Increased Intravascular Flow Rate Triggers Cerebral Arteriogenesis

Abstract

Peripheral arteriogenesis is distinctly enhanced by increased fluid shear stress. Thus, the aim of this study was to investigate in the rat brain whether increased fluid shear stress can also stimulate cerebral arteriogenesis. To increase fluid shear stress in the cerebral circulation, we developed different shear stress models as the ligature of both common carotid arteries (Double-Ligature model), bilateral carotid ligature followed by creation of a unilateral arterio-venous fistula (two-stage protocol, Ligature-Shunt model), and unilateral arterio-venous fistula-creation alone (Solo-Shuntmodel). Blood flow changes were monitored in vivo by quantitative magnetic resonance imaging-analysis. Cerebral arteriogenesis was analyzed by magnetic resonance imaging and contrast agent-angiography. For proliferation and accumulation of mononuclear cells, immunohistochemistry was performed. During the 14 days-observation period, blood flow increased maximal by 5.5-fold in the A. basilaris and 10.3-fold in the fistula-sided A. cerebri posterior of the Ligature-Shunt model. Considerable vessel growth was found in all shear stress-stimulated arteries. Comparative analysis of vessel length and diameter versus blood flow indicated a correlation between the growth of cerebral collaterals and rising intravascular flow rates (R² = 0.90/0.96). Immunohistochemistry showed the typical phases of arteriogenesis and accumulation of mononuclear cells. In conclusion, we provide evidence that fluid shear stress is not only the pivotal trigger of peripheral but also of cerebral arteriogenesis.

Introduction

Besides cardiovascular diseases and cancer, stroke is the third leading cause of death in industrialized countries and frequently produces disability in patients. Clinical studies demonstrated the importance of an adequate hemodynamic compensation via the circle of Willis in patients with high-grade internal carotid artery stenosis or occlusion (Bisschops et al, 2003; Hendrikse et al, 2001, 2002).

The term arteriogenesis was established by Schaper and his colleagues (Arras et al, 1998) to discriminate between the growth of collateral arteries and angiogenesis—the development of small-calibre capillary vessels in ischemic tissue. However, arteriogenesis describes the remodeling of preexisting arterio-arteriolar anastomoses or arteries outside the ischemic region to functional vessels, which can replace the conductance capacity of a larger artery (Grundmann et al, 2007; Schaper and Scholz, 2003; Scholz et al, 2001).

However, it was shown in animal models that peripheral collateral growth reaches only approximately 40% to 50% of the maximal conductance of the replaced artery (Hoefer et al, 2001; Ito et al, 1997). Our group demonstrated in previous studies in a pig and rabbit hindlimb model that fluid shear stress (FSS) is the pivotal trigger of peripheral arteriogenesis and that long-term FSS-stimulated collaterals can completely restore and even surpass the physiologic function of the occluded artery (Eitenmuller et al, 2006; Pipp et al, 2004).

To chronically elevate FSS after femoral artery ligature, collateral flow was directly drained into the venous system by creating a side-to-side anastomosis [arterio-venous (AV) fistula] between the distal stump of the occluded femoral artery and the accompanying vein. The direct connection of the collateral system to the low venous pressure system provided several beneficial consequences as the minimization of systemic flow resistance and the prevention of the dropping of FSS during later phases of arteriogenesis (Eitenmuller et al, 2006; Pipp et al, 2004).

The morphology of the cerebral arteries is very similar between humans and rats, including anomalies and structure of these vessels as well as their morphologic changes associated with cerebral vascular disease (Lee, 1995). It was demonstrated by Busch et al, 2003 that arteriogenesis can be induced in the adult rat brain. By developing the three-vessel occlusion (3-VO) model (occlusion of one carotid artery and both vertebral arteries), they could show a significant enlargement of the ipsilateral posterior cerebral artery causing a significant reduction of experimentally induced stroke volume (Busch et al, 2003, 2006; Buschmann et al, 2003; Schneeloch et al, 2004).

On the basis of our previous AV fistula-experiments in the rabbit hindlimb, we established several models to increase FSS in the rat cerebral collateral circulation. Thus, the aim of this study was to determine whether increased FSS is also the pivotal trigger of cerebral arteriogenesis.

We evaluated the effects of shear stress stimulation on the cerebrovascular circulation in vivo by magnetic resonance imaging (MRI)-phase contrast angiography as well as postmortem by contrast agent-angiography, and correlated the results to the quantitative MRI-based measurement of the cerebrovascular blood flow. Proliferation of vascular cells as well as accumulation of mononuclear cells was monitored by immunohistochemistry.

Materials and methods

Animal Models

The present study was performed with the permission of the State of Hessen, according to Section 8 of the German Law for the Protection of Animals. The investigation conforms to the Guide of Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996).

All surgical procedures were performed under anesthesia with ketamine (100 mg/kg body weight) and xylazine (4 mg/kg) administrated i.p., and buprenorphine (0.03 mg/kg) was applied to prevent pain.

Male Sprague–Dawley rats (250–300 g; Charles River Laboratories, Sulzfeld, Germany) were randomly assigned to Sham, Double-Ligature, Ligature-Shunt, and Solo-Shunt group (n = 19 each). In the Double-Ligature group, both common carotid arteries were ligated by 4–0 USP nonabsorbable sutures. In the Ligature-Shunt group, rats were allowed to recover for 48 h after bilateral ligature of the common carotid arteries. Subsequently, the left common carotid artery was cut immediately distally to the ligature and its distal stump was anastomosed end-to-side with the adjacent jugular vein, thereby creating an AV fistula using 10–0 USP sutures. In the Solo-Shunt group, ligature of the left common carotid artery and AV fistula-creation (as described above) were performed in one session leaving the right carotid artery unaffected. Central arterial pressure was determined in the nonligated carotid artery, central venous pressure in the jugular vein, and carotid backpressure in the cut distal stump of the ligated carotid artery before AV fistula-creation. Animals that had developed cerebral oedema or AV fistula-reocclusion in the postoperative MRI-control (five animals of the Ligature-Shunt group) were excluded from the study. Control animals were sham-operated by dissecting the two carotid arteries without any further surgical treatments. After surgery, animals were returned to their cages with free access to food and water and were allowed to move freely.

Magnetic resonance imaging analysis and quantitative MRI-flow measurement were performed 1 day (n = 10), 7 days (n = 10), and 14 days after surgery (n = 5). After MRI-analysis at experimental day 7 and day 14, rats were euthanized by bleeding after an overdose of ketamine and postmortem angiograms were performed (n = 5 each). Immunohistochemical evaluation was conducted 1, 3, and 7 days after surgery (n = 3 each at every point of time).

Magnetic Resonance Imaging-analysis and Quantitative Magnetic Resonance Imaging-flow Measurement

One, 7, and 14 days after surgery, rats were placed in an MRI-scanner (Bruker PharmaScan 7.0T, 16 cm; Bruker Corporation, Ettlingen, Germany; n = 10 each at 1 and 7 days and n = 5 each at 14 days). Respiratory rate was monitored during the imaging protocol. Isoflurane concentration was varied between 2.0% and 3.0% to keep the respiratory rate between 60 and 80/mins. Control of central arterial pressure during isoflurane anesthesia showed constant values of 100 to 110mmHg. Body temperature was maintained at 37°C via a feedback-regulated hot-water spool.

The magnetic resonance tomograph operates at 300.51 MHz for ¹H imaging and is equipped with a 300-mT/m self-shielding gradient system. Thereby, a linear polarized volume resonator (diameter, 35mm) was used. Localizer images were acquired using a spin-echo sequence and correction of slice angulation was performed, if necessary. Rapid acquisition with relaxation enhancement sequences in axial and sagittal slice orientation were used to determine the exact positioning of the rat brain [20 contiguous slices, 1-mm thickness, repetition time (TR) = 2500 ms, time to echo (TE) = 36.7 ms]. Afterwards, a coronal three-dimensional phase contrast angiogram of the whole rat brain was performed to detect the intracerebral, vascular system [TR = 35 ms, TE = 3.9 ms, field of view (FOV) = 70 × 37mm, slab-thickness = 15.00mm, matrix size = 256 × 200 × 64, velocity encoding (venc) = 30 cm/sec]. Three-dimensional reconstruction [maximum intensity projection (MIP)] was made with the use of the Image Processing Tool of the Paravision 4.1 software (Bruker PharmaScan).

The slice orientation of the following flow-quantification sequence was set strictly perpendicular to the target vessel. Parameters of the flow-quantification sequence were TR = 20 ms, TE = 3.5/3.9 ms, flip angle = 30°, venc = 50 to 150 cm/sec (depending on the target vessel). For sagittal orientation, the remaining parameters were NEX= 12, FOV= 70 × 35mm, slab-thickness = 8.06mm, matrix size = 512 × 256 × 16, for axial orientation NEX= 17, FOV= 33 × 35mm, slab-thickness = 5.54mm, matrix size = 257 × 256 × 10, resulting in an in plane resolution of approximately 130 × 130 µm². Thus, a vessel with a diameter of, for example, 400 µm is covered by 7 to 8 pixels. For estimation of the significance of the partial volume effects on the accuracy of the flow measurement, the following function has to be considered:

$${\phi }_t=\text{arctan}\left(\frac{\text{sin}(\phi v)\cdot (1-x)}{\text{cos}(\phi v)\cdot (1-x)+x}\right)$$

(1)

with φ_v, the signal phase originating from the velocity encoding of the moving spins; x, the relative signal from the unmoved spins; and φ^t, the calculated signal phase of the magnetization in the whole voxel. It calculates the total signal phase in dependency of the phase of the moving spins and the relative signal amount from unmoved (x) and moved (1–x) spins. This function is nearly linear in the range of x = 0.0–1, at least for ${\phi }_v⩽\pi /2$ . That is at least for half the maximal encoding velocity. Therefore, the calculated overall velocity of a voxel with moved and unmoved spins is inversely proportional to the volume containing unmoved spins. At least at the vessel border, where partial volume effects are important and the velocity is relatively low, partial volume effects do not affect the accuracy of the flow measurement significantly. Naturally, this does not apply for the average velocity or vessel area alone.

In Paravision 4.1, the slice positioning on projected images is not possible. Owing to the small size of the target vessels and the resulting difficulties in identifying the vessels in the native images, a slice positioning using the native images was not possible in a reliable way. To keep the measurement time in a reasonable range, the number of slices in the following flow measurement was relatively low (16 or 10, respectively) with a slice thickness of approximately 500 µm. Because the covered range of the rat brain was rather small and the density of small vessels in the considered area relatively high, it was difficult to identify the target vessels on the images of the flow map. Therefore, it was necessary to calculate the physical coordinates in the magnet coordinate system from the coordinates of selected points in the MIP of the angiogram.

The physical coordinates were used for the calculation of the slice orientation and position of the flow measurement sequence. Three points in the MIP of the angiogram were selected, which were situated in maximal possible distance of each other on a straight part of the vessel. The connecting line of the two outer points defined the normal of surface of the new slices and, consequently, the new slice angles, the included third point, the position. The in plane shift was calculated for centring the third point in the new image plane.

For the determination of all three coordinates of a selected image point in the MIP of the angiogram, we got an algorithm from Bruker that not only calculates the MIP but also saves the slice number of the image point with maximal signal used for the construction of the projection. This algorithm functions only perpendicular to the native slices, inhibiting the typical rotation of MIP images but was sufficient for the examined morphology. The calculation of the physical coordinates from the image points and the sequence parameters as well as the calculation of the parameters of the new slices were performed with a program written by one of the authors.

The quantitative flow values were then obtained by integrating across manually drawn regions of interest (ROI) that enclose the vessels (Paravision 4.1 ROI-Tool, Bruker PharmaScan). Blood flow was measured in the left A. cerebri anterior, both A. cerebri posteriores, A. basilaris, left A. carotis interna (to determine flow of the AV fistula) and the right A. carotis interna (only in the Solo-Shunt group) (Figure 2B).

Figure 2

Verification of cerebral integrity and MRI-based blood flow measurement. (A) Representative angiogram (Double-Ligature model), spin-echo T2- and diffusion-weighted MRI 1 day after bilateral ligature of the common carotid arteries. Images showed the absence of both common carotid arteries after ligature on the angiogram, but they did not show any hyperintensity as visualization of a cytotoxic oedema on the spin-echo T2- and diffusion-weighted sequence, proving cerebral integrity. (B) MRI-phase contrast angiography of a sham-treated rat (Control) demonstrating the intracerebral measuring points for MRI-based blood flow measurement. (C) Quantitative MRI-flow measurement of the A. basilaris and Aa. cerebri posteriores showed, compared with the preoperative flow, a significant increase of blood flow 1, 7, and 14 days after indicated treatment (P<0.05). Note the significant flow differences between the different treatment options ‘AV fistula—ligature—sham’ in the Aa. cerebri posteriores of the Ligature-Shunt and Solo-Shunt group. (preop: preoperative; values expressed as mean±s.e.m.; n=10 each at 1 and 7 days and n=5 each at 14 days).

Cerebral integrity was verified 1 day after bilateral carotid ligature or AV fistula-creation. For detection of oedema, a spin-echo Proton- and T2-weighted sequence was used (16 contiguous slices, 2-mm thickness, FOV= 37 × 37mm, matrix size 512 × 256, TR = 3000 ms, effective echo times TE1 = 25.7 ms and TE2 = 77.1 ms, two echoes for proton and four echoes for T2-weighting with an echo spacing of ΔTE = 17.1 ms, imaging time = 12.48 mins). To detect ischemic regions, a diffusion-weighted echo planar imaging sequence was performed (6 contiguous slices, 2-mm thickness, FOV= 32 × 32mm, matrix size 128 × 128, TR = 5000 ms, TE = 46.3 ms, imaging time = 5.3 - mins, diffusion gradient duration δ = 7 ms, diffusion gradient separation Δ = 14 ms, b-value = 1500 s/mm²).

Cerebral Postmortem Angiograms

Cerebrovascular anatomy was studied 7 and 14 days after surgery in each study group (n = 5 at every point of time). After maximum adenosine (Sigma-Aldrich, Taufkirchen, Germany) induced vasodilatation, cerebrovascular circulation of euthanized rats was perfused with 37°C-warm, gelatine-barium-based contrast medium according to the formula developed by Fulton (Fulton, 1965) and colored by bromophenol blue (Sigma-Aldrich). Injection of contrast medium was performed through the right carotid artery (contralateral to the fistula-side) at a constant, physiologic pressure of 80mmHg. Right and left atrium was incised to facilitate outflow. To harden gelatine, animals were placed on crushed ice for 20 mins. Brains were carefully removed, and pictures were taken using a Carl Zeiss-microscope (OPMI 1 FR) with a digital camera. Vessel length was measured between stem and re-entry, external diameters at three different levels using ImageJ 1.38p software (http://rsb.info.nih.gov/ij/).

The vasodilator effects of adenosine and adenine nucleotides were first recognized by Drury and Szent-Gyorgyi in 1929. Since then, many studies demonstrated that adenosine has the capability to dilate many vascular beds including heart, skeletal muscle, and brain. Although it was shown for the brain that the intraluminal application of adenosine is less effective than the extraluminal application, the intracarotid adenosine infusion proved as an effective tool for a robust and profound augmentation of cerebral blood flow without any adverse hemodynamic side effects. Increased adenosine levels in the brain were shown to be caused by hypoxia or ischemia, but there were no alterations in the adenosine concentration found when the arterial pressure was manipulated over the normal autoregulatory range in the absence of hypoxia (Collis, 1989; Gordon et al, 2008; Joshi et al, 2002; Miekisiak et al, 2008). Because animals, which had developed cerebral oedema caused by hypoxia or ischemia, were excluded from the study, adenosine-induced vasodilatation and injection of the contrast medium at a constant physiologic pressure ensured demonstration of the correct anatomical vessel size for each of the models.

Immunohistochemistry

Three rats of each study group were euthanized 1, 3, and 7 days after surgery without previous adenosine-dilation. Brains were frozen in 2-methylbutan (precooled in liquid nitrogen) and immunostaining of the A. basilaris was performed on 5 µm-cryosections as previously described (Rodriguez et al, 2005). Primary antibodies: Ki67/MIB-5 (1:20, Dako, Hamburg, Germany) and CD11b (1:100, AbD Serotec, Düsseldorf, Germany). Secondary antibody: Cy3-conjugated anti-mouse IgG (1:300, Chemicon/Millipore, Schwalbach, Germany). Actin was stained using Phalloidin FITC-labeled (Sigma-Aldrich) and nuclei were stained with Draq5 (Alexis Biochemicals, Lörrach, Germany). Sections were viewed with a confocal scanning laser microscope (Leica TCS SP). Confocal images were obtained using concomitant multichannel scanning. Each recorded image was taken using two confocal detectors for reflected fluorescence and consisted of 1024 × 1024 pixels. Series of eight confocal optical sections were taken through the depth of the tissue sample. To improve the image quality and to obtain a high signal/noise ratio, each image from the series was two-fold signal-averaged. After data acquisition, the images were transferred to a Silicon Graphics Indy or Octane workstations (Silicon, Grabsbrunn, Germany) for image restoration and reconstruction using Imaris, the multichannel image processing software (Bitplane, Zürich, Switzerland). The principles of this method have been previously described (Kostin et al, 2004; Polyakova et al, 2008). For determination of the proliferation index in growing basilar arteries Ki67-positive nuclei of endothelial, medial, and adventitial cells were counted and referred to the total number of nuclei of each particular vessel layer.

Statistical Analysis

All values are expressed as mean±s.e.m. D'Agostino and Pearson omnibus normality test (Prism, GraphPad Software, Inc.) was used to test normal distribution. Group differences were analyzed for statistical significance using one-way ANOVA (Prism, GraphPad Software, Inc.). Post hoc comparisons were performed using Bonferroni correction. Probability values less than 0.05 were considered as statistically significant.

Results

Novel Rat Models were Established to Increase Fluid Shear Stress in the Cerebral Circulation

Under conditions of laminar flow, FSS is directly related to blood flow and inversely related to the third power of the growing radius (Schaper and Scholz, 2003; Scholz et al, 2002). Enhanced blood flow in the cerebral collateral circulation can be reached, in principle, by either a reduction of the flow resistance or an increase of the pressure gradient.

Because the main blood supply to the brain is physiologically provided by the two carotid arteries (6.35±1.14 mL/min via each internal carotid artery versus 1.38±0.59 mL/min via the A. basilaris; Table 1) first choice to reduce resistance was to ligate both common carotid arteries (Double-Ligature model). This treatment was expected to lead to a prompt flow increase in the A. basilaris and both Aa. cerebri posteriores (Figure 1). The approach to enhance blood flow by increasing the pressure gradient via an AV fistula (steep pressure gradient between the arterial and venous system) has been used for many years in different species and parts of the vascular system (Ben Driss et al, 1997; Eitenmuller et al, 2006; Pipp et al, 2004; Tohda et al, 1992; Tronc et al, 1996). So, in the Ligature-Shunt model we used both, bilateral ligature of the common carotid arteries and creation of an AV fistula between the distal stump of the ligated carotid artery on the left side and the adjacent jugular vein (end-to-side anastomosis). According to the resulting blood pressure conditions between the arterial (100–110mmHg) and venous system (3–6mmHg) a further increase of blood flow in the posterior cerebral collateral circulation should be the direct consequence (Figure 1). In this model, carotid backpressure in the cut distal stump of the ligated carotid artery dropped to 20 to 25mmHg (Figure 1). To avoid hemodynamic stroke, AV fistula-creation in this model was performed 2 days after bilateral carotid ligature. In the Solo-Shunt model, nonligated carotid artery on the one and AV fistula on the other side should cause a blood flow increase in the posterior and anterior cerebral collateral circulation (Figure 1). Carotid backpressure after ligature of only one carotid artery was 40 to 45mmHg (Figure 1).

Figure 1

Rat models to increase blood flow in the cerebral circulation. Schematic diagrams of physiologic, cerebral blood flow (Control) and expected flow changes in the different shear stress models. After ligature of both carotid arteries in the Double-Ligature model, a blood flow increase in the posterior cerebral collateral circulation should be the direct consequence. Arterio-venous fistula-creation on the left side, additionally to Double Ligature, was expected to increase blood flow even more because of the steep pressure gradient between the arterial and venous system (Ligature-Shunt model). In the Solo-Shunt model, right carotid artery remained nonligated and an AV fistula was created on the left side that should cause a blood flow increase in the posterior and anterior cerebral collateral circulation. Central arterial pressure was determined in the nonligated carotid artery, central venous pressure in the jugular vein, and carotid backpressure in the cut distal stump of the ligated carotid artery before AV fistula-creation. Red: arteries with intravascular blood flow; pink: ligated arteries with interrupted blood flow; orange: arteries with expected shear stress elevation.

Table 1

Blood flow changes in the rat cerebral collateral circulation (absolute: ml/min and relative to preoperative: fold changes)

Model	Time point	A. basilaris	A. cerebri post. right	A. cerebri post. left	A. cerebri ant. left	A. carotis interna right
Preoperative		1.38±0.59 1.0-fold	0.48±0.36 1.0-fold	0.48±0.36 1.0-fold	2.03±0.62 1.0-fold	6.35±1.14 1.0-fold
Double-Ligature	1 day	2.42±0.33 1.8-fold	0.64±0.43 1.5-fold	0.70±0.41 1.5-fold	0.82±0.17 0.4-fold
	7 days	5.69±1.33 4.3-fold	2.46±1.12 5.1-fold	2.34±0.17 4.9-fold	1.04±0.26 0.5-fold
	14 days	6.68±0.89 4.8-fold	2.48±0.89 5.2-fold	2.28±1.40 4.8-fold	1.63±0.41 0.8-fold
Ligature-Shunt	1 day	4.74±1.00 3.4-fold	1.06±0.20 2.2-fold	1.18±0.10 2.5-fold	0.42±0.23 0.2-fold
	7 days	6.43±0.54 4.7-fold	2.38±0.38 5.0-fold	3.60±0.59 7.5-fold	0.67±0.35 0.3-fold
	14 days	7.61±1.12 5.5-fold	2.61±0.70 5.4-fold	4.95±0.45 10.3-fold	1.17±0.33 0.6-fold
Solo-Shunt	1 day	3.02±0.61 2.2-fold	0.35±0.20 0.7-fold	0.83±0.26 1.7-fold	2.47±0.71 1.2-fold
	7 days	4.32±0.82 3.1-fold	0.44±0.14 0.9-fold	1.76±0.41 3.6-fold	2.96±0.46 1.5-fold	10.66±2.23 1.7-fold
	14 days	5.01±0.76 3.6-fold	0.53±0.37 1.1-fold	2.06±0.83 4.3-fold	4.07±1.46 2.0-fold	15.75±3.60 2.5-fold

Values are mean±s.e.m.

n=10 (1 or 7 days) and n=5 (14 days); post.=posterior and ant.=anterior.

Left side=AV fistula side in the Ligature-Shunt and Solo-Shunt group.

Cerebral Integrity as well as Functionality of the Arterio-venous Fistula was Verified by Magnetic Resonance Imaging

Control of cerebral integrity was performed by spine-cho T2- and diffusion-weighted MRI 1 day after bilateral carotid ligature or AV fistula-creation, respectively. Diffusion-weighted MRI is very sensitive to show early ischemic changes in the acute stage after stroke. The cytotoxic oedema visualized as hyperintensity on the diffusion-weighted images can be shown within 5 mins to 1 to 3 h after onset of symptoms (Mitsias et al, 2002; van Everdingen et al, 1998). In our models, cerebral integrity could be proved in all animals of the Double-Ligature and Solo-Shunt group (Figure 2A). None of the rats developed any neurologic symptoms during the further observation period. In the Ligature-Shunt group, 15.8% (3/19) of the animals developed signs of cerebral oedema indicating that the cerebral collateral system in this model comes close to its limits (data not shown).

Functionality of the AV fistula was confirmed by MRI-phase contrast angiography and quantitative MRI-flow measurement 1 day after surgery. Fistula-flow was 1.37±0.61 mL/min in the Ligature-Shunt and 3.28±1.34mL/min in the Solo-Shunt group. During the following 14 days-observation period, fistula-flow did not change significantly (data not shown). AV fistula-reocclusion (10.5% or 2/19) was only seen in the Ligature-Shunt group.

Magnetic Resonance Imaging-Analysis Demonstrated Significantly Increased Blood Flow after Surgery

One, 7, and 14 days after bilateral carotid ligature or AV fistula-creation, blood flow changes in the A. basilaris, Aa. cerebri posteriores, A. cerebri anterior, and A. carotis interna were monitored by quantitative MRI-flow measurement (Figure 2B; Table 1). Quantitative MRI-flow measurement was recently introduced as a noninvasive method for detecting blood flow insufficiency in patients who may require revascularization (Conway et al, 2008; Rutgers et al, 2000). Therefore, we first had to adapt this technique to the rat model (vessel diameter > 0.5mm).

When compared with the preoperative flow, MRI-analysis verified in all models a significant blood flow increase for the A. basilaris at every point of time (Figure 2C; Table 1). Significant differences in the basilar flow between the different models were observed 1 and 14 days after surgery (Figure 2C; Table 1). In both Aa. cerebri posteriores significant flow increase, compared with the preoperative flow, was confirmed in all models 7 and 14 days after bilateral carotid ligature or AV fistula-creation, respectively (Figure 2C; Table 1). MRI-flow measurement also revealed in these two collateral arteries a significant flow difference between the different treatment options ‘AV fistula—ligature—sham’ at 7 and 14 days, even in one animal (Figure 2C; Table 1). Further, in the Solo-Shunt group, blood flow increase was demonstrated during the 14 days-observation period in the left A. cerebri anterior toward the AV fistula by 2.0-fold and in the nonligated, contralateral carotid artery by 2.5-fold (Table 1), which was consistent with the existing studies using the carotid-jugular AV fistula (Tohda et al, 1992).

Increased Fluid Shear Stress Enhances Cerebral Arteriogenesis

Stimulation of vascular growth by the different shear stress models was analyzed postmortem by contrast agent-angiography at maximum vasodilatation as well as in vivo by MRI-phase contrast angiography to ensure that the postmortem visualization is truly reflecting the in vivo situation (Figures 3A and 3B).

Figure 3

Growth evaluation after 14 days of shear stress stimulation. Representative MRI-phase contrast angiographies and postmortem angiograms 14 days after indicated treatment. MRI was performed in vivo followed by visualization of the cerebral angioarchitecture postmortem under maximum vasodilation. (A) Details focus on the Aa. cerebri posteriores (arrows). Note the significant growth differences between the different treatment options ‘AV fistula—ligature—sham’ in the Ligature-Shunt and Solo-Shunt group. Scale bars: 1 mm. (B) Representative images of a Sham- and Ligature-Shunt-treated animal. Considerable vessel growth and the typical corkscrew formation of growing collaterals were also observed focussing on the shear stress stimulated A. basilaris and Aa. vertebralis (arrows). Scale bars: 1 mm. (C and D) Quantitative analysis of the postmortem-angiograms 14 days after surgery using ImageJ 1.38p software. Stimulation of cerebral collateral arteries by increased FSS caused a significant increase of vessel length (mm) and external diameter (µm). Vessel length was measured between stem and reentry, external diameters at three different levels. Contr, Control group; Doub-Lig, Double-Ligature group; Lig-Shunt, Ligature-Shunt group; Solo-Shunt, Solo-Shunt group; R, right side; L, left side; Lr, ligature right; Ll, ligature left; L, ligature; AVf, AV fistula; Sh, sham; values are expressed as mean±s.e.m.; n=5 each at every point of time).

A considerable, FSS-dependent enlargement of the Aa. cerebri posteriores and A. basilaris in length and especially in diameter appeared in all shear stress models 7 days (Table 2) and 14 days after surgery (Figures 3A–3D; Table 2). Maximum increase of length and diameter of the A. basilaris was found in the Ligature-Shunt group (Figures 3B and 3D; Table 2). Concerning the Aa. cerebri posteriores, analysis demonstrated a significant difference in vessel length and diameter between the different treatment options ‘AV fistula—ligature—sham’ (Figures 3A and 3C; Table 2). Enlargement of the A. cerebri anterior was only seen ipsilateral to the AV fistula in the Solo-Shunt group (Table 2).

Table 2

Growth of rat cerebral collateral arteries after 7 and 14 days of shear stress stimulation (length: mm and diameter: µm)

Model	Time point	Growth parameters	A. basilaris	A. cerebri post. right	A. cerebri post. left	A. cerebri ant. left
Control		Length	6.8±0.6	3.2±0.4	3.1±0.3
		Diameter	249.3±32.1	103.9±25.5	111.5±39.3	242.1±42.7
Double-Ligature	7 days	Length	7.3±0.6	3.8±0.2	3.6±0.4
		Diameter	291.8±29.2	234.2±26.5	232.9±29.3
	14 days	Length	7.5±0.7	3.9±0.5	3.8±0.4
		Diameter	326.5±33.4	281.8±23.8	275.6±36.2
Ligature-Shunt	7 days	Length	7.5±1.0	4.2±0.4	4.6±0.2
		Diameter	364.4±41.7	286.3±42.0	327.4±37.8
	14 days	Length	8.1±0.9	4.2±0.2	4.9±0.4
		Diameter	400.6±39.8	291.1±22.4	346.6±46.2
Solo-Shunt	7 days	Length	6.9±0.4	3.1±0.3	3.5±0.1
		Diameter	270.6±37.4	89.2±6.8	205.4±15.7	285.0±52.6
	14 days	Length	7.05±0.5	3.5±0.2	4.0 ±0.4
		Diameter	333.2±33.7	134.0±12.9	248.3±39.6	335.0±36.5

Values are mean±s.e.m.; n=5; post.=posterior and ant.=anterior.

Left side=AV fistula side in the Ligature-Shunt and Solo-Shunt group.

For the Aa. cerebri posteriores, a correlation was found between vessel length/external diameter and blood flow (Figure 4). The enlargement of the vessel length followed a linear and the enlargement of external diameter a potential trend line, both with high correlation (coefficient of determination R²: 0.9025 and 0.9555; Figure 4).

Figure 4

Flow-dependent growth of cerebral collaterals. Charts of vessel length (mm) and external diameter (µm) of the Aa. cerebri posteriores versus blood flow (ml/min) at equal points of time. A correlation was found between vessel length/external diameter (analyzed on postmortem angiograms by ImageJ 1.38p software at 7 and 14 days) and blood flow (determined in vivo by quantitative MRI analysis at 7 and 14 days). Vessel length: linear trend line, R²=0.90; external diameter: potential trend line, R²=0.96 (Values are expressed as mean±s.e.m.; n=5 each at every point of time).

Immunohistochemistry Showed Typical Signs of Arteriogenesis

We performed Ki67-immunostaining of the A. basilaris at multiple sections to ensure that the observed enlargement of the vessels is really because of arteriogenesis and verified active proliferation 1, 3, and 7 days after surgery in all shear stress models. Proliferation started from 1 day in adventitial cells and was subsequently followed by endothelial (from 3 days) and smooth muscle cells (from 7 days) (Figures 5A and 5B).

Figure 5

Increased proliferation activity and perivascular accumulation of mononuclear cells. (A) Representative images of the A. basilaris (Double-Ligature group). Active proliferation of vascular cells was detected by Ki67-immunostaining. Immunofluorescence confocal microscopy of shear stress stimulated A. basilaris showed the typical phases of arteriogenesis: Proliferation started from 1 day in adventitial cells and was subsequently followed by endothelial (from 3 days) and smooth muscle cells (from 7 days). Vessels for immunohistochemistry were not adenosine dilated. Red: Ki67, green (actin): Phalloidin-FITC, blue (nuclei): Draq5; Scale bars: 30 µm. (B) For determination of the proliferation index at 1, 3, and 7 days Ki67-positive nuclei of endothelial, medial, and adventitial cells were counted and referred to the total number of nuclei of each particular vessel layer (Values are expressed as mean±s.e.m.; n=3 each at every point of time). (C) Representative images of the A. basilaris (Double-Ligature group). Accumulation of mononuclear cells (CD11b-positive), a typical sign of arteriogenesis, was demonstrated. Red: CD11b, green (actin): Phalloidin-FITC, blue (nuclei): Draq5; Scale bars: 3 µm.

Immunohistochemical evaluation of monocyte accumulation showed a considerable increase of mononuclear cells perivascular to the growing cerebral collaterals (Figure 5C).

Discussion

In analogy to the observations in the coronary and peripheral circulation, cerebral collaterals that maintain perfusion beyond the site of an occluded artery have long been appreciated as a critical factor to reduce cerebral ischemia and stroke risk in the setting of cerebrovascular disease. Besides the leptomeningeal anastomoses, the circle of Willis, a low-resistance connection between the four main supplying arteries of the brain, is of great importance for brain integrity in patients with carotid or vertebral artery disease (Bisschops et al, 2003; Busch et al, 2003; Hendrikse et al, 2001, 2002; Liebeskind, 2004; Proweller et al, 2007; Todo et al, 2008).

As our group demonstrated in previous studies that FSS is the pivotal trigger of peripheral arteriogenesis leading to a distinctly increased collateral growth (Eitenmuller et al, 2006; Pipp et al, 2004), we tested the hypothesis that increased FSS is also a major stimulus of cerebral arteriogenesis.

According to the resulting blood pressure conditions, creation of an AV fistula distally to the ligature was the critical step to chronically elevate FSS in the peripheral collateral circulation (Eitenmuller et al, 2006; Pipp et al, 2004). Formation of an AV fistula between the common carotid artery and the jugular vein has already been described in the literature but was so far only applied to study the effects of blood flow increase on the remodeling of the common carotid artery (Tohda et al, 1992; Tronc et al, 1996). Moreover, these fistulas were created between the open common carotid artery and the jugular vein exhibiting a much higher shunt flow (Tohda et al, 1992; Tronc et al, 1996). The fistulas in the Ligature-Shunt and Solo-Shunt model on the contrary were created between the distal stump of the ligated carotid artery and the jugular vein. Because we could adapt the quantitative MRI-flow measurement to the rat model, our group was, to our knowledge, the first using the AV fistula for investigations on the effect of an increased blood flow on the remodeling of intracerebral vessels.

Arteriogenesis is defined as the remodeling of preexisting arterio-arteriolar anastomosis or arteries outside the ischemic region to functional vessels, which can completely replace the conductance capacity of a larger artery (Grundmann et al, 2007; Schaper and Scholz, 2003; Scholz et al, 2001). In the rat brain, this definition applies best to the two Aa. cerebri posteriores, arteries with marginal blood flow under physiologic conditions that enlarge after bilateral carotid ligature to provide sufficient brain perfusion. Blood flow increased significantly in all models reaching maximum intravascular flow rates in the Ligature-Shunt model after 14 days. However, carotid backpressure dropped to 40 to 45mmHg after ligature of one and to 20 to 25mmHg after ligature of both common carotid arteries indicating that the observed growth effects were caused by alterations of blood flow but not pressure (Korff et al, 2008).

It has already been proved in former studies that arterial remodeling is in principle related to hemodynamics (Langille, 1996; Tohda et al, 1992; Tronc et al, 1996). For the Aa. cerebri posteriores, a correlation was found between external diameter and blood flow according to a potential trend line. In addition, intracerebral collaterals not only increased in diameter but also flow dependently in length (linear trend line) and developed the typical corkscrew formation of collateral arteries. Both, length and tortuosity are known to contribute to the total resistance to flow limiting a more intense blood flow increase (Schaper and Scholz, 2003; Scholz et al, 2001, 2002).

Meanwhile, it is a well-known fact that collaterals (coronary, peripheral, and cerebral) grow by DNA synthesis and mitosis of endothelial and smooth muscle cells—cells that are quiescent in normal adult arteries but can rapidly convert to G¹ under abnormal conditions, as well as that the remodeling of the adventitia is essential to create the necessary space for the enlarging collaterals (Busch et al, 2003; Grundmann et al, 2007; Pipp et al, 2004; Schaper et al, 1971; Schaper and Ito, 1996; Scholz et al, 2002). Immunohistochemical evaluation showed the typical phases of arteriogenesis for all growing collaterals in our shear stress models as well as the accumulation of mononuclear cells in the adventitia, a typical sign of arteriogenesis (Arras et al, 1998; Busch et al, 2003; Grundmann et al, 2007; Pipp et al, 2004; Scholz et al, 2002).

Evidence that arteriogenesis can be induced in the circle of Willis of adult rats was already supplied in previous studies by Busch et al (Busch et al, 2003, 2006; Buschmann et al, 2003; Schneeloch et al, 2004). To investigate cerebral arteriogenesis, they developed the 3-VO model (unilateral common carotid artery occlusion in combination with bilateral vertebral artery occlusion). Comparison of our 1-week results with that of the 3-VO of Busch et al (2003) showed similar external diameters of the A. cerebri posterior at the ligature-sides (260 µm 3-VO versus 234 to 286 µm ligature-side of Double-Ligature and Ligature-Shunt). In the Ligature-Shunt model, increased FSS at the fistula-side enhanced cerebral arteriogenesis even further (up to 327 µm at 7 d). Comparable external diameters concerning the ipsilateral A. cerebri anterior were found between the 3-VO and Solo-Shunt model (292 µm 3-VO versus 285 µm Solo-Shunt).

Although we chose a different endpoint for growth evaluation (2 weeks instead of 3) all four models, 3-VO, Double-Ligature, Ligature-Shunt, and Solo-Shunt, proved as highly suitable to study cerebral arteriogenesis. In addition, with our new cerebral shear stress models, we could provide evidence that FSS is not only the pivotal trigger of peripheral and coronary (Eitenmuller et al, 2006; Grundmann et al, 2007; Pipp et al, 2004; Schaper and Scholz, 2003; Scholz et al, 2001) but also of cerebral arteriogenesis, and that cerebral arteriogenesis can be further enhanced by increased FSS.

One limitation of this study is the lack of the investigation of the leptomeningeal anastomoses as important preexisting collateral arterioles. Furthermore, we did not include functional investigations on the cerebral hemodynamic reserve capacity. However, no serious neurologic symptoms appeared as a consequence of the different interventions.

In conclusion, we demonstrated that the growth of cerebral collaterals correlates with rising intravascular flow rates. We showed evidence for the first time that FSS is not only a major stimulus of peripheral but also cerebral arteriogenesis. So our shear stress models offer several applications for further studies, for example, concerning FSS-dependent gene expression or FSS-activated pathways in cerebral arteriogenesis.