ITGαL (rat IgG_2a; BD Biosciences, Mississauga, Ontario, Canada), anti‐ITGαM (rat IgG_2b; BD Biosciences), anti‐ITGα4 (rat IgG_2a; R&D Systems, Minneapolis, MN, USA), anti‐ITGβ1 (rat IgG_2a; R&D Systems), anti‐ITGβ2 (rat IgG₁; BD Biosciences) and anti‐ITGβ7 (rat IgG_2a; R&D Systems). Control mice were injected with isotype‐matched non‐specific antibodies.

EAE induction and clinical evaluation

Mice were injected subcutaneously on days 0 and 7 with 200 µL (100 µL/site) of emulsion containing 300 µg of myelin oligodendrocyte glycoprotein peptide 35–55 (AnaSpec, Fremont, CA, USA) dissolved in saline and mixed with an equal volume of complete Freund's adjuvant containing 500 µg of killed Mycobacterium tuberculosis H37 RA (BD Biosciences). The animals were also injected intraperitoneally with PTX at a dose of 20 µg/kg immediately and 48 h after the first immunization. The clinical signs were monitored daily and scored as follows: 0, no detectable sign; 1, tail flaccidity; 2, hindlimb weakness and poor righting ability; 3, hindlimb paralysis/paresis; 4, hindlimb paralysis and forelimb paraparesis; 5, moribund or dead.

L1‐10

EAE mice were injected subcutaneously every other day with 4 mg/kg of L1‐10 (kindly provided by Amgen Inc., Thousand Oaks, CA, USA) diluted in PBS. Control mice were injected with PBS.

RNA isolation and cDNA synthesis

Total RNA was isolated from tissue samples and cell cultures using the TRI Reagent (Sigma‐Aldrich) or GenElute Mammalian Total RNA Miniprep Kit (Sigma‐Aldrich), respectively. RNA integrity and yield were assessed by capillary electrophoresis using the Bioanalyzer 2100 (Agilent Technologies, Mississauga, Ontario, Canada). Complementary DNA was generated from 2 µg of total RNA using random primer hexamers and Superscript II (Invitrogen, Carlsbad, CA, USA). The RNA template was digested with RNase H (Invitrogen).

Microarray analysis

Total RNA (200 ng) was used to produce biotinylated single‐strand DNA targets using the GeneChip Whole Transcript Sense Target Labeling Assay (Affymetrix, Santa Clara, CA, USA). The targets were hybridized to Affymetrix GeneChip Mouse Gene 1.0 ST arrays for 16 h at 45°C with constant rotation at 60 rpm. The arrays were washed and stained with streptavidin–phycoerythrin (10 µg/mL) and biotinylated goat anti‐streptavidin (3 mg/mL) using the Affymetrix Fluidics Station 450 (protocol FS450‐0007), then read using the Affymetrix GeneChip Scanner 3000 G7. The data were processed using the RMA algorithm in Expression Console software (Affymetrix) and filtered using the following criteria: (i) expression level >100 in at least one sample; and (ii) mean fold change >1.5 between at least two groups. Using the MeV software (31), the filtered data were log₂‐transformed and analyzed using the LIMMA algorithm with an alpha level of 0.05. The list of modulated genes was filtrated for redundant entries and used to create a Venn diagram with GeneVenn (27).

Quantitative polymerase chain reaction (PCR)

Quantitative PCR (qPCR) was conducted in a final volume of 15 µL containing 1 × Universal PCR Master Mix (Applied Biosystems, Foster City, CA, USA), 5 µL cDNA, 250 nM Amplifluor Uniprimer probe (Chemicon, Temecula, CA, USA), 10 nM Z‐tailed primer (18S rRNA, 5′‐ACTGAACCTGACCGTACACGGTACAGTGAAACTGCGAATG‐3′; ICAM1, 5′‐ACTGAACCTGACCGTACACAGGACCGGAGCTGAAAAGTT‐3′; VCAM1, 5‐ACTGAACCTGACCGTACAGCCCACTCATTTTAATTACTGGATC‐3′) and 100 nM untailed primer (18S rRNA, 5′‐CCAAAGGAACCATAACTGATTTAATGA‐3′; ICAM1, 5′‐CTGGAGACGCAGAGGACCTT‐3′; VCAM1, 5′‐AGGTGGAGGTCTACTCATTCCCT‐3′). The amplification was performed using the ABI PRISM 7900 Sequence Detector (Applied Biosystems) under the following conditions: 2 min at 50°C, 4 min at 95°C, followed by 55 cycles of 15 s at 95°C and 30 s at 55°C. Ribosomal 18S RNA was used to normalize each sample.

Immunostaining

Immunostaining was performed as described previously (2) using the following primary antibodies: rat anti‐CD3 (1:500; Serotec, Raleigh, NC, USA), rat anti‐CD45 (1:1000; BD Biosciences), rabbit anti‐CD45 (1:500; Santa Cruz Biotechnology, Santa Cruz, CA, USA), rat anti‐Ly6G/C (GR1) (1:500; Invitrogen), rat anti‐Ly6G (1:5000; BD Biosciences), hamster anti‐ICAM1 (1:500; BD Biosciences), rat anti‐VCAM1 (1:500; BD Biosciences).

In situ hybridization

Brain sections were analyzed by radioisotopic in situ hybridization as described previously (37). Riboprobes were synthesized from linearized plasmids containing a fragment of the mouse cDNA of Angpt2 (1544 bp), ICAM1 (1474 bp) or VCAM1 (1507 bp).

Stereology

As described previously (2), cell number and volume estimates were obtained using the optical fractionator and Cavalieri methods, respectively.

Microscopy

Confocal images were acquired using an Olympus Fluoview microscope (BX61) by sequential scanning using a z‐separation of 0.1 µm. For each fluorochrome, the upper and lower thresholds were set using the range indicator to minimize data loss. Bright‐field and dark‐field micrographs were taken using a Retiga EX monochrome camera mounted on a Nikon E800 microscope. Images were adjusted for contrast, brightness and sharpness using Photoshop 8 (Adobe Systems, San Jose, CA, USA).

Enzyme‐linked immunosorbent assay

Blood samples were obtained by intracardiac puncture, allowed to clot and centrifuged for 20 min at 2000 × g. Serum was collected and analyzed for IFNγ, IL1β, IL6 and TNF using enzyme‐linked immunosorbent assay (ELISA) kits (R&D Systems).

Flow cytometry

Blood samples were blocked for 15 min with 5 µg/mL anti‐CD16/CD32 antibody (BD Biosciences) and stained for 30 min on ice with 2 µg/mL anti‐Ly6G/C (Gr1) antibody. After hemolysis and fixation with the Whole Blood Lysing Reagent Kit (Beckman Coulter, Mississauga, Ontario, Canada), cells were analyzed with a BD FACSCanto II flow cytometer and FlowJo software (Tree Star, San Carlos, CA, USA).

Statistical analyses

Data are expressed as mean ± standard error. Means were compared using the unpaired Student's t‐test, one‐way ANOVA or two‐way ANOVA. Alternatively, the Wilcoxon rank‐sum test was used when the distribution was abnormal. Bonferroni or Student's t‐tests were performed for post hoc multiple comparisons. All these analyses were performed using JMP (SAS Institute, Cary, NC, USA) or Prism 4 (GraphPad Software, La Jolla, CA, USA) with a significance level of 5%.

RESULTS

Crawling phagocytes are recruited in the cerebral vasculature during PTX exposure and EAE

To determine whether crawling leukocytes can be recruited in the cerebral vasculature in response to stimuli other than LPS, we performed immunohistochemistry for the pan‐hematopoietic marker CD45 on brain sections from mice euthanized either 6 h after intraperitoneal injection of PTX (20 µg/kg) or 12–16 days after induction of EAE by active immunization against myelin oligodendrocyte glycoprotein (Figure 1A). As estimated using stereological procedures, the number of crawling leukocytes (ie, CD45⁺ intraluminal leukocytes with a rod‐shaped or rounded morphology) was at least twice higher in the cerebral cortex of these mice compared with that of sham‐treated mice (Figure 1B–C). For comparison, we also analyzed brain sections from mice treated with LPS and found a roughly similar increase when LPS was administered at a high dose (1 mg/kg), but no difference when administered at a low dose (2 µg/kg) (Figure 1B). These results are consistent with our previous study showing an increase in crawling leukocytes during endotoxemia that peaks at 6 h and that lasts for at least 3 days (2).

Figure 1.

Increased numbers of crawling leukocytes in the cerebral vasculature during PTX exposure and EAE. A. Nomarski interference micrographs of crawling leukocytes with a rod‐shaped or rounded morphology in mouse brain sections stained for CD45 by immunohistochemistry. B. Counts of CD45⁺ crawling leukocytes in the cerebral cortex of mice euthanized 6 h after intraperitoneal injection of PBS, LPS or PTX. *Significantly different from PBS according to post hoc Student's t‐tests. ANOVA: P = 0.012 (rod cells) or 0.0005 (round cells). Sample size: four to nine mice per group. C. Counts of CD45⁺ crawling leukocytes in the cerebral cortex of mice euthanized before (day 0) or 12–16 days after EAE induction. *Significantly different from day 0 according to post hoc Student's t‐tests. ANOVA: P = 0.011 (rod cells) or 0.31 (round cells). Sample size: three to six per group. D. Confocal micrographs of a rod‐shaped leukocyte doubly labeled for GR1 (Ly6C/G) and CD45 by immunofluorescence. E. Percentages of CD45⁺ crawling leukocytes positively labeled with the GR1 antibody. *Significantly different from PBS according to post hoc Student's t‐tests. ANOVA: P < 0.0001. Sample size: 6 per group. F. Micrograph of a rod‐shaped leukocyte immunostained with the GR1 antibody. G. Counts of GR1⁺ leukocytes in the cerebral cortex of mice treated with PBS, LPS, or PTX. *Significantly different from PBS according to post hoc Student's t‐tests. ANOVA: P = 0.0032 (rod cells) or 0.039 (round cells). Sample size: five to six per group. H. Micrograph of a rod‐shaped leukocyte immunostained with an anti‐Ly6G antibody. I. Counts of Ly6G⁺ leukocytes in the cerebral cortex of mice treated with PBS, LPS or PTX. *Significantly different from PBS according to post hoc Student's t‐tests. ANOVA: P = 0.0078 (rod cells) or 0.027 (round cells). Sample size: five to six per group. J. Counts of Ly6G⁺ leukocytes in the cerebral cortex of EAE mice. *Significantly different from PBS according to post hoc Student's t‐tests. ANOVA: P = 0.036 (rod cells) or 0.14 (round cells). Sample size: three to six per group.

To clarify the identity of the newly recruited crawling leukocytes, we first double‐stained brain sections using antibodies against CD45 and Ly6C/G (Figure 1D). The latter antibody, called GR1, labels both granulocytes and inflammatory (type 1) monocytes (13). Under basal conditions, the ratio of GR1⁺ to GR1^‐ rod‐shaped leukocytes was roughly 1:3. Interestingly, this ratio was reversed to nearly 3:1 after injection of PTX or LPS (Figure 1E). No significant change was observed in the case of round leukocytes (Figure 1E). The increases in GR1⁺ cells were confirmed by stereological counting using adjacent sections stained for GR1 by immunohistochemistry (Figure 1F–G). We next addressed the question of whether the newly recruited GR1⁺ cells were granulocytes or monocytes using an antibody against the granulocyte‐specific antigen Ly6G 12, 24 (Figure 1H). The results revealed increases in Ly6G⁺ cells (Figure 1I) that were comparable with the increases in GR1⁺ cells (Figure 1G) and CD45⁺ cells (Figure 1B). Significant increases in rod‐shaped Ly6G⁺ cells were also recorded in EAE mice (Figure 1J). Finally, we examined whether the population of crawling leukocytes included T cells by performing immunohistochemistry for the lymphocytic marker CD3. No CD3⁺ cell was detected in the brains of mice treated with either one of the toxins, whereas a significant fraction of rod‐shaped leukocytes were CD3⁺ in EAE mice killed 14 or 16 days after immunization (72–90 CD3⁺ cells/mm³, representing ∼29%–38% of the total population of rod‐shaped leukocytes) (Figure S1). Taken together, these observations indicate that the increase in the population of crawling leukocytes is a non‐specific response triggered by diverse immune stimuli and attributable largely to the recruitment of granulocytes.

No involvement of Angpt2 during PTX exposure and EAE

We have previously reported that the recruitment of crawling phagocytes in the cerebral vasculature during LPS exposure occurs through an Angpt2‐dependent mechanism and can be blocked using an anti‐Angpt2 peptide‐Fc fusion protein called L1‐10 (2). To determine whether PTX and EAE trigger this mechanism, we performed in situ hybridization for Angpt2 mRNA on brain sections from mice euthanized 3, 6, 12 or 24 h after PTX injection or euthanized when a clinical scores of 1 to 4 were obtained after EAE induction (n = 4–6 mice per group). No hybridization signal was detected in any sections (Figure 2A). In contrast and as observed previously (2), strong signals were detected in control mice euthanized 6 h after LPS injection (Figure 2A). Similar results were obtained by qPCR (Figure 2B). To definitively exclude a role for Angpt2 in EAE, we injected EAE mice once every other day with L1‐10 (4 mg/kg), a regimen previously shown to inhibit Angpt2 activity 25, 36. No significant difference in disease severity was detected in these mice compared with sham‐treated animals (Figure 2C). Therefore, these results suggest that PTX and EAE induce the recruitment of crawling phagocytes by a mechanism that is independent of Angpt2 and distinct from the one previously described for LPS (2).

Figure 2.

Angpt2 is not expressed in the brain in response to PTX and is not involved in EAE. A. Dark‐field micrographs showing in situ hybridization signals for Angpt2 mRNA in the brain of a mouse treated with LPS (upper‐right panel) but not in the brains of mice treated with PBS or PTX or subjected to EAE. The animals were euthanized 6 h after injection or when a clinical score of 4 was reached after EAE induction. B. Quantitative PCR quantification of Angpt2 mRNA in the brains of mice subjected to the indicated conditions. *Significantly different from PBS. ANOVA, P < 0.0001. Sample size: five to seven per group. Clinical scores of the EAE mice: from 2 to 3.5. C. No difference in the severity of EAE was detected in mice treated with the Angpt2 inhibitor L1‐10 (4 mg/kg, once every other day) compared with mice treated with PBS. Two‐way ANOVA with repeated measures: P = 0.93. Sample size: eight per group.

PTX induces vascular changes involved in phagocyte adhesion

In seeking to understand the mechanism by which PTX promotes the recruitment of crawling phagocytes, we hypothesized that PTX might act directly on the vasculature as does LPS (9). To test this possibility, we performed microarray analysis on RNA samples from cerebral endothelial cells cultured for 6 h in the presence of PTX (100 µg/mL), LPS (10 µg/mL) or PBS. The raw data (Table S1) were filtered to retain only the genes that were significantly expressed in at least one sample (hybridization signal >100) and whose average expression varied significantly between at least two conditions by a factor >1.5 (P < 0.05). As illustrated in Figure 3A,B, PTX altered the expression of only 14 genes by a magnitude ranging from 1.5 to 3. In contrast, LPS, at a lower molarity, produced a greater effect, with 233 genes that were modulated by a magnitude of 1.7 to 18. All of the genes modulated by PTX were also modulated, to a greater extent, by LPS. To validate these data, we quantified by qPCR two genes known to be involved in leukocyte recruitment, namely VCAM1 and ICAM1 (16). The results confirmed that PTX induced only modestly VCAM1 expression in cultured cerebral endothelial cells but did not affect at all ICAM1 expression (Figure 3C).

Figure 3.

PTX induces few transcriptional changes in cultured cerebral endothelial cells. A. Venn diagram showing the number of genes identified by microarray analysis as being modulated in bEnd.3 cerebral endothelial cells incubated for 6 h in the presence of PTX (100 ng/mL) or LPS (10 ng/mL) as compared with cells treated with PBS. B. Hybridization signals of the 14 genes modulated by both PTX and LPS. The color scale indicates the intensity of the signals in arbitrary unit. The dendrograms indicate the degree of similarity among the genes or samples. They were produced by hierarchical clustering using Euclidean distance as the distance metric. C. Quantification of VCAM1 and ICAM1 mRNAs in bEnd.3 cells by qPCR. *Significantly different from PBS according to post hoc Student's t‐tests. Two‐way ANOVA: time effect, P < 0.0001; treatment effect, P < 0.0001; interaction, P < 0.0001.

To confirm the latter observations in vivo, we performed in situ hybridization for VCAM1 and ICAM1 mRNAs on brain sections from mice treated for 6 h with PTX (20 µg/kg) or LPS (2 µg/kg), thus respecting the concentration ratio tested in vitro (ie, 10:1). As a positive control, we examined the brains of mice injected with a high dose of LPS (1 mg/kg). Surprisingly, PTX strongly induced the expression of VCAM1 and ICAM1 mRNAs in blood vessels, whereas LPS induced their expression only when administered at a high concentration (Figure 4A). Similar observations were made using adjacent sections stained with antibodies against VCAM1 or ICAM1 (Figure 4B). However, an interesting difference in the distribution of these proteins was noted: VCAM1 was restricted to large blood vessels, whereas ICAM1 was also present on capillaries (Figure 4B). In summary, the differences between our in vitro and in vivo results suggest that PTX promotes vascular changes necessary for leukocyte adhesion by acting indirectly via a mechanism distinct from that of LPS. Our observations also suggest that ICAM1 rather than VCAM1 is involved in the recruitment of phagocytes in the cerebral microvasculature.

Figure 4.

PTX induces VCAM1 and ICAM1 expression in cerebral blood vessels. A. Dark‐field micrographs showing in situ hybridization signals for VCAM1 and ICAM1 mRNAs in the brains of mice killed 6 h after intraperitoneal injection of PBS, LPS or PTX. B. Bright‐field micrographs showing immunohistochemical staining for VCAM1 or ICAM1 in adjacent brain sections.

IL‐6 mediates the effect of PTX on ICAM1 expression and phagocyte recruitment

To identify a mediator through which PTX might act on the vasculature, we measured, in serum of mice exposed to PTX or LPS, the levels of cytokines classically known as regulators of ICAM1 expression in the brain, namely TNF, IL1β and IFNγ(18). We also measured the levels of IL6, whose effect on ICAM1 expression has been studied in vitro 22, 30 but not in the brain. As is well known, all these cytokines were increased in response to a high dose of LPS (Figure 5). In contrast, only IL6 was significantly increased in response to PTX (Figure 5A).

Figure 5.

PTX induces the expression of IL6 but not of IL1β, TNF and IFNγ. A–D. Quantification of each cytokine by ELISA in serum collected from mice 6 h after intraperitoneal injection of PBS, LPS or PTX. *Significantly different from PBS according to post hoc Bonferroni tests. Kruskal–Wallis test: P < 0.0001. Sample size: eight per group.

To determine whether IL6 mediates the effects of PTX, we first counted the number of crawling leukocytes in the cerebral cortex of IL6‐knockout and wild‐type mice injected with PTX or LPS as a control. As shown in Figure 6A, a normal increase in rod‐shaped leukocytes was observed in the knockouts treated with LPS, whereas no increase was observed in those treated with PTX. This result was not caused by an intergroup difference in the number of circulating myeloid cells, as confirmed by flow cytometry (GR1⁺ blood cells: IL6‐knockout mice, 22 ± 2%; wild‐type mice, 19 ± 2%; Student's t‐test, P = 0.22). Next, we examined ICAM1 expression at the mRNA and protein levels by in situ hybridization and immunohistochemistry (Figure 6B–C). Quantitative analysis revealed no difference in the mRNA levels of ICAM1 between the knockout and wild‐type mice after treatment with PTX or LPS (Figure 6D). However, the protein levels of ICAM1 were not increased in the knockouts treated with PTX, whereas they were normally increased in those treated with LPS (Figure 6E). Overall, these results indicate that IL6 is necessary for the recruitment of rod‐shaped phagocytes in response to PTX but is dispensable in the context of LPS. They also suggest that IL6 produces this effect, at least in part, by increasing ICAM1 expression at the post‐transcriptional level.

Figure 6.

IL6 mediates the effects of PTX on the recruitment of crawling leukocytes and the post‐transcriptional expression of ICAM1 in the brain vasculature. A. Counts of CD45⁺ crawling leukocytes in the cerebral cortex of IL6‐knockout (KO) and wild‐type (WT) mice euthanized 6 h after intraperitoneal injection of PBS, LPS or PTX. *Significantly different from the wild types according to post hoc Student's t‐tests. Two‐way ANOVA: genotype effect, P = 0.29 (rod cells) or 0.34 (round cells); treatment effect, P < 0.0001 (rod cells) or = 0.0004 (round cells); interaction, P = 0.0083 (rod cells) or 0.66 (round cells). Sample size: seven to nine per group. B. Dark‐field micrographs of in situ hybridization signals for ICAM1 mRNA in brain sections from IL6‐deficient and wild‐type mice treated with PTX. C. Bright‐field micrographs of brain sections derived from IL6‐deficient and wild‐type mice treated with PTX and stained for ICAM1 by immunohistochemistry. D. Quantification of the hybridization signals in the cerebral cortex by optical density (OD) readings. ANOVA: genotype effect, P = 0.42; treatment effect, P < 0.0001; interaction, P = 0.64. Sample size: six to eight per group. E. Quantification of the volume occupied by ICAM1⁺ blood vessels in the cerebral cortex. *Significantly different from PBS according to post hoc Student's t‐tests. Two‐way ANOVA: genotype effect, P < 0.0001; treatment effect, P < 0.0001; interaction, P ≤ 0.0001. Sample size: eight or nine per group.

Phagocytes crawl on the cerebral endothelium using ITGαM and ICAM1

It is known that leukocytes can bind to ICAM1 through the integrin dimers αMβ2 (Mac1) or αLβ2 (LFA1) and to VCAM1 through α4β7 (LPAM1) or α4β1 (VLA4). To examine the importance of such interactions in the recruitment of crawling phagocytes in response to PTX, we counted these cells in the cerebral cortex of mice intravenously injected with neutralizing antibodies directed against each of the abovementioned integrin monomers. We also counted these cells in ICAM1‐knockout mice, which are viable, as opposed to VCAM1‐knockout mice, and which had normal counts of circulating myeloid cells, as confirmed by flow cytometry (%GR1⁺ blood cells: ICAM1‐knockout mice, 22 ± 3%; wild‐type mice, 23 ± 4%; Student's t‐test, P = 0.75). The results revealed a complete blockage of the recruitment of rod‐shaped leukocytes in response to PTX after treatment with an anti‐ITGαM antibody but not with any of the other antibodies (Figure 7A–B). Consistently, a partially attenuated recruitment of rod‐shaped leukocytes was observed in ICAM1‐knockout mice exposed to PTX (Figure 7C). Altogether, these results suggest that phagocytes require ITGαM to adhere to the cerebral endothelium through ICAM1 and probably also another ligand that remains to be identified.

Figure 7.

Crawling leukocytes depend on ITGαM and ICAM1 for adhesion to the cerebral vasculature in response to PTX. A, B. Counts of CD45⁺ crawling leukocytes in the cerebral cortex of mice treated with different neutralizing antibodies against integrins and euthanized 6 h after intraperitoneal injection of PBS or PTX (20 µg/kg). *Significantly different from the isotype controls according to post hoc Student's t‐tests. ANOVA for A: P < 0.0001 (rod cells) or = 0.096 (round cells). ANOVA for B. P = 0.36 (rod cells) or 0.75 (round cells). Sample size: 5–10 per group. C. Counts of CD45⁺ crawling leukocytes in the cerebral cortex of ICAM1‐knockout (KO) and wild‐type (WT) mice euthanized 6 h after injection of PBS or PTX. *Significantly different from the wild types according to post hoc Student's t‐tests. Two‐way ANOVA: genotype effect, P = 0.05 (rod cells) or 0.035 (round cells); treatment effect, P = 0.005 (rod cells) or 0.10 (round cells); interaction, P = 0.05 (rod cells) or 0.32 (round cells). Sample size: 5–10 per group.

DISCUSSION

How phagocytes are attracted to the brain and how PTX promotes EAE are two important questions that confront neuroimmunologists. As illustrated in Figure 8, the present study helps to answer these questions by showing that PTX promotes vascular changes required for leukocyte adhesion. In contrast to LPS, PTX exerts few direct effects on endothelial cells and must act indirectly through mediators distinct from TNF, IL1β and IFNγ, which are classical regulators of endothelial adhesiveness. Instead, PTX acts through IL6, which contributes to stimulate ICAM1 expression on the capillary surface. In response, appropriately activated phagocytes, mainly granulocytes, adhere to the endothelium, at least in part, through interaction involving ITGαM and ICAM1. In summary and as discussed below, this study (i) identifies the existence of an alternative, non‐classical mechanism by which phagocytes are recruited at the blood‐brain interface; (ii) it demonstrates the importance of a ligand‐receptor couple in this process; and (iii) it suggests that PTX might promote EAE by increasing the patrolling of the CNS vasculature by immune cells capable of crawling on its luminal surface.

Figure 8.

A proposed mechanism by which phagocytes are recruited at the blood‐brain interface in response to PTX. (1) In peripheral tissues, PTX stimulates cells to secrete IL6. Ostensibly, another mediator, different from TNF, IL1β and IFNγ, is also secreted and remains to be identified. (2) IL6 circulates in the blood and contributes to stimulate ICAM1 expression on the endothelial surface by acting at the post‐transcriptional level. The as yet unknown mediator is responsible for triggering ICAM1 gene transcription. (3) IL6 might also induce integrin activation on phagocytes, as suggested by a recent study (7). (4) Activated phagocytes, mainly granulocytes, adhere to cerebral capillaries by binding to ICAM1 and an as yet unknown ligand through integrin αMβ2.

PTX induces vascular changes required for leukocyte adhesion

It was long believed that PTX promotes EAE by increasing the permeability of the blood–brain barrier 19, 20. However, a few years ago, it was discovered that this effect is secondary, and that PTX rather acts by increasing the adhesiveness of the cerebral endothelium by inducing P‐selectin expression (17). The underlying mechanism was not fully elucidated but was found to be dependent on TLR4. Our results extend these observations by showing that PTX induces the expression of other adhesion molecules, namely ICAM1 and VCAM1. It will be important in future studies to consider the following two hypotheses. First, it is plausible that PTX acts through a receptor different from that of LPS and absent in endothelial cells. This is suggested by our observation that PTX exerts few direct effects on endothelial cells, while the latter were responsive to LPS and are known to express TLR4 (9). Second, it is possible that the relative importance of the adhesion molecules involved in phagocyte recruitment varies according to vessel caliber. On the one hand, we did not count leukocytes in large blood vessels, where they can be confounded with perivascular macrophages caused by their amoeboid morphology. On the other hand, Kerfoot et al could not examine leukocytes in capillaries, because they are too deeply dispersed in the nervous tissue to be visualized by real‐time microscopy through an intracranial window. Therefore, it is not possible to definitively conclude that P‐selectin has the same importance in capillaries as in larger vessels (eg, arterioles) and vice versa for ICAM1.

IL6 contributes to the effects of PTX

Our observation that PTX did not affect ICAM1 expression in cultured cerebral endothelial cells but strongly induced its expression in vivo lead us to examine the possibility that PTX acts indirectly via soluble mediators. ICAM1 is classically regulated by IFNγ and factors that stimulates the nuclear factor‐kappa B (NF‐κB) signaling pathway, such as TNF and IL1β(18). The present study shows that these three mediators are dispensable, and that IL6 is absolutely required for mediating the effect of PTX. The ability of IL6 to induce ICAM1 expression has been demonstrated in vitro 22, 30 but not in the brain. Interestingly, the fact that ICAM1 expression was altered in IL6‐knockout mice only at the protein level and not at the mRNA level suggests that IL6 stimulates the post‐transcriptional machinery that governs ICAM1 production, while another soluble mediator is responsible for stimulating the transcriptional machinery. Further work is needed to identify this mediator and demonstrate how it cooperates with IL6 to increase endothelial adhesiveness.

Adhesion molecules used by crawling phagocytes

ICAM1 and VCAM1 are well‐known adhesion molecules, but their relative importance in the recruitment of phagocytes in the brain was unclear. Our observation that VCAM1 was only expressed on large blood vessels during endotoxemia, whereas ICAM1 was also expressed on capillaries, led us to believe that ICAM1, but not VCAM1, is involved in the recruitment of crawling phagocytes. This was confirmed in PTX‐treated mice deficient in ICAM1 or injected with an antibody against ITGαM, which is an integrin that forms with ITGβ2 a receptor for ICAM1 called Mac1. No role was found for ITGαL, which forms with ITGβ2 another receptor for ICAM1 called LFA1. This negative result was not likely caused by an inability of our antibody to neutralize ITGαL, because this antibody was proved to be effective in another study in which it was shown to detach type 2 (GR1^‐) monocytes that crawl in post‐capillary venules of the ear in steady‐state conditions (3). This difference between our results and those of Auffray et al could arise from the types of leukocytes and vessels that were studied (ie, inflammatory vs. resident leukocytes and cerebral capillaries vs. large blood vessels of peripheral tissues). Nevertheless, ICAM1 is ostensibly not the only ligand used by crawling phagocytes in the brain, as suggested by the fact that their increase was totally blocked in mice treated with the anti‐ITGαM antibody, but only partially blocked in mice lacking ICAM1. Further work will be required to test whether Mac1 interacts with an alternative ligand exposed on the surface of cerebral capillaries.

Recruitment of crawling phagocytes via two different mechanisms

We have previously shown that Angpt2 is needed for the recruitment of crawling leukocytes in response to LPS (2). Angpt2 is believed to be a gatekeeper of endothelial activation. It sensitizes the endothelium toward TNF, thereby enhancing the expression of ICAM1 and VCAM1 (11). This effect might be attributable to the ability of Angpt2 to block the interaction between Tie2 and ABIN‐2, an inhibitor of the NF‐κB signaling pathway (15). Considering that TNF, an important regulator of Angpt2 expression (2), was not produced in PTX‐treated mice, it is not so surprising that Angpt2 was also not produced in these animals. There are thus at least two different mechanisms that induce phagocyte adhesion in response to different stimuli: one dependent and the other independent of the TNF‐Angpt2 axis. It will be interesting in the future to examine why different mechanisms are required, whether the leukocytes recruited by these mechanisms exhibit phenotypic differences and whether there is an alternative gatekeeper to Angpt2 that sensitizes the endothelium toward IL6.

Reinforced patrolling of the vasculature in response to alarm signals

It is interesting to speculate that PTX might promote EAE, at least in part, by increasing the patrolling of the CNS vasculature by leukocytes able to crawl on the endothelium. These leukocytes could be not only innate immune cells, but also T lymphocytes, as suggested by our finding of CD3⁺ cells with a rod‐shaped morphology in the brains of EAE mice. The ability of T cells to crawl in the cerebral vasculature of EAE mice and to interact with phagocytes has been observed in another study using intravital microscopy (4). It therefore seems plausible that microbes and other toxins can produce a similar effect in MS patients, thereby affecting disease severity and progression. It would be exciting to examine in the future whether an anti‐ITGαM antibody, or any other drugs designed to block phagocyte crawling, could be used to attenuate MS, as it was the case for EAE (14).

CONCLUSION

Neuroimmunologists have traditionally used LPS as a simple model to study the development of immune responses in the nervous system with the hope of gaining insight into neuroinflammatory diseases, such as EAE and MS. The present study warrants that LPS may not be appropriate for such a purpose, as it induces leukocyte recruitment via an Angpt2‐dependent mechanism that is not involved in EAE. Future studies on the alternative mechanism triggered by PTX and uncovered in this study may allow for the identification of new potential targets for the treatment of MS.

Supporting information

Figure S1. Micrograph of a rod‐shaped leukocyte immunostained with an anti‐CD3 antibody.

Table S1. Genes differentially regulated in mouse bEnd.3 cerebral endothelial cells in response to PTX and/or LPS, as determined using Affymetrix DNA microarrays.

Supporting info item

Supporting info item