Neuronal Protection in Stroke by an sLex-Glycosylated Complement Inhibitory Protein

See allHide authors and affiliations

Science  23 Jul 1999:
Vol. 285, Issue 5427, pp. 595-599
DOI: 10.1126/science.285.5427.595


Glycoprotein adhesion receptors such as selectins contribute to tissue injury in stroke. Ischemic neurons strongly expressed C1q, which may target them for complement-mediated attack or C1qRp-mediated clearance. A hybrid molecule was used to simultaneously inhibit both complement activation and selectin-mediated adhesion. The extracellular domain of soluble complement receptor–1 (sCR1) was sialyl Lewis x glycosylated (sCR1sLex) to inhibit complement activation and endothelial-platelet-leukocyte interactions. sCR1 and sCR1sLexcolocalized to ischemic cerebral microvessels and C1q-expressing neurons, inhibited neutrophil and platelet accumulation, and reduced cerebral infarct volumes. Additional benefit was conferred by sialyl Lewis x glycosylation of the unmodified parent sCR1 molecule.

Interrupting blood flow to the brain, even for relatively short periods, can trigger inflammatory events within the cerebral microvasculature that can exacerbate cerebral tissue injury. The tissue damage that accrues is amplified by activation of both inflammatory and coagulation cascades within postischemic cerebral microvessels, impairing recovery of blood flow and causing collateral damage to bystander neurons. In a murine model of stroke, increased expression of the glycoprotein adhesion receptors P-selectin (1) and ICAM-1 (2) promotes leukocyte recruitment. Mice lacking these adhesion receptors exhibit reduced leukocyte recruitment and improved postischemic perfusion and are moderately protected from cerebral injury in stroke. Microvascular thrombosis, which occurs in situ in cerebral microvessels downstream from the site of primary occlusion, can be inhibited by the blockade of platelet glycoprotein IIb/IIIa receptor (3).

Although neuronal blood supply may be diminished by microvascular thrombosis and leukostasis, and neurons may be damaged by their proximity to inflammatory foci, it is also possible that inflammatory mechanisms may subject neurons to direct attack. Although neurons do not normally express complement proteins, in Alzheimer's disease, neurons synthesize and express C1q (4), a critical initiator of the classical pathway of complement activation and a ligand for C1qRp, a receptor on myeloid cells (including microglia) that enhances phagocytosis (5). A role for complement activation in stroke has not been elucidated, though it has been shown to play a role in ischemia-induced damage in other tissues (6,7).

Brain tissue sections obtained from mice subjected to 45 min of middle cerebral artery (MCA) occlusion followed by 23 hours of reperfusion showed prominent neuronal staining for C1q in the ipsilateral cerebral cortex (Fig. 1A). Neurons in the contralateral cerebral cortex expressed C1q at barely detectable levels (Fig. 1B), and neurons from nonischemic control brain did not express any detectable C1q (Fig. 1D). As a specificity control for these experiments, murine C1q was purified (8) and then mixed with the primary antibody to C1q (anti-C1q) before staining. These experiments confirmed a specific increase in C1q expression in neurons in the ischemic cerebral cortex but not in the nonischemic cerebral cortex (Fig. 2A). Protein immunoblot analysis demonstrated increased density of a band that was immunoreactive for murine C1q in ipsilateral but not contralateral hemispheric tissue extracts 24 hours after stroke (amounts of C1q protein in nonischemic tissue were similar to that observed in nonischemic control brain). Not only was this band immunoreactive for C1q, but it comigrated with a purified C1q-positive control at the expected relative molecular mass of ∼28 kD (Fig. 2B).

Figure 1

Ischemia-driven neuronal C1q expression, and localization of sites of sCR1 and sCR1sLexbinding. Adjacent tissue sections were incubated with either a primary rabbit antibody to mouse C1q [62 μg/ml, 1:150 dilution in phosphate-buffered saline (PBS); this antibody is a purified immunoglobulin G (IgG) with murine C1q as the rabbit immunogen], or with rabbit affinity-purified anti-sCR1 (19) (42 μg/ml, 1:50 dilution in PBS) at 37°C for 60 min. (Aand B) Sections of brain from a control mouse subjected to 45 min of cerebral ischemia and 23 hours of reperfusion. Immunostaining with anti-C1q demonstrates immunoreactivity on neuronal surfaces in the ipsilateral (A) cerebral cortex, with minimal staining in the contralateral (nonischemic) cortex (B). (C andD) Section of brain from an untreated (nonoperated) mouse. Neither sCR1 (C) nor C1q (D) immunoreactivity can be detected. (E and F) Adjacent brain sections from the ipsilateral hemisphere of an sCR1-treated mouse that underwent stroke immunostained with either anti-sCR1 (E) or anti-C1q (F). Arrowheads indicate neurons that exhibit immunoreactivity for both sCR1 and C1q in adjacent sections. Arrows point to a cerebral microvessel, which stains intraluminally for sCR1 but not C1q. (G andH) Ipsilateral and contralateral brain sections of an sCR1-treated mouse that underwent stroke. The ipsilateral section (G) demonstrates detailing of neuronal bodies and axonal processes from immunoreactivity to anti-sCR1. The contralateral section (H) taken from the same mouse shows minimal immunoreactivity for sCR1. (Iand J) Ipsilateral sections of a brain from an sCR1sLex-treated mouse that underwent stroke. Neuronal immunoreactivity for sCR1 is shown (arrowheads) in the ipsilateral cerebral cortex (I), as well as some immunoreactivity within a cerebral microvessel (arrow). Contralateral cerebral cortex from the same mouse (J) shows minimal immunoreactivity for sCR1.

Figure 2

Effect of cerebral ischemia on neuronal C1q expression. (A) The minimum concentration of IgG that gave positive immunostaining was determined (1:400 dilution), then mixing experiments were performed with an excess of purified murine C1q to block the primary antibody to demonstrate specificity of the neuronal immunostaining. The contralateral (nonischemic) region did not stain with primary antibody alone (23 μg/ml; 1:400 dilution), although the ipsilateral (ischemic) region exhibited neuronal immunoreactivity under the identical staining conditions. When antigen (120 μg/ml, ∼fivefold excess) was mixed with the primary antibody (overnight incubation at 4°C) before experiments, ipsilateral staining was completely blocked. Similarly, no positive immunostaining was detected in the contralateral region in this blocking experiment (not shown). (B) Protein immunoblot analysis for murine C1q. (Reducing conditions were used, 12% SDS-PAGE, 140 μg of total protein per lane, and a dilution of primary antibody of 1:1000). Sites of primary antibody binding were detected with a donkey horseradish-peroxidase–conjugated secondary anti–rabbit IgG (1:2000) and the enhanced chemiluminescent method of immunodetection. Purified murine C1q (250 ng) was used as a positive control (8). Tissue samples from two normal, untreated mice (lanes marked control brain) showed only slight immunoreactivity, similar to that seen in the contralateral hemispheric extracts from two separate mice (mouse A and mouse B) taken 24 hours after stroke. In contrast, a strongly immunoreactive band was seen for both ipsilateral (ischemic) hemispheres from these same mice.

To examine the functional effects of complement activation in the context of ischemia- reperfusion–driven expression of neuronal C1q, we studied the effects of a potent inhibitor of complement activation, soluble complement receptor–1 (sCR1) (7). In nonstroke (control) brain, an affinity-purified antibody to human sCR1 failed to stain any structures (Fig. 1C). However, we used this same antibody to localize sites of sCR1 binding in stroke. When sCR1 was infused into mice just before stroke and brain tissue was then studied after reperfusion, sCR1 was found to colocalize to neurons that express C1q (Fig. 1, E and F, arrowheads), as well as intraluminally in the cerebral microvessels where C1q was not expressed (Fig. 1, E and F, arrows). Both neuronal bodies and their axonal processes in the ischemic region bound sCR1 (Fig. 1G). In contrast, neurons examined in the contralateral cortex from the same mouse bound little if any sCR1 (Fig. 1H).

Infusion of sCR1 into mice immediately before stroke caused a modest reduction in cerebral tissue injury (9) (Fig. 3, A and B). However, because leukocyte adhesion receptor expression (1, 2) and thrombosis (3) in the postischemic cerebral microvasculature may occur independent of complement activation yet still contribute to cerebral tissue injury in stroke, we sought to modify sCR1 to inhibit these additional deleterious inflammatory and thrombotic events. Because selectin expression is up-regulated in stroke (1, 10), we hypothesized that a single molecule that inhibits both complement activation and selectin-mediated adhesive events might simultaneously limit both inflammatory and coagulant reactions in stroke. We therefore covalently modified sCR1 by sialyl Lewis x (sLex) glycosylation (11). We expressed sCR1 in CHO LEC11 cells, which are capable of incorporating the sLex tetrasaccharide in the posttranslational glycosylation of proteins. This glycoprotein product, sCR1sLex, includes at least 10 sLexmoieties per sCR1 molecule (11). In vitro experiments demonstrated that sCR1sLex, but not sCR1, binds cell surface E-selectin and also blocks P-selectin–mediated cellular adhesion (11).

Figure 3

Effect of sCR1 or sCR1sLex administration on outcomes after stroke. For the surgical procedure, neurological scoring, and infarct volume analysis, operators were blinded to the treatment regimen. (A) Compared with sham-operated animals (n = 10), neurological deficit was greater after middle cerebral arterial occlusion and reperfusion in vehicle-treated animals (n= 27, * = P < 0.005). Neurological deficit was less in sCR1-treated (15 mg/kg, n = 37) and sCR1sLex-treated (15 mg/kg, n = 12) animals compared with vehicle-treated controls; ** = P < 0.05, and *** = P < 0.05, respectively). (B) Infarct volumes were determined by staining serial cerebral sections with triphenyl tetrazolium chloride (TTC) and performing computer-based planimetry of the negatively (infarcted) staining areas to calculate infarct volume (with National Institutes of Health image software). Infarct volumes, with dose-response relation of sCR1 and sCR1sLex [n = 27 for vehicle-treated animals (0 mg/kg) (striped bar)]; n = 16, 10, and 37 for sCR1 (gray bars) and n = 10, 10, and 12 for sCR1sLex (black bars), for the 5, 10, and 15 mg/kg doses respectively. Significantly lower infarct volumes were observed in 10 mg/kg and 15 mg/kg doses of sCR1sLex relative to vehicle (* = P < 0.05). Survival figures are indicated below each bar as the number of animals surviving until they were killed at 24 hours (numerator) over the total number of experiments performed (denominator). Panels (C to F) illustrate the effect of preoperative sCR1 (15 mg/kg) or sCR1sLex (15 mg/kg) administration on neutrophil accumulation, platelet accumulation, relative cerebral blood flow, and intracerebral hemorrhage 24 hours after stroke in mice. (C) 111In-labeled PMNs administered preoperatively demonstrate progressively decreasing accumulation in the ipsilateral versus contralateral hemisphere in sCR1- and sCR1sLex-treated animals compared with vehicle treated controls (* = P < 0.03; ** = P < 0.003; *** = P < 0.0001). Ten mice were used to obtain pooled PMNs; n = 8, 7, 5, and 6 for sham, vehicle, sCR1, and sCR1sLex, respectively. Neutrophil accumulation after focal cerebral ischemia and reperfusion was determined by preoperative injection of 1.5 × 105 111In-labeled PMNs administered intravenously immediately before surgery. Brain harvest was performed 24 hours after middle cerebral artery occlusion, and the determination of the differential extent of neutrophil accumulation in the affected ipsilateral hemisphere was calculated as ipsilateral/contralateral counts per million (cpm). (D) Platelet accumulation studies with111In-labeled platelets demonstrate no difference in accumulation in ipsilateral versus contralateral hemispheres in sham mice. sCR1- and sCR1sLex-treated mice show progressively diminished rates of platelet accumulation compared with vehicle-treated controls (* = P < 0.05; ** = P < 0.005; *** = P < 0.0002 versus vehicle-treated controls). Ten mice were used to obtain pooled platelets;n = 6, 6, 5, and 4 for sham, vehicle, sCR1, and sCR1sLex, respectively. Platelet accumulation was determined with 111In-labeled platelets, collected and prepared as previously described (3). Mice received an intravenous injection of 5.0 × 106 111In-labeled platelets immediately before surgery. Brains were harvested at 24 hours after middle cerebral artery occlusion, and deposition in the affected cerebral hemisphere was determined by the calculation of ipsilateral/contralateral counts per minute (cpm). (E) Relative cerebral blood flow was measured at baseline (PreOccl), immediately after middle cerebral artery occlusion (Occl), 45 min later immediately after withdrawal of the occluding suture (Reperf), and immediately before sacrifice at 24 hours (Sac) [n = 7, 12, and 9 for vehicle (striped bars), sCR1 (gray bars), and sCR1sLex (black bars), respectively; * =P < 0.05 versus vehicle]. We obtained measurements of relative cerebral blood flow as previously described (1–3,14) using a straight laser Doppler flow probe placed 2 mm posterior to the bregma, and 6 mm to each side of midline using a stereotactic micromanipulator, keeping the angle of the probe perpendicular to the cortical surface. These cerebral blood flow measurements, expressed as the ratio of ipsilateral to contralateral blood flow, were obtained at baseline, immediately after withdrawal of the occluding suture, and before sacrifice at 24 hours. (F) Mice were treated with vehicle, sCR1 (15 mg/kg), or sCR1sLex (15 mg/kg) immediately before surgery, and intracerebral hemorrhage measured at 24 hours with a spectrophotometric assay that has been validated for use in mice with stroke (20). The means ± SEM of five animals are shown.P = not significant for all comparisons.

sCR1sLex given before stroke could be detected bound to neurons and within vascular lumina 24 hours after stroke (Fig. 1, I and J). In its ability to minimize the volume of infarcted cerebral tissue and reduce neurological deficit, sCR1sLex was more potent than sCR1 (Fig. 3, A and B). Mice treated with sCR1 before middle cerebral artery occlusion demonstrated less neurological deficit at 24 hours than vehicle-treated controls (Fig. 3A). sCR1sLextreatment diminished neurological impairment even further. Mean infarct volumes were lower in the sCR1-treated animals, although large standard errors precluded ascribing statistical significance to these values. However, sCR1sLex markedly diminished infarct volume in a dose-dependent fashion, with up to an 11-fold reduction in cerebral infarct volume observed at the highest dose tested (15 mg per kilogram of body weight) (Fig. 3B) (12).

Because sCR1 and sCR1sLex localize not only to C1q-expressing neurons but to ischemic cerebral microvessels, we examined vascular mechanisms by which these compounds may confer protection to cerebral tissue in stroke. We administered111In-labeled neutrophils immediately before surgery and then quantified their accumulation in ischemic cerebral tissue. Compared with animals undergoing sham surgery, animals subjected to middle cerebral artery occlusion and reperfusion exhibited a nearly threefold increase in ipsilateral leukocyte accumulation (Fig. 3C). Pretreatment with sCR1 (15 mg/kg) caused a 1.7-fold reduction in leukocyte accumulation. However, sCR1 treatment did not reduce neutrophil accumulation to baseline, implicating additional noncomplement-dependent mechanisms of leukocyte recruitment in stroke. sCR1sLex further reduced polymorphonuclear leukocyte accumulation in the ipsilateral cerebral cortex, to a level just slightly above baseline.

When the accumulation of 111In-labeled platelets was similarly tracked, sCR1sLex caused even greater inhibition of platelet accumulation than sCR1 alone (Fig. 3D). These data support the observation that leukocytes and platelets demonstrate substantial adhesion-receptor-dependent interactions (13), which may amplify microvascular thrombosis. Postischemic cerebral hypoperfusion, which is dependent upon the severity of the cerebral ischemic insult and which may be caused by microvascular thrombosis, leukocyte capillary plugging, vascular damage, or abnormal vasoreactivity (1–3, 14, 15), is also a feature of the stroke model used in these studies. Laser Doppler measurements showed equivalent blood flows between ipsilateral and contralateral cerebral cortices at baseline, upon occlusion of the middle cerebral artery, and immediately upon withdrawal of the occluding suture (Fig. 3E). However, by the time the mice were killed, ipsilateral blood flow was significantly higher in sCR1sLex-treated animals, and at an intermediate level in the sCR1-treated animals. The failure of blood flow to return completely back to baseline suggests the recruitment of additional effector mechanisms beyond those that can be inhibited by complement and selectin blockade. Because of concerns that an antiselectin agent, which inhibits platelet recruitment, may increase intracerebral hemorrhage, we performed experiments to examine this possibility. We saw no increase in intracerebral hemorrhage after therapy; there was even a trend toward a decrease with the antiselectin agent (Fig. 3F). Other physiological variables examined did not suggest that the tested compounds had any other adverse effects in mice (16). Even after longer observation periods (72 hours), the beneficial effects of sCR1sLex administration were sustained (Fig. 4A).

Figure 4

Effect of complement/selectin blockade on delayed survival and effect of postischemic administration. (A) Effects of treatment on survival beyond 24 hours. sCR1sLex (15 mg/kg, n = 23) or vehicle (n = 22) was administered immediately before surgery to a separate cohort of animals, and the delayed effects on cerebral infarct volumes and survival were recorded. By 72 hours, all mice that had not died beforehand were killed due to humane considerations (secondary to reduced oral intake and weight loss, and as mandated by our Institutional Animal Care and Use Committee-approved protocol). (B) Effect of postischemic administration of sCR1 or sCR1sLex on cerebral infarction volumes. Compounds (15 mg/kg) or vehicle were administered 45 min after occlusion immediately after withdrawal of the middle cerebral arterial occluding suture. Preocclusion data are those shown in Fig. 2B (15 mg/kg dose data), repeated here for comparison to postischemic administration (* =P < 0.03 and ** = P < 0.05 versus vehicle); survival figures are indicated below each bar as the number of animals surviving until sacrifice at 24 hours (numerator) over the total number of experiments performed (denominator). Striped bars, vehicle; gray bars, sCR1; black bars, sCR1sLex.

In clinical stroke, patients present after the onset of the ischemic event. Any therapeutic intervention must therefore be effective after the onset of ischemia. Both compounds (15 mg/kg) were administered 45 min after ischemic onset immediately after removal of the occlusion. sCR1sLex maintained cerebroprotective efficacy, as reflected in diminished infarct volumes, albeit less than that seen during preischemic administration (Fig. 4B). Blockade of complement activation after the ischemic event with sCR1 was of intermediate efficacy between that of vehicle and that of sCR1sLex. Neurological function and relative cerebral blood flow at 23 hours of reperfusion were also improved by administration of sCR1sLex after occlusion. For control versus sCR1sLex-treated animals, respectively, the data were as follows: neurological deficit score 2.8 ± 0.3 versus 1.5 ± 0.3,P <0.02; cerebral blood flow of the ipsilateral hemisphere relative to that of the contralateral hemisphere was 41 ± 15% versus 75 ± 7%, P < 0.05. Although administration 45 min after ischemia in mice cannot be directly translated into clinical relevance without corroborative human data, a recent clinical trial showed that the average time from stroke onset to emergency room presentation was 57 min (although an additional 100 min were required for a CT scan before administration of tissue plasminogen activator) (17). If an “innocuous” agent can be given at presentation without the delay of an intervening CT scan, then an opportunity will open up for therapeutic intervention in stroke.

These studies demonstrate that ischemic neurons express C1q, presumably to flag the complement system to bring about their own demise or to aid in removing damaged cells, or both. It is not surprising that sCR1 targets and binds to C1q-expressing neurons, because CR1 has been shown to be a ligand for C1q (18). sCR1, a soluble, truncated form of the human complement receptor protein CR1, binds to proteins of the multisubunit C3 or C5 convertases formed as a result of complement activation to promote dissociation of the catalytic subunits (7). sCR1 is moderately cerebroprotective in the setting of reperfused stroke, reducing neutrophil and platelet recruitment. However, when this agent is glycosylated with sLex to additionally inhibit selectin-mediated events, it significantly protects the reperfused brain against leukocyte and platelet recruitment and neuronal injury.

  • * To whom correspondence should be addressed. E-mail: djp5{at}


View Abstract

Navigate This Article