|Home | About | Journals | Submit | Contact Us | Français|
Mechanisms of formation and growth of intracranial aneurysms are poorly understood. To investigate the pathophysiology of intracranial aneurysms, an animal model of intracranial aneurysm yielding high incidence of large aneurysm formation within a short incubation period is needed. We combined two well-known clinical factors associated with human intracranial aneurysms—hypertension and the degeneration of elastic lamina— to induce intracranial aneurysm formation in mice. Roles of matrix metalloproteinases (MMPs) in this model were investigated utilizing doxycycline, a broad-spectrum MMP inhibitor, and MMP knockout mice.
Hypertension was induced by continuous infusion of angiotensin-II for two weeks. The disruption of elastic lamina was achieved by a single stereotaxic injection of elastase into the cerebrospinal fluid at the right basal cistern. 77% of the mice that received 35 milli-units of elastase and 1000 ng/kg/min angiotensin-II developed intracranial aneurysms in two weeks. There were dose-dependent effects of elastase and angiotensin-II on the incidence of aneurysms. Histologically, intracranial aneurysms observed in this model closely resembled human intracranial aneurysms. Doxycycline, a broad-spectrum MMP inhibitor, reduced the incidence of aneurysm to 10%. MMP-9 knockout mice, but not MMP-2 knockout mice, had reduced the incidence of intracranial aneurysms.
In summary, a stereotaxic injection of elastase into the basal cistern in hypertensive mice resulted in intracranial aneurysms that closely resembled human intracranial aneurysms. The intracranial aneurysm formation in this model appeared to be dependent on MMP activation.
Intracranial aneurysms are considered to be common among the general population. Subarachnoid hemorrhage from ruptured intracranial aneurysms results in catastrophic consequences causing severe morbidity and high mortality.1 Despite recent advances in diagnosis and treatment, the mechanisms for the formation, growth, and subsequent rupture of intracranial aneurysms are not yet well understood.
Clinically, systemic hypertension is associated with intracranial aneurysm formation and subarachnoid hemorrhage from aneurysmal rupture.2–4 However, a causal relationship between hypertension and intracranial aneurysm formation or subarachnoid hemorrhage has not been fully established. Histologically, degeneration and disruption of the elastic lamina are key characteristics of human intracranial aneurysm.2, 5, 6 Degeneration or disruption of elastic lamina may be attributed to normal aging process or damages caused by hemodynamic stresses.7 Such changes in the elastic lamina have often been considered as pre-aneurysmal changes that eventually lead to the maturation of aneurysms.8, 9 Elastase-induced fragmentation of elastic lamina has been successfully used to induce aneurysms in the carotid artery10, 11 and aorta12 in animals. In these models, because of the ease of surgical or endovascular access of these blood vessels, elastase was applied intraluminally to induce degeneration of elastic lamina.
In this study, we present a new mouse model of intracranial aneurysms. To induce intracranial aneurysm formation, we combined hypertension and the degeneration of elastic lamina, two well-known clinical factors associated with human intracranial aneurysms. Hypertension was induced by continuous infusion of angiotensin-II; degradation of elastic lamina was induced by a single injection of elastase into the cerebrospinal fluid using a stereotaxic method. Intracranial aneurysms in this model histologically mimic human intracranial aneurysms. In addition, roles of matrix metalloproteinases (MMPs) in this model were investigated utilizing the MMP inhibition by doxycycline, a broad-spectrum MMP inhibitor, and MMP knockout mice.
In order to test whether the combination treatment of single elastase injection and pharmacologically-induced hypertension can cause the formation of intracranial aneurysms in mice, we treated C57BL/6J mice (8–10 weeks old) with a single stereotaxic injection of elastase at the right basal cistern, immediately followed by induction of systemic hypertension with continuous injection of angiotensin-II. After two weeks, we sacrificed the mice and perfused the animals with a bromophenol blue dye and gelatin mixture.
Our preliminary study indicated a wide variation in vessel diameter for the Circle of Willis and its major branches in mice. In order to have a conservative and consistent control for our experiment, we used the diameter of the basilar artery–one of the larger vessels in the brain circulation with relatively little variation—as a reference blood vessel. Aneurysms were operationally defined as a localized outward bulging of the vascular wall whose diameter is greater than 150% of the basilar artery diameter. The Circle of Willis or its major primary branches were assessed by two investigators in a blinded manner.
To assess the relationship between the elastase dose and the incidence of aneurysms, four groups of mice were prepared: Group 1: 35 milli-units of heat-inactivated elastase, Group 2: PBS (n=10), Group 3: 3.5 milli-units of elastase (n=10). Group 4: 17.5 milli-units of elastase (n=20), and Group 5: 35 milli-units of elastase (n = 44). Angiotensin-II was administered at 1000 ng/kg/min for two weeks in all groups.
To assess the relationship between the angiotensin-II concentration and the incidence of aneurysms, three groups of mice were prepared: Group 1: continuous infusion of PBS via osmotic pump (n=10), Group 2: 500 ng/kg/min of angiotensin-II (n=10), Group 3: 1000 ng/kg/min of angiotensin-II (n=44). This group (Group 3) is the same as the Group 5 in the above-mentioned experiment testing the dose dependency of elastase effects. 35 milli-units of elastase were administered to all groups.
To assess the effects of doxycycline, a broad-spectrum MMP inhibitor, on the incidence of aneurysms, 10 mice received the doxycycline through drinking water (40 mg/kg/day). Doxycycline treatment was started 3 days before the elastase injection and continued for 17 days (3 days + 2 weeks). Previously, this dose administered through drinking water was used to successfully suppress MMPs involved in various disease models in mice 13–15. For the control group, 10 mice received drinking water without doxycycline. These mice are different from the mice used in the dose-response studies describe above. To assess of roles of MMP-9 and MMP-2 specifically, we used MMP-9 knockout mice and MMP-2 knockout mice (n =10 in each group).16 All mice received a single injection of 35 milli-units of elastase and angiotensin-II (1000 ng/kg/min).
For continuous variables, we used analysis of variance (ANOVA) followed by Fisher LSD test, which was performed as a post-hoc analysis. When repeated measurements were performed, ANOVA with a repeated measurement design was used. For the incidence, we used chi-square test. Data were presented as mean ± S.D. Statistical significance was taken at P < 0.05.
Among the mice that received a single stereotaxic injection of elastase at 35 milli-units into the basal cistern and a continuous subcutaneous infusion of angiotensin-II at 1000 ng/kg/min, 77% of the mice (34/44) were found to have developed intracranial aneurysm formations along the Circle of Willis or its major branches.
Figure 1 shows a schematic view of normal brain vasculature (A), normal cerebral arteries from a control brain (B) and representative intracranial aneurysms (C–E). In order to visualize the Circle of Willis and its major branches, mice were perfused with blue dye (bromophenol blue). In the mice treated with angiotensin-II and elastase, large intracranial aneurysm formations were found mostly along the right half of the Circle of Willis and its major branches (C–E). Most of the aneurysms were larger than 500 μm, approximately 3 to 5 times larger than their parent arteries. A mouse that died at day 12 had fresh blood clots along the right middle cerebral artery (E), revealing subarachnoid hemorrhage. There was a larger aneurysm formation inside the blood clots of subarachnoid hemorrhage (E).
Figure 1F shows the locations of intracranial aneurysms in mice that received a single stereotaxic injection of 35 milli-units of elastase and a continuous infusion of Angiotensin-II at 1000 ng/kg/min. Majority of the intracranial aneurysms were located ipsilaterally to the elastase injection site. This was consistent with the methylene blue dye distribution found in the test dye injection. Small numbers of intracranial aneurysms were found on the contra-lateral side of the injection site as well as on the branches of the basilar artery.
Continuous infusion of angiotensin-II successfully induced systemic hypertension. At one week and two weeks after the initiation of continuous infusion of angiotensin-II at 1000 ng/kg/min, the blood pressure was significantly higher than the baseline (109 ± 11 vs. 142 ± 37 mmHg vs. 140 ± 30 vs, P < 0.05) (Figure 2C).
To investigate the kinetics of elastase activity after a single injection of elastase into the cerebrospinal fluid space, we analyzed elastase activity in the cerebrospinal fluid after elastase injection. Elastase activity levels were under the detection levels (less than 0.005 U/mL) at the baseline or after sham injection. At thirty minutes or at six hours after the injection of 35 milli-units of elastase (n = 3 in each group), elastase levels were under the detection level. As a next step, we injected 350 milli-units of elastase, a dose 10 times higher than what was actually used to induce aneurysms. At thirty minutes after injecting of 350 milli-units of elastase, the elastase levels were 0.025 ± 0.003 U/mL (n = 3). At six hours after injecting 350 milli-units of elastase, the elastase activity levels returned to levels under the detection level (n = 3). These data indicate that the elastase activity after a single injection of elastase was relatively short.
Figure 2A shows a dose-dependent relationship between the incidence of intracranial aneurysm and the concentration of elastase. The incidence of aneurysms in groups that received 35 milli-units of heat-inactivated elastase, 0 (PBS injection), 3.5, 17, and 35 milli-units of elastase were 0%, 0%, 10%, 60%, and 77%, respectively.
Figure 2B shows a dose-dependent relationship between the incidence of intracranial aneurysms and the concentration of angiotensin-II. The total incidence of aneurysms in groups that received 0 (PBS infusion), 500, and 1000 ng/kg/min of angiotensin-II were 0%, 20%, and 77%, respectively. There was a dose-dependent relationship between the angiotensin-II dose and systolic blood pressure (Figure 3). The blood pressure for groups that received 0 (PBS), 500, or 1000 ng/kg/min of angiotensin-II, was 111 ± 6.7, 127 ± 18.8, and 142 ± 37 mmHg, respectively, at one week (P < 0.05) (Figure 2C), showing a dose dependent effect of angiotensin-II on the blood pressure.
Histological assessment of intracranial aneurysm tissues from this model showed aneurysm formations with a varying degree of structural abnormalities that were similar to those reported in human intracranial aneurysms.2 Generally, intracranial aneurysms in this model had a vascular wall with thick segments. Figure 3 shows a normal basilar artery (A, C) and a representative intracranial aneurysm (B, D).
In the normal cerebral artery from control mouse (A, C), there were two to three layers of smooth muscle cells (A2) and a single, thin continuous layer of endothelial cells (A3). Fibroblasts were very scarce (A4). Elastica van Gieson and trichrome (C1, C2) staining showed one layer of elastic lamina as previously described.2 In an intracranial aneurysm (B, D), the thin vascular wall showed intact endothelial and smooth muscle cell layers, while the thick vascular wall showed discontinued endothelial cell layers and scattered, faint smooth muscle staining (B2, B3). Elastica van Gieson and trichrome stainings revealed severely disorganized elastic lamina in both thin and thick portions of the artery (D, E).
Potential roles of inflammation in the growth of intracranial aneurysms—both ruptured and unruptured— have been suggested by observational studies analyzing the presence of inflammatory cells or inflammatory markers in human intracranial aneurysm tissues and serum samples.6, 7, 17, 18 Therefore, we investigated the presence of inflammatory cells in our intracranial aneurysm samples. Figure 4 shows a normal cerebral artery and a representative intracranial aneurysm.
A normal basilar artery from the control mouse (A) showed absence of inflammatory cells. In the intracranial aneurysm tissue (B), leukocytes (CD45 positive cells) were present throughout the vascular wall (B1). Distribution of macrophages (CD68 positive cells) generally overlapped with the leukocyte distribution (B2). Small numbers of CD4 positive T-lymphocytes were present in the thin wall portion of the aneurysm. However, CD4 positive T-lymphocytes were not detected in the thick aneurysm wall (B4).
Presence of inflammatory cells in intracranial aneurysms tissues suggests potential roles of proteinases produced by inflammatory cells in the formation of intracranial aneurysms.7, 17 Studies showed elevation of serum elastase and collagenase in patients with intracranial aneurysms 19–22 and, matrix metalloproteinase-9 (MMP-9) and MMP-2 were detected in human intracranial aneurysm tissues.23 To test potential contributions from MMPs to the formation of intracranial aneurysms in this model, three lines of experiments were performed. First, MMP activity in intracranial aneurysm tissues was assessed using in situ zymography. Second, we treated the mice with doxycycline, a broad-spectrum MMP inhibitor. Third, intracranial aneurysm formation was assessed in MMP-9 and MMP-2 knockout mice.
Figure 5(A–F) shows in situ zymography and H&E stainings of a normal brain vasculature and a representative intracranial aneurysm in this model. While the normal brain vasculature lacked any appreciable gelatinase activity (B), intracranial aneurysms showed intense fluorescence, indicating robust gelatinase activity (A). Pretreatment of the tissues with 1,10-phenanthroline monohydrate (MMP inhibitor) abolished the gelatinase activity (C), showing that the gelatinase activity observed in the intracranial aneurysm tissues was from MMPs.
Figure 5G shows the incidence of intracranial aneurysms in the control group (wild type), doxycycline treated group (wild type), MMP-9 knockout mice, and MMP-2 knockout mice. All groups received the elastase injection and angiotensin-II infusion. The incidence of intracranial aneurysms in the control group (wild type) was 70%, consistent with the results from the dose-dependence studies described above. The doxycycline treatment reduced the incidence of intracranial aneurysms to 10% (P < 0.05), indicating potential roles of MMPs in the formation of aneurysms in this model. Similar results were previously reported using a different model of intracranial aneurysms in rats.24 The incidence of intracranial aneurysms was reduced to 40% in MMP-9 knockout mice compared to the wild type mice (P < 0.05). However, there was no difference in the incidence of aneurysms between the wild type mice and MMP-2 knockout mice (70% vs 60%).
In this study, we showed that the combination of hypertension and a single stereotaxic injection of elastase into the cerebrospinal fluid at the basal cistern resulted in the formation of intracranial aneurysms that recapitulated key features of human intracranial aneurysms. The single stereotaxic injection of elastase was used to induce disorganization or disruption of elastic lamina, a structural change that is considered to be pre-aneurysmal change.10, 11, 25 To induce systemic hypertension, we utilized continuous infusion of angiotensin-II via subcutaneously implanted osmotic pump. This is a well-established and easily reproducible method for hypertension in mice.26, 27 Although neither systemic hypertension nor degeneration of elastic lamina alone could cause intracranial aneurysms in our model, the combination of both factors resulted in large aneurysm formation. These findings suggest the synergistic effects of these two factors in the formation of intracranial aneurysms in this model. In addition, we were able to show dose dependent effects of angiotensin-II and elastase on the formation of aneurysms, further supporting the causative roles of hypertension and degeneration of elastic lamina in aneurysm formation in this model.
Aneurysms formed in this model were generally large and easily distinguishable from the normal arteries without using histological assessment. Aneurysm formations in this model can serve as a simple and easily interpretable outcome for future studies that utilize various inhibitors, knockout mice, or transgenic mice to test roles of specific molecules and pathways in the pathophysiology of intracranial aneurysms.
Histologically, intracranial aneurysms observed in this model closely resembled human intracranial aneurysms.2 The intracranial aneurysms in this model showed thin and thick vascular walls with partial loss of smooth muscle cells; elastic laminas showed various degenerative changes ranging from a partial disruption to a complete loss. In addition, the infiltration of leukocytes, especially macrophages, into the aneurysm wall was clearly present in our model, consistent with the observational studies that analyzed human intracranial aneurysm tissues and serum samples.6, 17
In this model, intracranial aneurysms formed within two weeks. Such short incubation time for this model’s intracranial aneurysm formation would allow practical and feasible screening of various molecular targets. However, it should be cautioned that aneurysm formation and maturation are believed to require a much longer time frame in humans. The relatively shorter time span of aneurysm formation in this model may be due to the higher magnitude of physiological insults—hypertension and elastase-induced inflammation—generated to create the model. In animal models, inciting factors that lead to the disease often need to be exaggerated or accentuated to achieve severe disease phenotype at high incidence within a practical time frame. However, by doing so, the model may fail to recapitulate or skip events that play key roles in the human disease.
Although angiotensin-II induced hypertension is a well-established, well-characterized model of hypertension in mice,26, 27 the effects of angiotensin-II are diverse. Angiotensin-II can exert various effects on the vasculature in addition to its hypertensive effect, including induction of reactive oxygen species and promotion of inflammation.28–30 Such non-hemodynamic effects of angiotensin-II may have contributed to intracranial aneurysm formation in this model. In abdominal aortic aneurysms induced by angiotensin-II in apolipoprotein E (ApoE)-knockout mice, the non-hemodynamic effects appeared to be major causative factors. 31, 32 Further studies are needed to elucidate relative contributions from non-hemodynamic effects and hypertensive effects of angiotensin-II in this model.
Potential roles of proteinases, especially MMPs, in the formation of intracranial aneurysms have been suggested by observational, genetic and experimental studies.19–21, 22, Kim, 1997 #13247, 33 In our study, we detected increased MMP activity in intracranial aneurysm tissues, and the treatment with doxycycline, a broad-spectrum MMP inhibitor, dramatically reduced the incidence of intracranial aneurysms. Interestingly, the incidence of aneurysms was significantly reduced in MMP-9 knockout mice, but not in MMP-2 knockout mice. MMP-9 is known to be produced by inflammatory cells, especially macrophages, and its expression is up-regulated in vascular diseases including abdominal aortic aneurysms.34, 35 On the other hand, MMP-2 is constitutively expressed by fibroblasts and smooth muscle cells in vascular tissues. Hemodynamic stress induced by systemic hypertension can trigger an inflammatory process by activating endothelial and inflammatory cells and up-regulate leukocyte adhesion molecules.36 In addition, the degeneration of elastic lamina by elastase in this model can trigger vascular inflammation. MMP-9 produced by inflammatory cells may destabilize the vascular wall and facilitate excessive vascular remodeling, causing aneurysm formation. The constitutive nature of MMP-2 expression may suggest that MMP-2 is not critical for dynamic vascular remodeling processes that may be occurring during aneurysm formation. Alternatively, unknown developmental compensation may have masked effects of the lack of MMP-2 on intracranial aneurysm formation in this model.
It should be noted that the greater reduction of the incidence of aneurysms in doxycycline treated mice compared to MMP-9 knockout mice may suggest potential contributions from other MMPs to aneurysm formation in this model. In addition, doxycycline has other effects in addition to inhibiting MMPs, including modulations of various aspects of inflammation.37, 38 These additional effects of doxycycline may have diminished or masked its MMP inhibitory effects. Nevertheless, these data showed the feasibility of utilizing this model to study underlying mechanisms for intracranial aneurysm formation.
Nobuo Hashimoto et al. at Kyoto University pioneered the development of an elaborate intracranial aneurysm model that combines three surgical and pharmacological manipulations.5, 25, 39, 40 They combined (1) renovascular hypertension induced by ligation of posterior branches of the bilateral renal arteries and loading with high salt diet, (2) continuous administration of a lathrogen, beta aminoproprionitrile, and (3) unilateral ligation of common carotid artery. Animals in this model developed pre-aneurysmal changes or microscopic intracranial aneurysm formations at the olfactory artery-anterior cerebral artery bifurcation after three to four months. Not only did their model and our model share the histological characteristics of aneurysms, but both models also showed that pharmacological inhibitions of MMPs result in the reduction of intracranial aneurysm formation.24 While the initial events that lead to aneurysm formation in these two model may be different, both models may be sharing common downstream pathways, including MMP activation, that lead to the same end phenotype—aneurysm formation.
We established a new mouse model of intracranial aneurysm that recapitulates key characteristics of human intracranial aneurysms. In this model, the combination of hypertension and the degeneration of elastic lamina induced by a single stereotaxic injection of elastase into the cerebrospinal fluid resulted in large aneurysm formations with the incidence as high as 77%. The aneurysm formation in this model appeared to be dependent on MMP activity. This model can be used to study molecular mechanisms that lead to aneurysm formation and growth.
We wish to thank Dr. William L. Young for providing mentoring and insightful suggestion and Mr. Mark Weinstein for his skillful technical assistance.
This study was funded by NIH R01NS055876 (TH) and American Heart Association Grant-in-Aid 0755102Y (TH)
Conflict of interest