A role for leukocyte-endothelial adhesion mechanisms in epilepsy

Experimental data from animal models as well as human evidence indicate that seizures can lead to neuronal damage and cognitive impairement^{2, 3}. However, the molecular mechanisms leading to seizures and epilepsy are not well understood. Recent data suggests that inflammation may play a role in the pathogenesis of epilepsy^{4, 5}. For instance, elevation in inflammatory cytokines are seen in the central nervous system (CNS) and plasma in experimental models of seizures and in clinical cases of epilepsy^{4, 5}. Moreover, CNS inflammation is associated with breakdown in the blood-brain barrier (BBB), and BBB leakage has been implicated both in the induction of seizures and in the progression to epilepsy^6–9. Leukocyte recruitment is a hallmark of and a point of therapeutic intervention in tissue inflammation^10,11, but a role for leukocyte-endothelial interactions in seizure pathogenesis has not been explored. Here we have employed an established mouse model of epilepsy to ask whether leukocyte-endothelial interactions can also contribute to seizure pathogenesis.

Systemic administration of pilocarpine induces status epilepticus (SE) in mice typically lasting a few hours^{12, 13}. After a latent (seizure-free) phase, mice that experienced SE go on to develop epilepsy characterized by spontaneous recurrent seizures (SRS)^{12, 13}. In addition to hippocampal damage, pilocarpine-induced seizures in rodents also cause widespread cortical and subcortical lesions¹⁴.

Seizure activity in this model led to induction of vascular adhesion molecules for leukocytes. Vessels in control brains expressed low amounts of ICAM-1 (intercellular adhesion molecule-1), but VCAM-1 (vascular cell adhesion molecule-1), E-selectin and P-selectin were below the limits of detectability (Fig. 1a). As shown in Figure 1a, b ICAM-1 and VCAM-1 were dramatically upregulated following seizure induction, with the highest levels observed 24 h and 7 d post-SE (Fig 1a, b, Supplementary Table 1a). Pharmacologic suppression of SE with Diazepam, a benzodiazepine with CNS depressant properties, prior to pilocarpine injection substantially but not completely prevented the upregulation of VCAM-1 (Fig. 1b), suggesting that electrical hyperactivity contributes to high VCAM-1 expression in brain vessels potentially in combination with a direct vascular effect of pilocarpine itself⁹. P-selectin and E-selectin were also induced on venular endothelium after seizures (Fig. 1a, b and Supplementary Table 1a). Thus elicited seizure activity is associated with inflammatory changes in the CNS vasculature, changes that could support enhanced adhesion of circulating leukocytes and consequent amplification of vascular damage.

Direct pilocarpine treatment of leukocytes does not render them hyperadhesive (Supplementary Fig. 1a), but pilocarpine induction of seizures dramatically enhanced leukocyte adhesion to CNS vessels in vivo (Fig. 1c, d). Intravital microscopy studies performed in cortical venules¹⁵ showed that PMN (polymorphonuclear leukocytes) displayed efficient rolling and arrest after SE onset (Fig. 1c, d). Resting lymphocytes did not interact, (Fig. 1c), while T_H (T helper)1, polarized cells rolled and stuck efficiently after seizures (Fig. 1c, d). PMN arrest was consistently higher at 6 h, while T_H1 cells interacted more efficiently at 24 h than at 6 h after seizures, consistent with progression from acute to subacute inflammation during the 24 h period after SE-induced damage (Fig. 1c). Rolling velocities (Vroll) and the frequency distribution of Vroll in velocity classes are provided at Supplementary Fig. 1b and Supplementary Table 1. Interestingly, in vitro polarized T_H2 cells, known to lack functional P-selectin ligand, did not roll or stick in brain vessels (Fig. 1c).

Rolling interactions and arrest of PMN cells after SE were dramatically inhibited by antibody blockade of α4 integrin or its vascular counter ligand VCAM-1, or by blockade of P-selectin glycoprotein ligand-1 (PSGL-1) or its principal vascular ligand P-selectin (Fig. 1e; Supplementary Fig. 1c and d, Supplementary Discussion [1]). Rolling and arrest of T_H1 cells were also reduced by α4 or VCAM-1 blockade after SE (Fig. 1f), and were almost completely blocked by antibodies to PSGL-1 and P-selectin, as well (Fig. 1f and data not shown). MAb (monoclonal antibody) to MAdCAM-1 (mucosal addressin cell adhesion molecule-1) had no significant effect, excluding a role for the intestinal trafficking receptor α4β7 integrin (Fig. 1e, f and Supplementary Figs. 1c). These results demonstrate that the integrin α4β1 and the selectin ligand PSGL-1 mediate vascular-leukocyte adhesive interactions in cerebral vessels following SE. We therefore set out to ask whether these leukocyte adhesion mechanisms might contribute to seizure induction and/or epileptogenesis in this model.

To mimic therapeutic intervention in response to severe acute seizures, we initially evaluated the effect of adhesion blockade after SE on the subsequent development of recurrent seizures. One hour of SE is sufficient to lead to chronic spontaneous seizures in the pilocarpine model in mouse and rat^{13, 16, 17} (supplementary Discussion [2]). Therefore, pilocarpine-injected C57Bl/6 mice were allowed to convulse for one hour (stage 5 of Racine’s Scale¹⁸), and then were treated with α4 integrin-specific mAb i.p (Fig. 2). Mice also received α4 integrin-specific mAb every other day for 20 days. As expected, treatment with antibody to α4 integrin after SE onset had no effect on the character and duration of ongoing SE as assessed visually or by EEG analysis after pilocarpine treatment (Fig. 2g, Supplementary Fig. 2b and data not shown). However, video monitoring revealed a dramatic reduction of spontaneous convulsions during the chronic period (from day 5 to 20 after pilocarpine injection) in mice treated with α4 integrin-specific antibody (Fig. 2a). EEG telemetry confirmed a consistent reduction in frequency of spontaneous recurrent seizures (SRS) during the chronic phase in the treated mice (Fig. 2e). The latency of the first SRS, and the duration of individual seizure events when they occurred, was similar in the treatment and control groups (Fig. 2d, f, h). Neuropathological findings showed post-SE alterations including increased ventricle volume and hippocampal shrinkage, but there was a statistically significant decrease in neuronal cell loss in somatosensory cortex and hippocampus subfields (Figure 3). In addition, 30 days after SE, animals treated with antibody to α4 integrin exhibited a slight reduction of exploration behavior compared to normal animals, but a significant preservation of the behavior compared to epileptic mice (Supplementary Fig. 2 and Supplementary Table 2a). Post-SE antibody against the α4β1 ligand VCAM-1 was also effective in preventing spontaneous convulsions during the chronic phase (Fig. 2b). Thus our data suggest that α4 and its ligand VCAM-1 contribute to pathogenic events required for epileptogenesis in this model.

Surprisingly, when antibody to α4 integrin was given 2 h before injection of pilocarpine, it completely prevented convulsions (Fig. 2a). In the first hour after pilocarpine injection, anti-α4 integrin pretreated mice displayed sporadic tremors and oral mastication (stage 2 of the Racine’s Scale¹⁸), but no further tremors or SE developed, and no delayed spontaneous convulsions were detected from day 5 through day 20 (mice also received 200 µg mAb to α4 integrin every other day for 20 days, to mimic a therapeutic regimen) (Figure 2a). Moreover, pre-treated mice displayed no EEG alterations indicative of SE or SRS following pilocarpine-injection (Figure 2d–h, Supplementary Fig. 2a). Predictably, initiation of anti-α4 integrin treatment prior to pilocarpine led to a significant decrease of neuronal cell loss in somatosensory cortex and hippocampus and prevented delayed cognitive impairment (Figure 3, Supplementary Fig. 2e, Supplementary Table 2a). In support of these results, blockade of αL integrin, or of its endothelial ligand ICAM-1 consistently reduced leukocyte-endothelial interactions and prevented convulsions (Supplementary Figure 1d, e). In contrast, control anti-CD45 mAb or anti-α4β7 or anti-CD4 mAbs had no effect on SE induction, chronic disease, or cognitive deterioration (Figure 2a, Supplementary Fig. 2e and Supplementary Table 2b for anti-CD45 mAb and data not shown).

Intravenously administered antibody should be largely excluded from the CNS by the BBB prior to SE, but we could not formally exclude an effect of leaked anti-α4 or anti-VCAM-1 mAbs on nervous tissue. To address this issue as well as to assess the role of selectin-mediated interactions, we took advantage of genetically modified mice deficient in PSGL-1 (Psgl-1^−/− mice) or in α-1,3-fucosyltransferases (FucTs), FucT-VII and FucT-IV, enzymes required to generate functional selectin-binding carbohydrates (Fut4^−/− and Fut7^−/− mice)¹⁹. Consistent with the inhibitory effects of anti-α4 treatment we observed a dramatic reduction in the incidence of SE and of subsequent spontaneous convulsions and of SRS as monitored visually or electroencephalographically in Psgl-1^−/− and Fut^−/− mice (Figure 2c and Fig 2d–h, and Supplementary Figure 2c, d, Supplementary movie). Psgl-1^−/− and Fut7^−/− animals also showed a significant reduction of neuronal damage and a considerable preservation of the behavior compared to epileptic mice (Figure 3, Supplementary Fig. 2e, Supplementary Table 2c). Since PSGL-1 is selectively expressed on leukocytes and endothelium²⁰, inhibition of seizures in Psgl-1^−/− mice further supports a direct or indirect role for leukocyte adhesion in seizure generation in this model, and rules out primary neuronal or glial effects of anti-adhesion treatments. To further confirm the role of leukocytes, we used a granulocyte-specific antibody to deplete neutrophils before pilocarpine administration. Neutrophil depletion caused a significant reduction of SE and spontaneous recurrent convulsions (Supplementary Fig. 2f), confirming that leukocytes contribute to seizure pathogenesis.

Recent studies have shown that BBB breakdown induces epileptiform activity^{6–9, 21}. Moreover, pilocarpine induced seizure activity is not only associated with BBB leakage in vivo, but may in fact require leakage of serum components: pilocarpine induces neuronal hyperactivity in ex vivo brain slices, but only in the context of elevated K⁺ levels mimicking plasma leakage into the CNS⁹. BBB leakage could also enhance access of pilocarpine itself to the extravascular compartment, although this is unlikely to be sufficient for epileptogenesis in the absence of plasma leakage⁹ (Supplementary discussion [3]). We therefore evaluated BBB leakage in our model. Pilocarpine-induced BBB opening was largely prevented in mice with leukocyte adhesion blockade, whether by antibody or by genetic interference with leukocyte adhesion pathways (Fig. 4a, b).

Taken together, the demonstrated elicitation of neuronal hyperactivity by BBB leakage^{6–9, 21} and the inhibition of BBB breakdown by blockade of leukocyte-endothelial interaction, shown here, suggest a mechanism for synergistic and potentially self-reinforcing interactions between endothelial activation, leukocyte adhesion, BBB opening and seizure generation. A cycle of seizure-induced inflammation and inflammation-mediated stimulation may amplify the initial effects, leading to chronic changes and recurrent seizure activity (Supplementary Fig. 3a, Supplementary Discussion [4]). The general relevance of our findings remains to be determined in other models of epilepsy.

We evaluated human cortical CNS tissue from subjects with epilepsy and found elevated brain leukocyte numbers compared to non-epileptic controls, suggesting a possible role for leukocyte adhesion in humans as well (Supplementary Figure 4 and Supplementary Table 3). Current pharmacological treatments for epilepsy inhibit neuronal excitability but do not address the inflammatory component of pathogenesis. Anti-adhesion therapies based on humanized antibodies^{10, 22,23} or other available approaches would allow clinical evaluation of adhesion pathways as targets for preventing epilepsy, particularly following inflammatory inciting events such as stroke or traumatic brain injury.

In conclusion, our results suggest that vascular inflammatory mechanisms and leukocyte-endothelial adhesion can contribute to the pathogenesis of seizures and epilepsy, and show that inhibition of leukocyte-vascular interactions can have preventive as well as therapeutic effects in a mouse model of this debilitating disease.