SDF‐1/CXCR4 signalling is involved in blood vessel growth and remodelling by intussusception

Abstract The precise mechanisms of SDF‐1 (CXCL12) in angiogenesis are not fully elucidated. Recently, we showed that Notch inhibition induces extensive intussusceptive angiogenesis by recruitment of mononuclear cells and it was associated with increased levels of SDF‐1 and CXCR4. In the current study, we demonstrated SDF‐1 expression in liver sinusoidal vessels of Notch1 knockout mice with regenerative hyperplasia by means of intussusception, but we did not detect any SDF‐1 expression in wild‐type mice with normal liver vessel structure. In addition, pharmacological inhibition of SDF‐1/CXCR4 signalling by AMD3100 perturbs intussusceptive vascular growth and abolishes mononuclear cell recruitment in the chicken area vasculosa. In contrast, treatment with recombinant SDF‐1 protein increased microvascular density by 34% through augmentation of pillar number compared to controls. The number of extravasating mononuclear cells was four times higher after SDF‐1 application and two times less after blocking this pathway. Bone marrow‐derived mononuclear cells (BMDC) were recruited to vessels in response to elevated expression of SDF‐1 in endothelial cells. They participated in formation and stabilization of pillars. The current study is the first report to implicate SDF‐1/CXCR4 signalling in intussusceptive angiogenesis and further highlights the stabilizing role of BMDC in the formation of pillars during vascular remodelling.

is intensively investigated. The domain comes to be multifaceted and contradictory data were sometimes arising. The discovery that mononuclear cells can home to sites of hypoxia and enhance neo-angiogenesis has faced the possibility of using isolated hematopoietic stem cells or endothelial progenitor cells (EPC) for therapeutic vasculogenesis. 3 However, infusion of EPC did not improve neovascularization 4,5 suggesting that a not-yet-defined functional characteristic (eg, chemokine or integrin receptors mediating homing) is essential for EPC-mediated vascular augmentation after ischaemia. 6 During endothelial repair after vascular injury and during tumour angiogenesis, BMDC do not seem to be involved in re-endothelialization, stressing their supportive role over trans-differentiation. 7 Ischaemia is believed to up-regulate VEGF or SDF-1 (CXCL12), 9 the latter in turn is released to the circulation and induces mobilization of progenitor cells from the bone marrow via a MMP-9-dependent mechanism. 10,11 Indeed, SDF-1 has been proven to stimulate recruitment of progenitor cells to the ischaemic tissue. 12 SDF-1 protein levels were increased during the first days after induction of myocardial infarction. 13 Moreover, overexpression of SDF-1 augmented stem cell homing and incorporation into ischaemic tissues. 12,13 Interestingly, hematopoietic stem cells were shown to be exquisitely sensitive to SDF-1 and did not react to G-CSF or other chemokines (eg, IL-8 and RANTES). 14 SDF-1/CXCR4 axis is crucial in the homing mechanisms of hematopoietic cells and metastasis of solid tumours. In the past few years, numerous studies have focused on unravelling the role of this signalling in angiogenesis and prove its angiogenic activity in organ repair and tumour development. However, the precise mechanisms by which SDF-1 exerts its pro-angiogenic effects are not fully elucidated. Since it is supposed to be an angiogenic growth factor, it is a good candidate for pro-and anti-angiogenic therapy. It was reported that transient disruption of the SDF-1/CXCR4 axis using CXCR4 blocking antibody blocked the recruitment of bone marrow-derived cells into the angiogenic sites of tumour tissue, and resulted in an inhibition of accelerated tumour growth. 15 Recently we have shown that inhibition of Notch signalling induces extensive intussusceptive angiogenesis by recruitment of mononuclear cells. 16 Notably, it was associated with increased levels of SDF-1 and CXCR4 chemotaxis factors. Intussusceptive angiogenesis is a process linked to both blood vessel replication and remodelling in development. 17,18 It is well documented as a mechanism of vascular adaptation in response to different environmental stimuli such as chronic systemic hypoxia 19 and prolonged inflammation. 20 It is also a mechanism of compensatory vascular growth. For example, capillary repair during kidney recovery in Thy1.1 nephritis proceeds through intussusceptive angiogenesis. 21 Similarly, the switch from sprouting to intussusceptive angiogenesis, found to occur in tumours after irradiation therapy, allows the vasculature to maintain its functional properties. 22,23 Potential candidates for molecular targeting of this angioadaptive mechanism are yet to be elucidated in order to improve the currently poor efficacy of contemporary anti-angiogenic therapies. Of major significance is the involvement of intussusceptive angiogenesis in pathological conditions such as liver nodular hyperplasia, 24 in the vasculature of experimental and clinical tumours, 25,26 in liver metastasis, 29 in metastatic tumours of the brain 30 and in breast cancer progression 31 among others.
Despite this variety of roles, attributed to intussusceptive angiogenic, most of the current research is focused on sprouting angiogenesis because the latter mechanism has been known since many decades ago and additionally, there are many experimental models related to sprouting angiogenesis. Here, we provide evidence that intussusceptive angiogenesis is regulated by SDF-1/CXCR4 signalling and suggest some intussusceptive angiogenic roles for CXCR4 and Tie-2 positive bone marrow-derived mononuclear cells (BMDC).

| Animals
Fertilized white leghorn eggs were obtained from commercial breeders (Fribourg, Switzerland). The eggs were incubated in-shell for 3 days at 37°C in humidified (65%) atmosphere, containing 1%-2% CO 2 . The eggs were opened on day 3 and further incubated at the same conditions using shell-free method 32 in petri dishes (Corning Incorporated, Corning, NY). The samples were divided into a control and experimental groups. There were at least six chicken embryos investigated in each group.
MxCre Notch1lox/lox mice on a C57Bl/6 background carrying the Cre-recombinase under the murine Mx1 promoter (described in 24). To induce recombination, 300 µg of polyiosinic-polycytidylic acid (pIpC) (InvivoGen, San Diego, CA) was injected intraperitoneally in 4-week-old mice at days 0, 3 and 6, resulting in efficient deletion of Notch1 in the liver already after 1 day. Notch1 deletion was consistent in liver sinusoidal endothelial cells (LSECs) and hepatocytes during the whole observation period.

| Reagents
Inhibition of Notch signalling was achieved as already described

| Morphometry
Evaluation of vascular parameters was accomplished with Tem Imaging Platform software (iTEM). Electronic images were acquired from normally developing CAV in order to obtain baseline data and a normal growth curve. From the experimental groups, images were

| Transmission electron microscopy
The samples obtained from the area vasculosa and PBS controls har-

| Vascular casting
Vascular casts were prepared as previously described. 33 Briefly, CAV vasculature and the vasculature of the murine livers were perfused with a freshly prepared solution of Mercox ® (Vilene Company, Tokyo, Japan) containing 0.1 mL of accelerator per 5 mL of resin.
One hour after perfusion, the specimes were transferred to 7.5% potassium hydroxide for digestion of tissue, which was effected over a course of 2-3 weeks. After washing, the casts were dehydrated in ethanol and dried in a vacuum desiccator. The samples were then sputtered with gold to a thickness of 10 nm and examined in a Philips XL-30 SFEG scanning electron microscope.

| Mononuclear cell counting
Semithin serial sections were obtained and images captured at magnification 40x using a light microscope (Leica, Leitz DM), equipped with Leica DFC480 camera. At least 10 images were taken per sample for further quantitative evaluation and at least 10 samples for each application were evaluated. The total number of adherent/ extravasated mononuclear cells per vessel circumference was assessed using analySIS Software 5.0 (Soft Imaging System, Muenster, Germany) by means of user-driven skeletonization of the vascular circumference.

| Bone marrow mononuclear cells isolation
To isolate bone marrow mononuclear cells, 4-to 6-week-old male

| Cell staining
The stock solution of CellTracker TM Green 5-Chloromethylfluorescein diacetate (10 mmol/L) was diluted to a final working concentration of 5 µmol/L in serum-free medium. The working solution was warmed to 37°C. The mononuclear BMDC were centrifuged to pellet them and aspirate the supernatant. The cells were gently re-suspended in pre-warmed working solution, then incubated for 15-45 minutes at 37°C and centrifuged. The probe solution was replaced with fresh, pre-warmed medium and incubated for another 30 minutes at 37°C.

| Adhesion assay
To perform adhesion assay, 10 5 labelled mononuclear BMDC were add per well of cultured endothelial cells (on coverslips) for 15 minutes. The wells were washed with PBS, coverslips were fixed with 3.7% formaldehyde for 15 minutes at room temperature, stained with Hoechst dye for 10 minutes, rinsed with water and mounted on a slide for counting by microscopy.

| Mononuclear cells isolation from CAV
Blood was washed out of the vessels of selected CAV samples. This was accomplished by injection of PBS. The treated tissue area was removed; making sure that for each embryo investigated the size of the tissue was the same. Tissues from at least 10 embryos for each test were then collected for each treatment and for controls. The tissue was cut into small pieces, digested in 500 µL 0.25% trypsin solution and 10 µL DNAse for 5 minutes at 37°C. The cell suspension was filtered through nylon gauze with a defined pore diameter of 30 µm, filled with 1X PBS and an equal volume of Histopaque-1077 was added gently. After 30 minutes of centrifugation, the white cellular ring floating over the Ficoll phase (containing mononuclear cells) was collected, filled with 1X PBS and centrifuged for 10 minutes at 400 g to wash. F I G U R E 1 Effects of Notch inhibition on intussusceptive angiogenesis in the chick area vasculosa are mediated by SDF-1/CXCR4 signalling. A, Fluorescein isothiocyanate microvasculature visualization after different targeted treatments. γ-secretase inhibitor (GSI) application or treatment with recombinant SDF-1 protein induced remarkably pillar formation (arrows), that is, intussusceptive angiogenesis. AMD3100 (CXCR4 antagonist) simultaneously applied with GSI demonstrated repressive effects on pillar formation. B, Bar graphs representing pillar density (*P < 10 -5 , **P < 10 -5 , ***P < 10 -6 ) and vessel area density (*P < 0.01, **P < 0.01, ***P < 0.01) after different applications

| Immunofluorescence
Sections from liver samples were obtained at 5 μm, deparaffinized, and rehydrated. On dewaxed and rehydrated slides heat induced epitope retrieval in citrate buffer pH6 (DakoCytomation) was carried out for 5 minutes to unmask the epitopes. This was followed by blocking with 1% casein for 10 minutes. Incubation with the primary mouse antibody SDF-1 and CXCR4 (Santa Cruz Biotechnology, Dallas, USA) overnight at 4°C (dilution 1:150) was followed by application of anti-mouse IgG-Cy3 (Sigma c2181) secondary antibody.
Counterstaining was performed with Hoechst dye.

| Statistical analysis
Probability associated with a Student's paired t test, with a two-tailed distribution, was considered in a given p value for each comparison. We performed immunofluorescent analysis for CXCR4 expression simultaneously with this one for SDF-1 in Notch1 KO mouse and positivity for the receptor was evident in the mononuclear cells recruited to the SDF-1-positive sinusoidal vessels ( Figure S2).

| Attraction of mononuclear cells to the endothelium and their plausible role in vessel remodelling
To examine one of the main functional effects of SDF-1-attraction of mononuclear cells, we did in vitro experiment with co-culture of bone marrow monocytes and endothelial cells. We were interested to see if the adhesion of BMD cells to the endothelium is comparable between the Notch inhibition subjects and SDF-1 treatment ones. The effect of adhesion after Notch inhibition and SDF-1 treatment was assessed by labelling mononuclear cells with Cell Tracker™ green and fluorescent microscopy. We detected comparable number of adherent mononuclear cells after GSI and SDF-1 treatment on endothelial cells ( Figure 3A; Figure S3). The inhibition of endothelial cells in culture (HUVEC) by GSI led to more than two times increase in SDF-1 endothelial expression ( Figure 3B). Successful GSI blocking was assessed by analysing the expression of Hes1 which is the most popular Notch target gene. These results suggest that the recruitment of mononuclear cells to the endothelium in GSI samples is due to SDF-1/CXCR4 signalling. There were considerable shape alterations in mononuclear cells due to SDF-1 application ( Figure 3C).

| Angiogenic expression pattern of extravasated mononuclear cells
Since the morphological analysis does not give us information about the molecular phenotype of the mononuclear cells, we specifically isolated these cells after GSI application from the in vivo model (CAV) and tried to characterize their expression pattern for some specific angiogenic markers, especially surface receptors ( Figure 5).

| Positive feedback relation between SDF-1 expression and shear stress
As intussusceptive angiogenesis can be initiated only in the presence of blood flow, an important aspect is the molecular link between shear stress and its effects on the regulation of angiogenesis. As the focus of our study here, we wanted to investigate the expression of SDF-1 under an exposure of specifically determined shear stress.  Figure 6A). Inversely, we tested if application of SDF-1 on endothelial cells increases the expression of eNOS-a marker of increased shear stress. We found more than three times increase in eNOS expression levels ( Figure 6B). These

| D ISCUSS I ON
Several studies in the recent past allowed distinguishing of subpopulations of mononuclear cells existing in the adult bone marrow and circulating in peripheral blood that support angiogenesis without being incorporated permanently into the newly formed vessels, the so-called circulating angiogenic cells (CAC). 34 Here, we showed that mononuclear cells of bone marrow origin are recruited to vessels in response to SDF-1 endothelial expression. Our previous studies demonstrated that extravasating mononuclear cells stabilize pillars, the hallmarks of intussusception, by formation of uropod-like protrusions and collagen production. 16 In the current study we have found that they express bFGF, supporting the previous observation. In addition, they were highly positive for the expression of CXCR4 and Tie-2 receptors. Pharmacological inhibition of SDF-1/CXCR4 signalling by small CXCR4 antagonist AMD3100 perturbs intussusceptive vascular growth (two times less pillar number) and abolishes mononuclear cell recruitment in GSI treated samples of CAV. In contrast, treatment of CAV by the mechanism of uropod formation was described, as the triggering event is cell polarization by chemotaxis molecules. 37 SDF-1 regulates adhesion, motility and cell shape in tumour progression. 38 They described morphological changes from round to polygonal shape, including the formation of neurite-like projections, increased membrane ruffling, and more frequent filopodia and uropod formation in response to SDF-1. We suggest that SDF-1 influences protrusion formation in mononuclear cells, facilitating their participation in pillar development during intussusceptive angiogenesis. This is the first study reporting connection between SDF-1 endothelial expression and intussusceptive mode of angiogenesis. A comprehensive study for SDF-1 expression in vessels was done by Salvucci et al. 39 They detected the presence of SDF-1 in endothelial cells of capillaries in bone marrow and skin, as well as in the endothelium lining umbilical veins, the chorionic villi, and the high endothelial venules in lymph nodes. Although SDF-1 is not expressed in the endothelia of capillaries from many organs in normal conditions (such as kidneys, brain, skeletal muscles, lung and liver), it is encountered in the capillaries from the same organs in pathological conditions, such as glioblastoma multiforme, infarcted brain tissue, Burkitt lymphoma tissue and lobular capillary hemangioma. This suggests that SDF-1 expression could be induced in endothelial cells during new vessel formation. We demonstrated SDF-1 positivity in the endothelium of liver sinusoidal vessels of Notch1 KO mice, representing regenerative hyperplasia by means of intussusception, but SDF-1 was not detected in WT mice with normal liver structure. The intraluminal endothelial protrusions were also positive for SDF-1 in the case of capillary hemangioma, described in the above-mentioned study. 39 SDF-1, when expressed in the bone marrow and various tissues, is able to regulate trafficking, localization and function of immature and mature leukocytes, including monocytes, neutrophils, dendritic cells and T lymphocytes. 40 All these immune cells play important roles in tumour angiogenesis and vascularization. It is well known that blocking of SDF-1/CXCR4 axis results in prevention or delay of tumour recurrence after irradiation by inhibiting the recruitment of CD11b+ monocytes/macrophages that participate in tumour revascularization. 41  Despite the important role of intussusception in vessel formation and remodelling, most of the existing studies are focused on the better known mechanism of sprouting angiogenesis. Thus, the mechanism of intussusceptive angiogenesis has not been adequately covered contemporary by investigators in the field of angiogenesis research. Intussusception is an alternative to the sprouting mode of angiogenesis. 44 The advantage of this mechanism of vascular growth is that blood vessels are generated more rapidly and the capillaries thereby formed are less leaky. 45 Regarding molecular regulation, very little is known of the molecular factors with potential significance in intussusceptive angiogenesis. Application of the essential angiogenic factors VEGF and bFGF in the arteriovenous loop model demonstrated advanced neovascularisation in the phase of remodelling by a higher incidence of intussusception, compared to controls. 46 It was found that neovascularization induced by VEGF plus bFGF is mediated by SDF-1/CXCR4 signalling and that SDF-1 neutralization reduced growth factor-triggered neovascularization by approximately 84%-86%. 39 In addition, our experiments on HUVEC demonstrated positive feedback between shear stress and SDF-1. Recently, it was proposed that Notch1 is atheroprotective and acts as a mechanosensor in adult arteries, where it integrates responses to laminar shear stress, 47 as well as that chemokine receptor CXCR4 pathway is the key regulator of Notch-dependent vessel growth. 48 In conclusion, our study is the first to implicate SDF-1/CXCR4 signalling in the mechanism of intussusceptive angiogenesis. Plausibly, BMDC play an important physical and stabilizing role in the formation of pillars during vascular augmentation. analysed the data and wrote the paper. DS contributed the knockout mice for the study and analysed the data.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.