Chemokines and Their Receptors in Intracerebral Hemorrhage

Abstract

Intracerebral hemorrhage (ICH) is a devastating clinical event, which results in a high rate of disability and death. At present no effective treatment is available for ICH. Accumulating evidence suggests that inflammatory responses contribute significantly to the ICH-induced secondary brain outcomes. During ICH inflammatory cells accumulate at the ICH site attracted by gradients of chemokines. This review summarizes recent progress in ICH studies and the chemoattractants that act during the injury, focuses on and introduces the basic biology of the chemokine monocyte chemoattractant protein-1 (MCP1) and its role in the progression of ICH. Better understanding of MCP1 signaling cascade and the compensation after its inhibition could shed light on the development of effective treatments for ICH.

1. Intracerebral Hemorrhage (ICH)

1.1. Introduction

ICH, caused by the rupture of weakened blood vessels and bleeding into the brain constitutes 15–20% of all strokes. It affects between 0.12 and 2 million people each year in the United States and worldwide, respectively [1–3]. Due to the aging of populations, these numbers are expected to grow. The primary causes of ICH are hypertensive arteriosclerosis and amyloid angiopathy [4, 5]. In addition, cerebrovascular malformations, neoplasia, coagulation disorders, and thrombolysis treatment also contribute to the incidence of ICH [4, 5]. The prognosis of ICH is poor: the one-year survival rate is only 38% [6] and about 90% of the survivors experience long-term physical and mental disability [7, 8].

1.2. Pathophysiology

At the onset of ICH, the initial bleeding leads to formation of hematoma and increase of intracranial pressure, which result in primary injury to the brain. A secondary injury is mainly caused by the subsequent inflammatory responses [9, 10]. The blood and its components at the injury site activate brain resident immune cells, microglia, which secrete inflammatory mediators [11, 12], which results in an amplification of the inflammatory response. Infiltrating peripheral polymorphonuclear leukocytes and monocytes as well as brain edema are evident [13–15]. In addition, cerebral mast cells have been shown to also contribute to the secondary injury after ICH [16, 17]. The migration into the CNS and the accumulation of the inflammatory cells results in the release of cytotoxic molecules that cause cell death in the peri-hematoma area. As the injury progresses, dead cells are cleared and eventually the hematoma can be resolved. These reparative processes, interestingly, are executed by the same cells that cause the secondary brain injuries (activated microglia and infiltrated leukocytes) [9, 18].

1.3. Animal Models of ICH and progression of pathology

There are several animal models of ICH, but two are the most widely used to mimic clinical symptoms and pathological changes of human ICH: the collagenase-induced ICH model and the whole-blood model. In the first model, collagenase, a bacterial enzyme, is injected to the brain (mostly in striatum), where it degrades collagen IV, a major component of blood vessel wall, causing rupture of the blood vessels and thus ICH [11, 12, 19]. This model is simple and replicates the blood vessel rupture pathology in ICH patients. Additionally, it is very reproducible. The size and location of hematoma reported by various laboratories are consistent [9, 20–22]. However, it should be noted that the collagenase induced ICH model has two major drawbacks. First, it does not replicate the vascular injury observed in human patients prior to ICH, such as hypertension/atherosclerosis induced vascular changes. Second, it introduces an exogenous enzyme, bacterial collagenase, into the brain. Although there is evidence suggesting that collagenase alone does not activate microglia or affect cell survival [11, 23, 24], it modifies the extracellular matrix around its delivery site, and may alter inflammatory or immune responses via other cell types or pathways and thus affect the progress of ICH. In the whole-blood model, blood either from the animal itself or a donor is injected into the brain (usually striatum). Such delivery replicates the pathology after rupture of blood vessels in human patients [25–29]. Like the collagenase induced ICH model, the whole-blood model also has some disadvantages: it lacks the pathologically injured vasculature, and does not cause rupture of blood vessels. The lesion size and location from this model is less reproducible compared to collagenase ICH. It has been reported that the whole blood model usually results in an umbrella-shaped narrower slit-like lesion [30], probably due to the rapid distribution of blood along corpus callosum or white matter tracts.

Both models have advantages and disadvantages. However, neither of them could mimic all the pathological changes of ICH in human patients. Recently, a spontaneous ICH model has been developed in rodents [31]. This new model induces ICH via acute hypertension, the most common etiology of hemorrhagic stroke. The pathophysiology and reproducibility of this new model needs further investigation.

In all ICH models blood leaks into the brain parenchyma. Similarly to the human condition, the accumulation of blood in the brain increases the intracranial pressure, causing immediate primary injury. In the collagenase model within 30 minutes the entrance of blood components, cellular and molecular, such as red blood cells, leukocytes, proteases, hemoglobin, iron deposition into the brain leads to inflammatory responses, which initially involve the microglia, the brain resident immune cells [32]. Microglia change their morphology, alter gene expression profile, secrete inflammatory mediators, proliferate and migrate to site of injury [11, 12]. Other cell types, such as astrocytes, infiltrating macrophages, monocytes and neutrophils, are also activated and migrate to the hematoma site [9, 33, 34]. Early after ICH, these activated cells secrete pro-inflammatory cytokines/chemokines, including TNF-α, IL-1β, and MCP1/CCL2, recruit additional inflammatory cells to the injury site, and form a barrier preventing the spread of injury to other sites [9–11]. With the progression of disease, inflammatory cells, especially microglia and macrophages, transit from pro-inflammatory to anti-inflammatory states. They clear cell debris at the site of injury by phagocytosis, thus promoting recovery [9, 10, 35], which in the collagenase model takes place within 15–20 days after the induction of ICH. With the resolution of hematoma, those activated inflammatory cells become resting again [11, 22]. The time course schematic of ICH is shown in Figure 1.

Figure 1. Time course of ICH.

Early after ICH, microglia become activated and migrate to the site of injury. Fast-responding neutrophils infiltrate into the brain and together with activated microglia, produce MMP and chemokines/cytokines. With the progression of disease, BBB is compromised and monocytes enter the brain. Activated microglia and infiltrated leukocytes generate pro-inflammatory cytokines, which together with ROS produced by lipid peroxidation induce cell death. At the later stage of ICH, the micro-environment changes from pro-inflammatory to anti-inflammatory. Activated microglia and infiltrated cells clear cell debris by phagocytosis, thus promoting recovery.

2. Chemokines

Chemokines are small basic proteins that activate and attract immune cells in response to insult [36–39]. Four types of chemokines are found based on the number and position of conserved cysteine on their primary sequences: C, CC, CXC, and CXXXC [40–42]. CXC chemokines recruit short-lived but fast responding neutrophils, whereas CC chemokines induce the migration of monocytes and cells of monocytic origin, like microglia [43]. Chemokines exert their functions via their high affinity receptors [44]. It has been shown that one receptor may have more than one ligands [9] and one ligand may interact with more than one receptors. This promiscuity makes the chemokine-receptor system very complex.

During ICH there have been many reports indicating that chemokines become upregulated [9, 45, 46], functioning to attract local and systemic immune cells into the CNS parenchyma. In a functional analysis study using human samples, many chemokines related to leukocyte extravasation, including CXCL8 (IL-8), CXCL2, CXCL5, CCL3, CCL4, and CCL20, have been found to be over-expressed in perihematomal areas [45]. In another study using human brain samples, besides the chemokines mentioned above, CCR1 and downstream effector molecules related to chemokine signaling were also reported to be activated [47]. These data suggest a pivotal role of chemokines and chemokine receptors on ICH.

2.1. MCP1

MCP1, a CC type chemokine (also called CCL2), is drastically upregulated and transiently expressed during injury or inflammation in the central nervous system (CNS), and acts on its receptor CCR2. Under physiological conditions, MCP1 is expressed at low levels by neurons, astrocytes and brain microvascular endothelial cells (BMEC) in the CNS [48–63]. Upon injury, however, the expression of MCP1 is dramatically increased. It is believed that, after it is released into the extracellular space, MCP1 forms a gradient to attract CCR2-expressing monocytes and microglia.

Microglia are the immune-competent cells in the brain. Under physiological conditions, they have a ramified morphology (smaller cell body surrounded by many long, thin, and highly dynamic processes). They extend and retract their processes continually to sense changes in the surrounding microenvironment [64]. When an injury occurs, microglia become activated. They change to an ameboid morphology, alter their gene expression profile, migrate to site of injury, and affect the outcome of injury [65–70].

Accumulation of monocytes and activated microglia has been found in many CNS injuries, such as ischemia, excitotoxicity and hemorrhage [70–77]. Lack of CCR2, however, abrogates the trafficking of microglia and leukocytes, indicating that MCP1-induced chemotaxis is dependent on CCR2 [78, 79]. In kainate-induced excitotoxicity model, we and others found that microglial migration/accumulation was attenuated in MCP1^−/− mice and wild-type mice administered with MCP1 blocking antibody [76, 80]. Interestingly, similar results were found in mice lacking plasminogen (plg) or tissue plasminogen activator (tPA), which converts plg to active plasmin [81, 82]. These results suggest that MCP1-CCR2 axis may crosstalk to the plg activation system. Further studies in our lab showed that plasmin cleaved MCP1 at K104, generating a N-terminal fragment with a much higher chemotactic potency [76, 83]. In our rescue experiments, we found infusion of plasmin-cleaved-, but not full length (FL)-MCP1 into the CNS restored excitotoxicity-induced microglial activation/migration and subsequent neuronal death in plg^−/− mice, suggesting plasmin-truncated MCP1 is the active form of MCP1. Further mechanistic studies revealed that plasmin-truncated MCP1 activated Rac1 and promoted the formation of lamellipodia more efficiently than FL-MCP1 [83].

MCP1 exerts its biological functions by binding to its high affinity receptor, CCR2, a G-protein-coupled receptor. Although MCP1 can only bind to CCR2, CCR2 has multiple ligands, including MCP1 (CCL2), MCP2 (CCL8), MCP3 (CCL7), MCP4 (CCL13) [84–86]. The relative contribution of each ligand to CCR2-mediated functions remains elusive [87, 88]. It should be noted that there is only CCR2 isoform in humans, whereas two CCR2 isoforms with different C-terminus are found in rodents [89], which may suggest different ways of regulation on MCP1 activity.

In the peripheral system, CCR2 is expressed in the so-called inflammatory cells, including monocytes, dendritic cells and memory Th1 cells [90–92]. In the CNS, CCR2 is expressed by various cell types, such as neurons, astrocytes, and microglia, and microvascular endothelial cells [14, 93–97]. Under physiological conditions, the CCR2 expression levels are very low. When an injury occurs, however, the expression of CCR2 is dramatically upregulated primarily in astrocytes and microglia [98–100].

Binding of MCP1 on microglial (or astrocytic) cell surface induces the internalization of CCR2, probably via endocytosis [13]. This mechanism has been suggested to regulate extracellular MCP1 levels [101–103] or facilitate the transcytosis of MCP1 across endothelial cells [97]. Many signaling molecules, including mitogen-activated protein kinases, protein kinase C, and Rho, are activated following MCP1-CCR2 interaction [14, 15, 104], suggesting that a variety of signaling pathways are involved.

2.2. MCP1 and ICH

Given that inflammation plays a crucial role in brain injury after ICH and that MCP1 is a potent chemoattractant for monocytes/microglia, we investigated the role of MCP1-CCR2 system in ICH. In collagenase ICH model, we found that mice deficient for MCP1 or CCR2 had smaller hematoma early after ICH. The size of hematoma at later stage, however, was bigger in the knockout mice [105], suggesting a delayed recovery. Consistent with the important role of MCP1 on microglial recruitment, the activation/accumulation of microglia was attenuated in the knockout mice early after injury. Surprisingly, 3 and 7 days post injury, more microglia were found in the knockout mice than in wt mice [105], suggesting activation of MCP1-CCR2 independent signaling events in the knockout mice. Besides, the infiltration of fast-responsive neutrophils and total leukocytes was decreased 1–3 and 7 days post injury, respectively, in both knockout mice. These data are consistent with our previous findings that MCP1^−/− and/or CCR2^−/− mice have a “tighter” BBB [13, 105]. Although not changed among different genotypes early after injury, the functional marker, inducible nitric oxide synthase (iNOS) kept at high level 7 days post injury in the knockout mice, indicating a long-lasting activation of ROS in the knockout mice. In addition, brain edema and neuronal loss decreased overtime in wt mice but the opposite was found in both knockout mice. Neurological functions were mirrored by these pathological changes [105]. Together, these data suggest that early inhibition of MCP1-CCR2 axis after ICH could be therapeutic. It should be noted that long-term inhibition of MCP1-CCR2 may be detrimental because lack of MCP1 or CCR2 delays the recovery of ICH.

Since plasmin cleaves and activates MCP1 and it is the active MCP1 that compromises the integrity of BBB in mice, we asked whether plasmin-mediated activation of MCP1 is involved in ICH. Infusion of plasmin-truncated MCP1 and FL-MCP1 back to MCP1^−/− mice significantly increased the injury volume 1 day post injury but speeded up the resolution of hematoma. Similarly, neutrophil (early after ICH) and leukocyte (later after ICH) infiltration was increased in these mice. Microglial activation/accumulation and iNOS expression increased early but decreased later after ICH. Brain edema, neuronal loss and neurological functions followed the same trend. It is worth noting that the changes are more significant in mice with plasmin-truncated MCP1 injection than that with FL-MCP1 injection. Infusion of plasmin-uncleavable MCP1, on the other hand, did not change the pathological patterns of MCP1^−/− mice over time. Taken together, our data suggest that plasmin-mediated truncation of MCP1 does play a critical role in ICH in mice. The role of MCP1 after ICH is summarized in Figure 2.

Figure 2. Role of MCP1 in ICH.

ICH induces up-regulation of MCP1, which is further activated by plasmin in rodents (by other mechanisms in humans). The activated MCP1 then activates/recruits microglia and induces leukocyte infiltration. Additionally, activated MCP1 disrupts the integrity of BBB, promoting the infiltration of leukocytes into the brain and cell death. The infiltrated cells and activated microglia secrete cytotoxic cytokines early after ICH, and clear cell debris probably via phagocytosis at later stages.

Since human MCP1 does not contain the C-terminal tail found in mouse MCP1, which is removed by plasmin [105], plasmin-mediated activation of MCP1 does not work in humans. Whether human MCP1 needs to be activated and what is the mechanism remains elusive. Based on the fact that rodents have only one CCR2, whereas humans have two alternatively spliced forms of CCR2 (CCR2A and CCR2B) [89], we speculate that the various expression level of the CCR2 isoforms in different cells may be a way to regulate MCP1 activity in humans. It has been shown that CCR2B is the major isoform for monocytes and activated NK cells, whereas mononuclear cells and vascular smooth muscle cells predominately express CCR2A [106]. Additionally, human MCP1 activity may also be regulated at transcription level or protein level (turnover rate) or by post-translational modifications. These data suggest that MCP1 and/or CCR2 instead of plasmin should be targeted for the treatment of ICH in humans.

3. Targets for ICH Treatments

Currently the therapy for ICH is primary supportive care and control of medical risk factors, such as high intracranial pressure [107, 108]. There are no effective treatments to attenuate the consequences of ICH. Thus new effective treatments are necessary. Since inflammation plays an important role in the secondary injuries after ICH, anti-inflammatory therapies may be a path to pursue. In fact, many inflammatory targets, such as microglial activation, leukocyte infiltration, matrix metalloproteinase (MMP) activation, and reactive oxygen species (ROS) production are being explored.

3.1. Microglial Activation

Microglia are the first non-neuronal cell type that responds to ICH in the brain. They become activated within an hour after the onset of ICH [29, 109], their activation peaks at 3–7 days [11, 12, 105, 110], and return to normal by 3–4 weeks [111, 112]. Many studies suggest that activated microglia contribute to both ICH-induced early [9, 18, 113] and secondary brain injury [9, 10, 35]. Activated microglia may damage the brain by releasing cytotoxic factors, such as iNOS, cytokines, chemokines, prostaglandins, cyclooxygenase-2, ROS, and proteases [9, 10, 105, 109]. However, it should be noted that activated microglia may also play neuroprotective roles by secreting growth factors and phagocytosing cell debris. Which role they play is largely dependent on the time that has elapsed since the injury. Previous work in our lab showed that microglia/macrophage inhibitory factor (the tripeptide MIF/TKP) inhibited the activation of microglia, decreased hematoma volume, and improved neurological function when given 2 days before or 2 hours after in collagenase-induced ICH model [11, 12]. Consistently, inhibition of microglial activation with minocycline, a neuroprotectant that inhibits MMPs [114], played a beneficial role by reducing brain edema and maintaining BBB integrity, although neuronal death was not changed [115–118].

3.2. Leukocyte Infiltration

Accumulating evidence shows that leukocytes infiltrate into the brain and accumulate in the penumbra after ICH [9, 33, 109]. In both collagenase and whole-blood ICH models, neutrophils were reported to infiltrate into the brain in as early as 4 hours and peak around day 3 in mice and rats [34, 105, 109, 111, 119]. In human postmortem brains, leukocytes were found in or around hematoma and their infiltration correlated with cell death in the CNS [120, 121]. In addition, the number of leukocytes in peripheral blood positively correlates with hematoma volume and has been considered an independent predictor of early clinical worsening in primary ICH [122, 123]. Similar to activated microglia, neutrophils may exacerbate brain injury by secreting pro-inflammatory cytotoxic mediators [124], producing ROS, and compromising the BBB integrity [125]. It has been shown that lack of CD18, a surface molecule important for the infiltration of leukocytes, decreases brain neutrophil number, brain edema, and mortality in collagenase induced ICH model [126].

3.2.1. MMP Activation

During inflammation MMPs become activated and mediate extracellular remodeling. MMPs are expressed at low level and in an inactive state in normal brain. Upon injuries, however, MMPs are upregulated and activated by other MMPs, plasmin, or tPA. Studies in our lab showed that MMP-2 and MMP-9 activities were increased 1–3 days post ICH in mice [11, 34]. The increase of MMP-9 activity has been confirmed in other ICH models [28, 127–131]. Other MMPs, including MMP-3 and −12 have also been reported to increase after ICH [116, 132]. It has been shown that mice deficient for MMP-3, −9, and −12 show less brain injury after ICH [28, 132, 133]; on the other hand, Tang and colleagues reported that MMP-9^−/− mice have enhanced hemorrhagic brain injury [19]. This discrepancy could be due to different experimental conditions. In addition to genetically modified mice, pharmacological inhibitors have also been used to block MMPs activities. It has been shown that the broad-spectrum MMP inhibitors BB-1101 and GM6001 reduce brain edema, decrease brain injury, and improve neurological functions after ICH [34, 127, 134]. Minocycline has been reported to reduce brain edema and improve functional recovery, although it does not reduce neuronal death [115, 135]. Minocycline, however, failed to promote functional recovery when given 3 hours after ICH [136]. This negative result could be due to different animal models or timing of drug administration. Studies using human samples mainly focused on MMP-9. Accumulating evidence showed that blood MMP-9 levels increased after acute spontaneous ICH and were associated with subsequent pathological changes [123, 137–139]. MMP-9 level has also been reported to associate with BBB integrity after stroke in humans [140]. Pathological studies using postmortem human brains (within 6 hours after death) revealed that MMP-9 was upregulated in neurons and reactive astrocytes in peri-hematoma area [20, 141]. Expression of MMP-9 was also increased in the ipsilateral hippocampus from 2 hours to 5 days post ICH, although it was also elevated in the contralateral hippocampus to a lesser extent [142].

3.3. ROS Production

After hemorrhagic injury, hemolysis of red blood cells leads to hemoglobin degradation and iron deposition in the brain [143]. Excess iron causes lipid peroxidation and the formation of free radicals [144, 145]. The ROS, released by activated microglia and infiltrating leukocytes [146, 147], contribute to brain injury after ICH [18, 148–151]. It has been shown that activation of Nrf2, a transcriptional factor that regulates the expression of antioxidant enzymes, improves stroke outcomes in rodents [152, 153]. Consistently, lack of Nrf2 leads to more severe brain damage in both permanent and transient stroke models [152, 154]. In both collagenase and whole-blood ICH models, Nrf2−/− mice showed increased ROS production, cytochrome c release and neurological deficit, suggesting a detrimental role of ROS in ICH [155, 156]. In accordance with these observations, deferoxamine, an iron chelator, was neuroprotective in whole-blood ICH model in rats and piglets [25, 27, 157–159], although one study showed that it was not beneficial in collagenase ICH model in rats [160]. In the collagenase-induced ICH model in mice, deferoxamine is neuroprotective, although it does not reduce lesion volume and brain edema [161]. 2, 2’-dipyridyl, a lipid-soluble iron chelator, has been shown to be beneficial in both ICH models in mice [162]. In addition, serum ferritin level associates with penumbra edema in patients 3–4 days post spontaneous ICH [163] and high serum ferritin level also indicates poor outcomes [164].

4. Future directions

Recent studies from our lab and others suggest that early inhibition of inflammatory responses (such as microglia/macrophage activation, leukocyte infiltration, inflammatory mediator release) may be beneficial but long-term inhibition delays the recovery. The early beneficial role of lack of MCP1/CCL2 is probably due to decreased cytokine & chemokine secretion, leading to attenuated neuronal death. The delayed recovery, however, may be caused by the dysregulation of inflammatory response. The activation of alternative chemokine signaling pathways and overactivation/accumulation of microglia/macrophages may prolong the detrimental phase of inflammation or prevent the transition from detrimental phase to beneficial phase. Thus, further investigations should focus on the time course of MCP1/CCL2-CCR2 activation, the compensation of inhibiting MCP1/CCL2-CCR2 activation, and the potential molecular mechanisms regulating its expression. To validate the data from genetic modified animals, the effect of pharmacological inhibition of MCP1-CCR2 axis on ICH should also be examined. In addition, given that little data about ICH are generated using human postmortem brains and inflammatory signaling events may be different in rodents and humans (rodent MCP1 needs to be activated by plasmin, whereas human MCP1 does not), ICH studies using human brain samples are in great need. Last but not least, combination treatments targeting multiple biological processes could be beneficial and should be investigated.