Experimental Cerebral Malaria Develops Independently of Endothelial Expression of Intercellular Adhesion Molecule-1 (ICAM-1)*

Background: Endothelium-expressed intercellular adhesion molecule-1 (ICAM-1) is considered critical for the development of cerebral malaria (CM). Results: ICAM-1 expression on leukocytes alone was sufficient for the development of experimental CM. Conclusion: Endothelial expression of ICAM-1 is not required for development of CM. Significance: Vascular occlusion in CM requires ICAM-1 expression on leukocytes but not endothelial cells. Cerebral malaria (CM) is a severe clinical complication of Plasmodium falciparum malaria infection and is characterized by a high fatality rate and neurological damage. Sequestration of parasite-infected red blood cells in brain microvasculature utilizes host- and parasite-derived adhesion molecules and is an important factor in the development of CM. ICAM-1, an alternatively spliced adhesion molecule, is believed to be critical on endothelial cells for infected red blood cell sequestration in CM. Using ICAM-1 mutant mice, we found that the full-length ICAM-1 isoform is not required for development of murine experimental CM (ECM) and that ECM phenotype varies with the combination of ICAM-1 isoforms expressed. Furthermore, we observed development of ECM in transgenic mice expressing ICAM-1 only on leukocytes, indicating that endothelial cell expression of this adhesion molecule is not required for disease pathogenesis. We propose that ICAM-1-dependent cellular aggregation, independent of ICAM-1 expression on the cerebral microvasculature, contributes to ECM.


Cerebral malaria (CM) is a severe clinical complication of
Plasmodium falciparum malaria infection and is characterized by a high fatality rate and neurological damage. Sequestration of parasite-infected red blood cells in brain microvasculature utilizes host-and parasite-derived adhesion molecules and is an important factor in the development of CM. ICAM-1, an alternatively spliced adhesion molecule, is believed to be critical on endothelial cells for infected red blood cell sequestration in CM. Using ICAM-1 mutant mice, we found that the full-length ICAM-1 isoform is not required for development of murine experimental CM (ECM) and that ECM phenotype varies with the combination of ICAM-1 isoforms expressed. Furthermore, we observed development of ECM in transgenic mice expressing ICAM-1 only on leukocytes, indicating that endothelial cell expression of this adhesion molecule is not required for disease pathogenesis. We propose that ICAM-1dependent cellular aggregation, independent of ICAM-1 expression on the cerebral microvasculature, contributes to ECM.
Cerebral malaria (CM) 2 is thought to arise from a confluence of inflammatory events in which infected RBCs (iRBC), activated leukocytes, and platelets are sequestered on inflamed endothelium due to increased expression of adhesion molecules (1). Among the large number of adhesion molecules implicated in CM development, ICAM-1 has long been known to bind and retain iRBCs in the central nervous system (CNS) microvasculature (2)(3)(4). ICAM-1 binds to Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) through its N-terminal Ig domain (5)(6)(7) and to several members of the ␤ 2 -integrin family of adhesion molecules including LFA-1, Mac-1, and p150,95 (8). ICAM-1 is expressed on essentially all cell types contributing to CM development including lymphocytes, myeloid cells, platelets, and endothelial cells (9) (8,10,11). The increased expression of ICAM-1 and other adhesion molecule receptor/ligand pairs within the microvasculature sets the stage for vessel occlusion under inflammatory conditions driven by proinflammatory cytokines produced during malaria infection (12,13).
The importance of ICAM-1 in CM is based on studies demonstrating the relationship between increased endothelial cell ICAM-1 expression and iRBC sequestration in human and murine CM (1,14). In experimental cerebral malaria (ECM), the animal model of CM, treatment with anti-ICAM-1 antibodies markedly inhibited rolling and iRBC sequestration in the central nervous system (3,4,15). Anti-ICAM-1 antibodies also inhibited adherence and rolling of iRBCs on LPS-primed brain sections or ICAM-1-transfected cells (3). Furthermore, there are reports that ICAM-1 polymorphisms are associated with severe forms of malaria, particularly CM, although this remains controversial (16 -18). Although these data indicate a critical role for ICAM-1 in iRBC binding to endothelium, no study has directly addressed the requirement of ICAM-1 in the development of ECM.
Multiple isoforms of ICAM-1, arising from alternative splicing, have been described in humans and mice (19 -24). However, their function, changes in expression, and relative expression on various cell types involved in the development of ECM remain unknown. We show here that ICAM-1 mutant mice deficient in all ICAM-1 isoforms (Icam1 null ) are highly resistant to the development of ECM. In contrast, mice expressing different combinations of three of the six known isoforms, but not the full-length molecule (Icam1 tm1Jcgr and Icam1 tm1Bay mice), are more susceptible to ECM. These data demonstrate that although the full-length ICAM-1 protein is not required for disease initiation, ECM is attenuated in its absence. We also report the unexpected finding that ICAM-1 expression on CNS microvasculature is not required for ECM development. For these studies, Icam1 null mice were bred with newly developed transgenic * This work was supported, in whole or in part, by National Institutes of Health Grants T32 AI07051 (to T. N. R.), F31NS077811 (to T. N. R.), and P30 CA13148 and P30 AR048311. This work was also supported by National Multiple Sclerosis Society Grant PP1058 (to S. R. B. and D. C. B.). 1  mice that express the full-length ICAM-1 isoform under the control of the leukocyte-specific promoter CD2. We observed that these transgenic mice, unlike Icam1 null mice, were highly susceptible to ECM. These observations suggest that ICAM-1-mediated aggregation of leukocytes, platelets, and iRBC within the vascular space, independent of expression on the microvasculature, is critical for promoting vessel occlusion during the development of ECM.
Flow Cytometry-CD2-restricted expression of ICAM-1 on leukocytes was determined by flow cytometry as described previously (32). Inbred C57BL/6 mice were used as controls for all experiments. T cell infiltration into brains at day 6 after infection was assessed by flow cytometry as described previously (33).
Statistical Analysis-Statistical significance of ECM survival was calculated using the log rank test using Prism 5 (GraphPad Software, Inc.). The Mann-Whitney test was used to determine significant differences in ECM clinical scores. Data are shown as mean Ϯ S.E. A value of p Ͻ 0.05 was considered statistically significant.

RESULTS AND DISCUSSION
Icam1 null Are Highly Resistant to ECM, whereas Icam1 tm1Jcgr and Icam1 tm1Bay Are Partially Resistant-Although changes in ICAM-1 expression are well documented in both severe and cerebral malaria (1,14), no study has directly assessed ECM severity and development using mice deficient in all ICAM-1 isoforms. For these studies, we used three different lines of ICAM-1 mutant mice, including Icam1 null mice, which lack all known ICAM-1 isoforms (Fig. 1A). We observed that Icam1 null mice were highly resistant to ECM, with Ͼ90% of mice surviving past day 10, whereas all wild type mice succumbed to dis- Wild type, Icam1 tm1Jcgr , and Icam1 tm1Bay mice were injected intraperitoneally with 5 ϫ 10 5 PbA-iRBC, and clinical scores and survival were monitored twice daily for 10 days as described in Ref. 37. A, schematic of ICAM-1 with isoforms missing in Icam1 null mice shown in light gray. B, Icam1 null mice (n ϭ 15) were significantly resistant to disease-induced mortality (p ϭ 0.0001, log rank test; 93% survival past day 10) as compared with wild type mice (n ϭ 12). C, Icam1 null mice had significantly reduced clinical signs of disease (p Ͻ 0.0001 from day 6 onward, Mann-Whitney test) as compared with wild type mice. D, schematic of ICAM-1 with isoforms expressed by Icam1 tm1Bay mice shown in black outline. E, Icam1 tm1Bay mice (n ϭ 15) were significantly resistant to disease-induced mortality (p ϭ 0.0001, log rank test; 60% survival past day 10) as compared with wild type mice (n ϭ 16). F, Icam1 tm1Bay mice had significantly reduced clinical signs of disease (p Ͻ 0.0001 from day 6 onward, Mann-Whitney test) as compared with wild type mice. G, schematic of ICAM-1 with isoforms expressed by Icam1 tm1Jcgr mice shown in black outline. H, Icam1 tm1Jcgr mice (n ϭ 17) were significantly resistant to disease-induced mortality (p ϭ 0.0001, log rank test; 41% survival past day 10) as compared with wild type mice (n ϭ 17). I, Icam1 tm1Jcgr mice had significantly reduced clinical signs of disease (p Ͻ 0.0001 from day 6 onward, Mann-Whitney test) as compared with wild type mice. Shown is the mean Ϯ S.E. of 3-4 independent experiments for all groups of mice. ease on or before day 9 (Fig. 1B, Table 1, p ϭ 0.0001, Log rank test). Icam1 null mice had a corresponding reduction in clinical scores (Fig. 1C, Table 1, cumulative disease index: 1.4 versus 18.3, p Ͻ 0.0001, Mann-Whitney test) and a marked reduction in CNS T cell infiltration (CD4 ϩ , 65% decrease; CD8 ϩ , 84% decrease) as compared with wild type mice. In contrast, two ICAM-1 mutant strains of mice that express only three of the six known ICAM-1 isoforms, but not the full-length isoform (Icam1 tm1Bay and Icam1 tm1Jcgr , Fig. 1, D and G, respectively), develop more severe ECM than Icam1 null mice. Icam1 tm1Bay mice were previously shown to develop attenuated ECM as compared with wild type mice (34), a finding we replicate in the present study (Fig. 1, E and F, Table 1). Icam1 tm1Bay mice, however, have lower survival and worse clinical scores as compared with Icam1 null mice (60% versus 93% survival, cumulative disease index 5.5 versus 1.4, respectively). T cell infiltration into the CNS was also reduced in these mice as compared with wild type mice (CD4 ϩ , 31% decrease; CD8 ϩ , 84% decrease). In contrast, Icam1 tm1Jcgr mice were most susceptible to ECM of the three ICAM-1 mutant mice, with the lowest survival rate and highest clinical scores (Fig. 1, H and I, Table 1). The extent of CNS T cell infiltration in Icam1 tm1Jcgr mice corresponded to increased disease severity with a modest reduction in CD4 ϩ T cells (30% decease), but an increase in CD8 ϩ T cells (12% increase) as compared with wild type mice.
ICAM-1 Isoform Expression Solely on Leukocytes Is Sufficient for ECM Development-It has been assumed that ICAM-1 contributes to CM predominately, if not exclusively, via expression on endothelial cells based on its function in the classic rolling/firm adhesion/transmigration paradigm used by leukocytes for trafficking to sites of inflammation or into lymphoid tissues (35)(36)(37)(38). To directly determine whether ICAM-1 expression on the microvasculature is required for ECM development, we generated transgenic mice with an ICAM-1-deficient background that expressed only the full-length ICAM-1 isoform on leukocytes (CD2-Icam1 fl /Icam1 null ). Expression of the full-length isoform on CD4 ϩ and CD8 ϩ T cells, CD19 ϩ B cells, and natural killer cells from CD2-Icam1 fl /Icam1 null mice was comparable with that seen on wild type mice (Fig. 2, A-D). We then performed ECM using wild type and CD2-Icam1 fl / Icam1 null mice. We observed that CD2-Icam1 fl /Icam1 null mice were fully susceptible to ECM; however, they progressed to fatal ECM significantly more slowly than wild type mice (p Ͻ 0.0001, Log rank test, ϳ2-day delay) with a corresponding decline in clinical scores (Fig. 2, E and F, Table 1). These results demonstrate that expression of a single ICAM-1 isoform, in this case the full-length ICAM-1 isoform, on leukocytes alone is sufficient to drive development of ECM, independent of expression on endothelium.
Our results indicate that ICAM-1 is critical to the development of ECM because deletion of all isoforms essentially prevents disease development. However, the disease phenotypes of the Icam1 tm1Jcgr and Icam1 tm1Bay mice raise two important points. 1) The absence of the full-length ICAM-1 isoform does not prevent the development of ECM, and 2) variable combinations of ICAM-1 isoforms in the absence of the fulllength isoform result in distinct ECM phenotypes, indicating that multiple isoforms can contribute to disease outcome. This is remarkable because Icam1 tm1Jcgr and Icam1 tm1Bay mice share two of the three expressed isoforms (Fig. 1, D and  G). It is worth noting that in experimental autoimmune encephalomyelitis (EAE), the disease phenotype of Icam1 tm1Jcgr and Icam1 tm1Bay mice is reversed as compared with what we report here for ECM (39). Thus Icam1 tm1Bay mice developed severe EAE, whereas Icam1 tm1Jcgr mice developed mild EAE. This contrast in disease phenotypes suggests differential utilization and/or signaling mechanisms based on the combination of ICAM-1 isoforms expressed on leukocytes and endothelium.
Perhaps the most interesting finding in our study was the observation that leukocyte expression of a single ICAM-1 isoform, in an otherwise Icam1 null background, was sufficient for ECM development. These data, combined with the results of the Icam1 tm1Jcgr and Icam1 tm1Bay studies (Fig. 1), demonstrate that expression of different ICAM-1 isoforms results in dramatically different ECM phenotypes, suggesting distinct effector functions for a given isoform or combination of isoforms. These effector functions likely result from the integration of isoform-specific signaling pathways initiated by binding to single or multiple receptors. In the setting of ECM, we hypothesize that leukocyte/platelet/iRBC aggregates form in an ICAM-1-dependent fashion through use of ␤ 2 -integrins (LFA-1 and Mac-1), which are then further stabilized by fibrinogen receptor binding (gpIIa/IIIb; CD41/CD61) on platelets. Such aggregates could form under the inflammatory conditions characteristic of malaria and CM and potentially occlude microvessels irrespective of the adhesive state of the endothelium. The powerfully protective effect of anti-LFA-1 treatment and in LFA-1-deficient mice in ECM suggests that LFA-1 is an important ICAM-1 counter receptor in aggregate formation (40 -43). In addition, recent studies have suggested that blood brain barrier opening due to vascular leakage contributes significantly to the development of ECM (42). The data we report here suggest that both mechanisms may be simultaneously at work, with brain edema initiated and/or exacerbated by sporadic to widespread vessel occlusion. Taken together, our results indicate a more central role for ICAM-1 in CM than previously appreciated and suggest that ICAM-1-based therapeutics may be an effective treatment strategy in CM.