CNS-associated macrophages contribute to intracerebral aneurysm pathophysiology

The formation of IAs was induced by local microinjection of elastase behind the MCA bifurcation (Fig. 1a; Additional file 1: Figure 1a) followed by subcutaneous Angiotensin II (Ang II) osmotic mini-pump insertion (Fig. 1a; Additional file 1: Figure 1a). Local microinjection of elastase induced vascular wall degradation while subcutaneous infusion of Ang II increased blood pressure from day 3, reaching a plateau by day 10 post pump implantation (Additional file 1: Figure 1b, c). The dose of elastase was chosen after a dose–response experiment (Additional file 1: Figure 2a). During our model development phase, we performed experiments without pump insertion and still observed the presence of aneurysms, however with smaller formation and rupture rate (Additional file 1: Figure 2b–d).

The development of IAs in mice was determined by contrast-enhanced T1-weighted (CE-T1-w) MRI following the injection of gadolinium (DOTAREM) as a contrast agent at three different time points (Day 5, Day 10, and Day 15; Fig. 1a, b; Additional file 1: Figure 1e, f). IAs were counted by two different experimenters observing the intravascular accumulation of gadolinium within the MCA using the CE-T1w MR scans (Fig. 1b; Additional file 1: Figure 1g). The 3D reconstructions of IAs from the MR scans revealed saccular subtype aneurysms (Fig. 1c). The spontaneous rupture of the IAs resulted in either subarachnoid or intracerebral bleeding (Additional file 1: Figure 1d). Forty percent of IAs were already present at the Day 5 (Fig. 1d), and 62% were ruptured by Day 15 (Fig. 1e). At the end of the protocol, 93% of mice developed IAs (Fig. 1d), and the overall survival was 76% (Fig. 1f). The average number of IAs per mouse was estimated to be 2 and 3.57 as observed through MRI and histological analysis, respectively (Fig. 1g). Mice were mainly asymptomatic even after rupture, probably due to the specific location of the aneurysms, i.e., the outer cortical area, and the relatively small hemorrhage volumes in the case of rupture (Fig. 1h).

Histological analysis of the brain sections was performed using the elastica van Gieson staining to visualize connective tissue and elastic fibers (Fig. 2a, b). Analysis of the vessel wall (intima-media) thickness and wall-to-lumen ratio showed significant increases in both parameters in the IAs compared to the contralateral arteries (Fig. 2c, d), confirming the data observed by MRI. Furthermore, the internal elastic lamina (IEL) was mostly damaged, fragmented, or completely degraded in our samples of induced mouse IAs (Fig. 2b). The collagen layer appeared thickened as compared to the control vessels and was most accentuated in the advanced IAs in mice (Fig. 2b, e). A large number of infiltrating cells was observed in all of our IA samples, with their numbers increasing in line with the severity of IA advancement (Fig. 2e).

MR scans were used as anatomical reference to detect the IAs in the histological and immunofluorescent (IF) brain sections (Additional file 1: Figure 2e). IF analyses confirmed the prominent IA features, most noticeably increased vessel wall thickness (Fig. 2f) and a low signal of both laminin and collagen type IV (Col IV) (Fig. 2g, h). α-SMA, a marker of smooth muscles cells (SMCs), was mostly increased in the IA walls due to the proliferation of SMCs (Fig. 2i). However, in some samples, almost complete degradation of SMCs was found, suggesting progressed vessel wall degeneration (Fig. 2j). All aneurysms showed a damaged endothelial cell (EC) layer as manifested by a discontinuous CD31⁺ signal where no cell–cell junctions were clearly visible (Fig. 2k). The presence of a fibrinogen signal in the vessel wall of some of the IA samples suggested BBB disruption (Fig. 2l). Both P-selectin⁺ and VCAM-1⁺ signals were found at the luminal EC side in the IAs whereas they were mostly absent in the contralateral arteries (Fig. 2m, n). The IAs formed at the elastase injection site in mice (Figs. 1a–c, 2b, e) had a pathophysiological appearance similar to that observed in the humans through radiology and histology (Fig. 3a–i), in terms of increased thickness of the vessel wall, inflammatory cell infiltration, and vessel wall decellularization (Fig. 3b, f–i).

To determine a potential link between endothelial cell activation and IA rupture, we performed in vivo longitudinal molecular MRI after the injection of microparticles of iron oxide (MPIOs) coupled to P-Selectin or VCAM-1 antibodies at different time points during pathology progression (Fig. 4a; Additional file 1: Figures 3, 4). No P-Selectin or VCAM-1 MPIO signal voids were observed prior to the MPIO injection (Fig. 4b, c) at any of the time points. MPIOs were progressively cleared from the brain, resulting in negligible background 24h after each MPIO-injection, allowing longitudinal studies by repetitive assessment of endothelial activation. Three days after IA induction, positive signal voids were observed specifically at the IA area for both MPIO-αP-selectin (Fig. 4b, d, f) and MPIO-αVCAM-1(Fig. 4c, e, f), with latter also partly detected at a relative distance from the IA site (Fig. 4c, e, f). Interestingly, MPIO signal voids found at Day 3 correlated with the hemorrhage volumes after IA ruptures observed at Day 5 (Fig. 4g–i), with an R² value of 0.91 for MPIO-αP-selectin (Fig. 4g) and 0.85 for MPIO-αVCAM-1 (Fig. 4h). Furthermore, the MPIO-αP-selectin signal present at earlier time point (Day 3) was situated at the site of IA formation at the subsequent time point (Day 5; Fig. 4j).

To assess the similarity in pathophysiological appearance between IAs in our mouse model and our human samples (Figs. 2, 3), we verified whether the number of CAMs found nearby and within the IA wall was similar both in human and mouse samples by using classical histological markers for CAMs such as CD163 (in human samples) or CD206 (in mouse samples), CD11b, CD68 and HLA. The human IA samples showed an accumulation of CD163⁺ and CD11b⁺ cells (Fig. 5a, b), as well as presence of HLA-DR⁺ myeloid cells (Fig. 5c) and CD68+ cells (Fig. 5d) within the IA wall. In addition, some of the CD163⁺ cells colocalized with CD11b⁺ (Fig. 5e).

Similar results were observed in the mouse IA samples, where a high accumulation of CD206⁺ cells was observed nearby and in the IA walls as early as day 5 of the IA development stage (Fig. 6a, b). In the contralateral hemisphere, CD206⁺ cells were found exclusively in the perivascular spaces (corresponding to PVMs) and the meninges (corresponding to MMs) (Fig. 6a). In contrast, in the IA area, CD206+ cells were highly accumulated (Fig. 6b) with a subset showing a weak positivity for the lysosomal marker CD68 (Fig. 6c, d). Taken together, these data indicate that CD163⁺ (humans)/CD206⁺ (mouse) macrophages proliferate and/or infiltrate the IA site in both humans and mice.

We also identified other blood-borne immune cell subtypes in the IA site at a late stage of the IA formation (15 days post IA induction surgery), including CD3⁺ T-cells (Fig. 7a, b) and Ly6G⁺ cells (neutrophils; Fig. 7c, d). Ly6G+ cells expressed myeloperoxidase (MPO) and citrullinated histone H3 (H3Cit), demonstrating the presence of neutrophil extracellular traps (NETs; Fig. 7e, f). Indeed, 85.62% and 87.36% of Ly6G⁺ cells were positive for MPO and H3Cit, respectively (Fig. 7g, h).

Microglial activity changes were determined both morphologically and phenotypically. Quantitative analysis revealed a significantly higher number of Iba1⁺/P2Y12R⁺ cells at the ipsilateral site compared to the contralateral side (Fig. 7i, j). Phenotypically, the expression of P2Y12R was decreased in areas close to the IA, whereas the expression of Iba1 was increased (Fig. 7j). CD68, a lysosomal marker associated with microglial phagocytic activity, was also highly expressed in the microglial cells in the ipsilateral hemisphere (Fig. 7j). Microglial response to IA was also confirmed with an unbiased automated analysis of microglial morphology, which revealed increased sphericity of the soma and a decrease in the number of processes, and filopodia at the IA site, as opposed to microglia assessed around the corresponding arteries in the contralateral hemisphere. (Fig. 7k, l).

Given the finding of CAM accumulation in the IA vessel walls (Fig. 6), we depleted CAMs by injecting clodronate (CLO) liposomes in the lateral ventricle 5 days before IA induction surgery to elucidate their role in IA pathophysiology (Fig. 8a). We observed a specific depletion of CAMs (CD11b⁺ CD45⁺ CD206⁺ cells) 10 days after the CLO injection (Fig. 8b, c), which did not affect microglia numbers (CD11b⁺ CD45low; Fig. 8d). As previously reported [11], clodronate liposomes reduced at least 70% of the total number of CAMs compared to the vehicle (Fig. 8b).

Longitudinal MRI was conducted on both PBS- (control) and CLO-injected mice to assess IA formation and hemorrhage volumes at day 5, day 10, and day 15 post-surgery (Fig. 8a). CAM-depleted mice developed fewer IAs compared to the control group, as indicated by both MRI and immunofluorescence analyses (Fig. 8e). Additionally, CAM-depleted mice had a significantly lower rupture rate (Fig. 8f, g) and significantly reduced hemorrhage volumes in cases of IA rupture (Fig. 8h) compared to the control mice. These effects did not appear to be due to differences in the size and morphology of aneurysms, as the intima-media thickness (Fig. 8i) and wall-to-lumen ratio (Fig. 8j) were similar in both groups. Interestingly, CAM-depleted mice not only exhibited a significantly lower accumulation of CD206⁺ cells (Fig. 9a–c) but also had significantly fewer neutrophils (Fig. 9d–f), compared to the control (PBS-treated) mice. Indeed, CAMs are in an ideal anatomical position to contribute to focal chemokine gradients, which may either be transported to the luminal surface of endothelial cells and/or act directly as a migratory signal to peripheral immune cells including neutrophils.

Formation of IAs was induced by local microinjection of elastase (1µL, 35 IU) behind the MCA bifurcation (Fig. 1a; Additional file 1: Figure 1a) and immediately followed by subcutaneous Angiotensin II (Ang II) osmotic pump (800 µg in 90 µL of saline solution; diffusion rate 37 ng/min/14 days) insertion (Fig. 1a; Additional file 1: Figure 1a). This model is a variation of the model previously described by Nuki et al. [37], inspired by models developed for abdominal aortic aneurysm studies [2]. The dose of elastase was chosen after a dose–response test where different parameters were studied (i.e., IA formation, occurrence of hemorrhage, survival rate etc.) (Additional file 1: Figure 2a).

To avoid including animals with potential surgically induced hemorrhages in the study, all animals were thoroughly assessed 24 h post-surgery by using the specific T2*-w sequence to visualize potential hemorrhages. Mice presenting spontaneous postoperative hemorrhages were excluded from further studies (Fig. 1a). Signal void appearing at the later stages (after aneurysm development) in T2*-w were considered aneurysm ruptures (Additional file 1: Figure 1e–h). To confirm that the vessel wall degradation was caused by the elastase micro-injection and was not due to potential micropipette-induced vascular damage, a control experiment was performed by injecting PBS (n = 5) or elastase (35 mU/µL) (n = 8). None of the PBS-injected mice developed aneurysms, and no hemorrhage was observed at any time throughout the protocol (Additional file 1: Figure 1i).

The studied aneurysm sample was resected by Dr. Frösen at the Kuopio University Hospital, following clipping of the aneurysm neck. The sample was from a large middle cerebral artery aneurysm with a diameter of 2cm. A summary of the preoperative imaging studies is presented in Fig. 3. An informed consent was obtained from the patient prior to the procedure, and the study was approved by the Ethical committee of the Hospital District of Northern Savo (TKU 52 2014). Following resection, the tissue sample was fixed overnight in 4% paraformaldehyde and embedded in paraffin so that the original shape of the aneurysm dome was preserved.

The resected human MCA aneurysm sample was cut into 4 µm thick sections with a microtome, following which these sections were deparaffinized and rehydrated using a standard alcohol series. After rehydration, some of the sections were stained with standard hematoxylin–eosin staining. For those sections that were immunostained, an antigen retrieval procedure was performed by incubating the sections in close to boiling temperature in HIER buffer (ThermoFisher Scientific, Watham, MA, USA) for 2 × 5min. After that, a protein block in 3% normal horse serum (Biowest, Nuaillé, France) in 0.1% PBST for 30 min at room temperature was performed. Following the antigen retrieval and serum block, the sections were incubated overnight at 4 °C with the primary antibody (Table 1) diluted in 1.5% normal horse serum (Biowest) in 0.1% PBST. Following this, after 3 × 5 min washes in PBS, the sections were incubated with the secondary antibody (Vector Laboratories, Burlingame, CA) (dilution 1:200) in 1.5% normal horse serum in 0.1% PBST for 30 min at room temperature. Secondary antibodies were chosen according to the species where primary antibodies were produced. After the incubation with the secondary antibody, the sections were washed for 3 × 5 min in PBS and underwent an endogenous peroxidase block with 3% hydrogen peroxide (ThermoFisher Scientific) in PBS for 20 min at room temperature. This was followed by 3 × 5 min PBS wash, following which the sections were incubated with a horseradish peroxidase conjugated avidin–biotin complex (Vector Laboratories) for 30min in room temperature. After subsequent 3 × 5min PBS washes, the positive signal was detected with DAB (3′–5′-diaminobenzidine) (Vector Laboratories). Hematoxylin was used for counter staining and the stained sections dehydrated and mounted with Depex.

Double immunofluorescence staining was performed to confirm the co-localization of markers CD11b and CD163. Protocol was similar to the one described above, except that normal goat serum (Vector Laboratories) in 0.1% PBST was used for protein blocking and antibody dilutions. Secondary antibodies were AlexaFluor 488 and 594 conjugated (dilution 1:200, ThermoFisher Scientific) for signal detection. Mounting was performed with Vectashield medium with DAPI (Vector Laboratories).

The sections chosen for the staining of human samples were adjacent sections from different depths of the IA, making it possible to compare co-localization of different inflammation markers. Immunoperoxidase stained sections were imaged with Olympus AX70 microscope (Olympus, Japan). For immunofluorescence staining, imaging was performed with LSM800 Zeiss confocal microscope system (Carl Zeiss Ag, Oberkochen, Germany) with 405/488/555 nm diode lasers and appropriate emission filters (20x/0.5 PlanApo objectives, 1024 × 1024 frame sizes). Image processing was performed by ImageJ (Rasband, W.S., National Institutes of Health, Bethesda, Maryland).

8 weeks old male Swiss (Janvier Labs) mice were housed (Centre Universitaire de Ressources Biologiques, Normandy University, Caen, France) at 21 °C in a 12 h light/dark cycle with food and water ad libitum. Animal care and manipulations complied with recommendations issued by the French and European guidelines for the care and use of laboratory animals (European directive 2010/63/UE) and were authorized by the ethical committee (authorization no. 27499). All experiments were performed following the ARRIVE guidelines (www.​nc3rs.​org.​uk). All the procedures needing anesthesia were performed by an initial exposure to 5% isoflurane followed by a maintaining phase of 1.5–2% isoflurane 30% O2/70% N₂O during experiments.

Depletion of CAMs was carried out 5 days before the MCA aneurysm induction surgery to minimize the pro-inflammatory effects of CLO per se, by injecting CLO intracerebroventriculary. Anesthetized mice were placed in a stereotaxic device, and after the skin was removed, a small craniotomy was performed (coordinates: − 0.2 mm anteroposterior; + 1 mm lateral; − 2 mm depth from the Bregma). With a glass micropipette, 10 µl PBS-liposomes (PBS group) or clodronate-encapsulated liposomes (CLO group) (purchased from clodronateliposomes.com) were gently injected into the left lateral ventricle for 20 min. CLO liposomes injected into the cerebral ventricles are transported by the cerebrospinal fluid to the perivascular spaces and phagocytized by CAMs. In the cytosol, CLO acts as a cytotoxic ATP analog, which impairs mitochondrial oxygen consumption and leads to cell death by apoptosis [29].

Blood pressure in mice was measured using the Visitech BP-2000 blood pressure analysis system to ensure accurate physiological data while minimizing stress on the animals. Measurements, including arterial blood pressure (systolic, diastolic, mean) and heart rate, were taken every third day using a volume pressure recording sensor and an occlusion tail-cuff system. The experimental setup involved activating the BP-2000 computer and control unit, setting the heating platform to 34 °C, and placing mice in a magnetic restraining device with their tails free. Mouse was gently placed into an integrated animal holder to ensure proper restraint during the procedure. The tail was inserted through the corresponding tail-cuff, taped down, and covered with an LED light sensor. During the analysis cycle, the blood pressure control unit pressurized and attempted to read a pulse, producing results at the end of each measurement cycle. Recording sessions, integrated with data acquisition, were repeated until experimental variability was minimized, typically requiring three sessions.

Brain samples for IHC were isolated 5 and 15 days after the MCA aneurysm surgery. Terminally anesthetized mice were transcardially perfused with cold heparinized saline (15 mL/min) and fixed with 50 mL of 4% paraformaldehyde phosphate buffer (pH 7.4). Brains were post-fixed with 4% paraformaldehyde phosphate buffer (18 h; 4 °C) and cryoprotected (sucrose 20% in PBS; 24 h; 4 °C) before freezing in CryomatrixTM medium, frozen in isopentane cooled with liquid nitrogen, and stored at − 80 °C.

For IF analysis 10 μm thick cryostat-cut sections were adhered and collected on poly-lysine coated microscope slides and stored at − 80 °C before processing. Sections were randomly selected and incubated overnight at room temperature, with primary antibodies listed in Table 1. Primary antibody solutions were prepared with PBS-Triton (25% v/v). To reveal primary antibodies, after rinsing three times in PBS (1x, pH = 7.4), brain sections were incubated with appropriate Donkey Fab’2 fragments conjugated with Alexa Fluor® 488, Alexa Fluor® 594 or Alexa Fluor® 647 (1:1000, Jackson ImmunoResearch, West Grove, USA).

After another set of rinsing in PBS (1x, pH = 7.4) and deionized water, slides were left to dry. Finally, sections were coverslipped with an antifade medium containing DAPI. Images were captured using Leica DM6000 epifluorescence microscope coupled camera and visualized with Leica MM AF 2.2.0. software (Molecular Devices, USA). Additionally, confocal images were taken on Leica TCS SP5 MP microscope using the 40X oil immersive objectives. Images were taken using LAS AF Leica Software.

Following transcardial perfusion with 4% PFA, 100 µm thick brain sections were cut with vibratome and stained for microglia (Iba1, Ancam Ab5076) and cell nuclei (DAPI, Sigma 10236276001). Confocal stacks with 0.2 µm/pixel size and Z-step of 0.4 µm were acquired using Nikon A1R confocal microscope with 60 × water immersion objective (Plan Apo VC NA = 1.2 WD = 0.31–0.28 mm FOV = 215.04 µm; Nikon Instruments Europe B. V., Amsterdam, the Netherlands). Images were acquired both on the ipsi- and contralateral side of the aneurysm. Stacks were then analysed using the Microglia Morphology Quantification Tool [20]. This automated analysis provides an unbiased, 3D analysis of microglia morphology.

Imaging was performed on a Pharmascan 7T/12 cm system using surface coils (Bruker, Germany). Mice were deeply anesthetized with 5% isoflurane and maintained with 1.5–2% isoflurane 30%O₂/70%N₂O during the acquisitions. T2-weighted images were acquired using a multislice sequence: TE/TR 33/2500 ms. Hemorrhage volumes were quantified using the ITK software. Mice showing lesions < 6mm3 at 24 h post-surgery were considered as surgical failure and excluded from the analyses.

For molecular MRI (see below) three-dimensional T2*-weighted gradient echo imaging with flow compensation (field-of-view (FOV) 70 × 70 × 70 mm³ interpolated to an isotropic resolution of 70 mm), TE/TR 13.2 ms/200 ms and a flip angle of 21° was performed to visualize MPIOs. T2*-weighted sequences were used to verify if IAs ruptured.

Micro-sized particles of iron oxide (MPIOs) (1.08 μm diameter; Dynabeads MyOne Tosyl Activated, Invitrogen) covalently conjugated to purified polyclonal goat anti-mouse antibodies for P-selectin (R&D Systems, clone AF737) and purified polyclonal mouse anti-rat antibodies for VCAM-1 (Clone MR106, BD Pharmingen) were prepared as previously described [32]. In brief, MPIOs were covalently conjugated to polyclonal antibodies in borate buffer (pH 9.5) at 37 °C for 48 h. The coating of the beads was made using 20 μg of VCAM-1 or 40 μg of P-selectin antibody per 1 mg of reactive MPIOs. After, beads were washed and incubated with PBS containing 0.5% BSA for 24 h at room temperature to block remaining active groups. MPIO solution is stored under continuous rolling at 4 °C until usage.

MRI acquisitions started immediately after the intravenous injection of MPIO-solution (200 µl of 2 mg Fe/kg of conjugated MPIOs). All T2*-weighted images presented are minimum intensity projections of six consecutive slices. The qualitative analysis and the distribution of the MPIO signal voids on 3D T2*-weighted images was assessed by ITK Snap Software [49].

Semi-automatic signal void quantification was performed similar to previously described [15]. Briefly, signal voids corresponding to MPIO presence were quantified in subject space within region-of-interest (ROI) masks of the aneurysm lesion and the ipsilateral hemisphere and their contralateral homologues. In MATLAB, histograms of voxel signal intensities in each T2*-weighted image in contralateral ROIs were generated and fitted to a Gaussian of certain peak center and peak width. Voxels in the brain with signal intensities lower than the value at the contralateral peak center minus its half width (i.e., signal intensity values lower than 6.25% of the value of the majority of contralateral reference voxels) were subsequently labeled as signal void. Per time point, signal voids were counted per ROI and expressed as a percent-difference of the count in the corresponding contralateral ROI.

Fc receptors were blocked with CD16/32 (553142, BD Biosciences) for 10 min at 4 °C before incubation with the primary antibodies. Cells were stained with antibodies directed against CD11b (M1/70, BioLegend), CD45 (30-F11, BD Biosciences), Ly6G (1A8, BD Biosciences), CD3e (145-2C11, BD Biosciences) and CD206 (C068C2, BioLegend) for 45 min at 4 °C. After washing, samples were analyzed by a FACSVerse flow cytometer or sorted by a FACSAria (BD Biosciences). Appropriate isotype control antibodies were used to establish sorting parameters. Data were analyzed with the FlowJo 7.6.5 software (TreeStar Inc.). Data are expressed as percentages.

For histological examination of brain tissue sections and visualization of connective tissue made up of elastic fibers Elastica van Gieson staining kit was used (Merck, Darmstadt, Germany). Slides with 10 μm thick brain sections were incubated for 13 min in resorcin fuchsin solution and then for 5 min in alcoholic hematoxylin solution and hydrochloric acid iron(III)nitrate solution in ratio 1:1. After each incubation slides were rinsed under running tap water for 1 min. Slides were then incubated in pirofuchsin solution (picric acid/acid fuchsin solution) for 2 min and rinsed in 70%, 96% and 100% ethanol (two times for 1 min in each percentage). Finally, slides were incubated in toluene for 10 min and coverslipped with mounting medium. Images were captured using Leica DM6000 epifluorescence microscope coupled camera and visualized with LAS X Leica Software (Molecular Devices, USA).

All images were analyzed using ImageJ 1.51k software. To determine the intima-media thickness and wall-to-lumen ratio, first, the diameters of both vessel and vessel lumen were measured. Intima-media thickness was calculated as a half value of the vessel diameter subtracted by the diameter of the lumen. The results are expressed in μm. To calculate wall-to-lumen ratio, the value of intima-media thickness was divided by the lumen diameter.

CD206⁺, Ly6G⁺, and CD3⁺ cells were quantified in a nonrestricted area around the aneurysms in the ipsilateral hemisphere and control arteries in the contralateral hemisphere. DAPI staining for cell nuclei was used to confirm that signal comes from the cell. Results for each cell type are presented as a number of cells around the aneurysm and contralateral artery. Microglial activation and proliferation were analyzed on images taken using 20X objective of Leica DM6000 epifluorescence microscope coupled camera, in an area of 400 × 400 μm. All quantifications were done manually, blinded to the treatment.

Images presented in the in Fig. 1; Figs. 2 and 3 and Additional file 1: Fig. 1 were adapted from BioRender with postprocessing using PowerPoint.