PECAM/eGFP transgenic mice for monitoring of angiogenesis in health and disease

For the monitoring of vascular growth as well as adaptive or therapeutic (re)vascularization endothelial-specific reporter mouse models are valuable tools. However, currently available mouse models have limitations, because not all endothelial cells express the reporter in all developmental stages. We have generated PECAM/eGFP embryonic stem (ES) cell and mouse lines where the reporter gene labels PECAM+ endothelial cells and vessels with high specificity. Native eGFP expression and PECAM staining were highly co-localized in vessels of various organs at embryonic stages E9.5, E15.5 and in adult mice. Expression was found in large and small arteries, capillaries and in veins but not in lymphatic vessels. Also in the bone marrow arteries and sinusoidal vessel were labeled, moreover, we could detect eGFP in some CD45+ hematopoietic cells. We also demonstrate that this labeling is very useful to monitor sprouting in an aortic ring assay as well as vascular remodeling in a murine injury model of myocardial infarction. Thus, PECAM/eGFP transgenic ES cells and mice greatly facilitate the monitoring and quantification of endothelial cells ex vivo and in vivo during development and injury.

Vascular growth is a key process during development and it also determines vessel remodeling in response to injury. Therefore transgenic models, in which endothelial cells are specifically labeled are valuable tools to monitor vascular network formation and to generate novel strategies for (re)vascularization. In the past different potentially endothelial-selective reporter mouse lines (tie1/eGFP 1 , tie2/eGFP 2 , flk1/eGFP 3 , flt1/eGFP 4 ) have been generated. However, their use is limited because they either show expression restricted to distinct developmental stages or they do not express the reporter in all endothelial cells or endothelial cells of all organs 1,2,4 . The respective Cre mice often displayed expression in non-endothelial cells mostly because the promoters are also active in other cell types during development [5][6][7][8][9] . To overcome these limitations and to generate a reporter gene model with endothelial-specific expression at all stages and in all vascular beds we have chosen the promoter of the platelet endothelial cell adhesion molecule 1 (PECAM) driving the eGFP reporter gene. PECAM is a cellular adhesion and signaling receptor, which acts as a mechanosensor and plays a major role during extravasation of leukocytes and the regulation of the vascular permeability barrier 10,11 . Because of the particularly strong expression on the endothelium PECAM is considered as the classical endothelial marker. We therefore thought that the PECAM promoter driving a reporter gene should be ideal for robust endothelial cell labeling and also for the monitoring of cell fate and vascular remodeling in injury models in vivo. Herein, we have generated PECAM/eGFP embryonic stem (ES) cells and the respective mouse line. This allowed direct visualization of the endothelium in all vascular beds at single cell resolution. Furthermore, we demonstrate the utility of the PECAM/eGFP mouse model to investigate angiogenesis in aortic outgrowth experiments and to monitor vascularization in the heart after experimental myocardial infarction.

Methods
Generation of PECAM/eGFP BAC vector. For the generation of the PECAM/eGFP vector, the 220 kbp bacterial artificial chromosome (BAC) RP23-120D9 (obtained by the Children's Hospital Oakland Research Institute, CHORI) covering the promoter region of the PECAM gene was transferred into SW105 E. coli cells by electroporation. This procedure we have reported recently 4 . Briefly, an eGFP-pA cDNA was created for insertion at the initiation codon in the first exon of the PECAM gene within the BAC. Homologous arms were generated by PECAM/eGFP transgenic mice. G4 PECAM/eGFP ES cell clones, which were screened for normal 40 chromosome karyotype, were used for aggregation with diploid CD1 embryos at morula stage as described previously 15 . Chimeric mice could be identified due to their spotted coat and were bred with CD1 mice to examine germline transmission. PCR of tail tips revealed the transgenic offspring: Following primers were used (fw: 5′ ACC TGC CAT CAC GAG ATT TC 3′, rev: 5′ TTC TCC TAC ACC GCT GTC TC 3′). The expression of the transgene was studied at day 9.5 and 15.5 of embryonic development (E9.5, E15.5) as well as in adult organs and tissues by fluorescence microscopy (AxioZoom V16, AxioObserver Z1 and Axiovert 200 M, Carl Zeiss, Oberkochen, Germany).

Isolation of adult cardiac endothelial cells via magnet-activated cell sorting (MACS). MACS
was performed as described before 21 . 6-to 8-week-old WT and PECAM/eGFP mice were sacrificed and hearts were isolated. Then, the aorta was perfused with 1.5 ml PBS before removing the atria. Ventricles were minced and enzymatically dissociated at 37 °C and 800 rpm for 25 minutes. For one heart 800 µl of enzyme buffer was used. To prepare the enzyme puffer 5 mg liberase III (Roche, Mannheim, Germany) was dissolved in 20 ml 10 mM HEPES solution in HBSS (Ca 2+ , Mg 2+ ) and aliquots of 800 µl were stored at −20 °C. Before the dissociation of tissue 1 µl 100 U/ml DNAse I (Life technologies, Darmstadt, Germany) was applied to the enzyme buffer aliquot. After the first dissociation the supernatant was removed and stored on ice while the remaining tissue was used for dissociation again. Cells in suspension were magnetically labelled and separated using PECAM microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's instructions. Purity was examined by flow cytometric quantitation of eGFP + /PECAM-PE + (BD Biosciences, Heidelberg, Germany) cell populations using a CyFlow space flow cytometer (Sysmex Partec, Görlitz, Germany). containing protease inhibitors (Roche) directly after MACS procedure. The protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Fischer Scientific). SDS-PAGE was performed in a 10% polyacrylamide gel. Afterwards, proteins were transferred to a nitrocellulose membrane using the semi-dry blotting procedure. Following primary antibodies were applied: anti-GFP (mouse, 1:1,000, Clontech), peroxidase-coupled anti-β-actin (1:10,000, Sigma Aldrich). A peroxidase-coupled anti-mouse antibody (1:10,000, Jackson ImmunoResearch) was applied as secondary antibody for eGFP detection. Chemiluminescence signals were detected using the SuperSignal TM West Pico Plus Substrate (Thermo Fischer Scientific) and a ChemiDoc TM MP imaging system (BioRad).

Western blot analysis.
Mouse aortic ring assay. Aortic ring assay was performed as described before 22 . 6-month-old PECAM/ eGFP mice were killed and thoracotomy was performed to expose and dissect the thoracic aorta. The aorta was immediately transferred to a cell culture dish filled with Opti MEM (Life technologies, Darmstadt, Germany) and the peri-aortic fibroadipose tissue was removed carefully. One-millimeter-long aortic rings were sectioned and starved in Opti MEM with 1% penicillin and streptomycin over night at 37 °C. Then the aortic rings were embedded in 250 µl collagen gel per well in an 8 well µ-Slide (ibidi GmbH, Martinsried, Germany). This gel was generated by mixing 1.7 ml DMEM (Invitrogen, Karlsruhe, Germany) with 600 µl 3.79 mg/ml collagen I (Merck Millipore, Billerica, USA) and 2 µl 5 M NaOH. Polymerization of the collagen gels was achieved by incubation at RT for 15 minutes and at 37 °C for 60 minutes. The polymerized gels containing the aortic rings were covered with 180 µl Opti-MEM containing 1% penicillin and streptomycin and 2% FCS which was exchanged every third day.
Cultures were kept at 37 °C and the explants were monitored for the outgrowth of endothelial cells on a daily basis using a stereomicroscope (AxioZoom V16, Carl Zeiss, Oberkochen, Germany).

Experimental myocardial cryoinfarction of PECAM/eGFP mice. Animals experiments were
approved by the local ethics committee and carried out according to the guidelines of the German law of protection of animal life with approval by the local government authorities (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen (LANUV), NRW, Germany) and in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Myocardial cryoinfarctions were generated as described before 17 . In order to examine the vascularization of the (peri-) infarct zone mice were sacrificed 3 or 6 days after the cryolesion. Hearts were harvested and immunohistochemical processing and staining was performed as explained above.

Generation of PECAM/eGFP embryonic stem (ES) cells.
To generate a reporter gene construct that labels PECAM + endothelial cells an eGFP expression construct was inserted into the first exon of a BAC spanning the murine PECAM promoter (Fig. S1). First, the expression pattern and specificity of the labeling were analyzed in murine ES cells. Therefore, the PECAM/eGFP vector was used for electroporation in G4 ES cells. As previously described for PECAM + ES cells 23 , PECAM/eGFP + cells were initially arranged in clusters (Fig. 1A). During differentiation in embryoid bodies (EBs) first eGFP + sprouts appeared around d 8-10 ( Fig. 1B,C, arrow heads). After plating of the EBs on d 10 eGFP + networks developed through d 10 + 4 to d10 + 7 ( Fig. 1D-F). Immunohistochemistry revealed that native eGFP expression of the sprouts (green, Fig. 1G,I) perfectly co-localized with PECAM staining (red, Fig. 1H,I), which confirms PECAM-specific expression of the reporter gene in the ES cell system.
Characterization of PECAM/eGFP transgenic mouse embryos. Next, we generated a transgenic mouse line by aggregation of PECAM/eGFP ES cells with morula stage wildtype embryos. The mice were developmentally normal and no adverse effects of PECAM/eGFP expression could be detected. When transgenic E9.5 embryos were compared with littermate controls PECAM/eGFP + embryos demonstrated eGFP + developing vascular networks, while controls showed no fluorescence (Fig. S2A). In particular, bright sprouts were found in the brain (Fig. S2B), the heart (Fig. S2C) and the caudal region of the embryo (Fig. S2D) The eGFP fluorescence in the E9.5 embryo showed an excellent overlap with PECAM whole mount stainings ( Fig. 2A,B). This could be also confirmed in cryosections where native eGFP (green) was co-localized with PECAM staining (red) throughout the embryo (Fig. 2C).
The strong eGFP labeling of the vasculature was also observed in cryosections derived from E15.5 embryos (Fig. 2D). At this stage, in particular highly vascularized organs such as the heart (Fig. 2E), lung (Fig. 2F), gut (Fig. 2G) and the paw (Fig. 2H) displayed bright native eGFP fluorescence (green). Also in these organs native eGFP expression and PECAM staining (red) showed a good overlap in cryosections ( Fig. S2E-P). At high resolution immunostainings of heart sections demonstrated that almost all eGFP + cells were also labeled with PECAM antibody (Fig. 2I-K) further confirming endothelial specific eGFP expression at E15.5.
Analysis of PECAM/eGFP adult mice. Then, we analyzed the expression pattern of the reporter gene in adult animals. Stereomicroscopy of adult mouse organs revealed that native eGFP expression was restricted to vascular network-like structures in all organs examined, such as the brain (Fig. 3A) and basilar artery (Fig. 3B), the outer eye (Fig. 3C) and retina (Fig. 3D), skin (Fig. 3E), skeletal muscle (Fig. 3F), heart (Fig. 3G), lung (Fig. 3H), liver (Fig. 3I), kidney (Fig. 3J), gut (Fig. 3K) and uterus (Fig. 3L). Immunostainings of cryosections using anti-PECAM antibody demonstrated that cutaneous vessels (Fig. 4A-C), the alveolar capillary network and peribronchial vessels (Fig. 4D-F), glomeruli and interlobular vessels of the kidney (Fig. 4G-I), vessels of the uterus (Fig. S3A-C) and gut (Fig. S3D-F) as well as cardiac capillaries (Fig. 4J-L) and coronary arteries (Fig. 4J-L, inset) displayed strong eGFP fluorescence and revealed almost a complete overlap with PECAM staining underscoring endothelial cell-specific eGFP expression. Within the heart, we analyzed eGFP expression in detail by cell counting in sections and found that 87.1 ± 0.02% of the labeled cells were eGFP + PECAM + , 9.5 ± 0.01% eGFP − PECAM + and only 3.4 ± 0.0% appeared to be eGFP + PECAM − (n = 3, Fig. 4M). Hence eGFP labeling in PECAM/eGFP adult hearts is highly specific and the vast majority of PECAM + cells show eGFP fluorescence. To analyze PECAM and eGFP expression in isolated endothelial cells, we have performed magnet-activated cell sorting (MACS) of adult hearts derived from WT and PECAM/eGFP mice using an anti-PECAM antibody. Flow cytometry analysis with anti-PECAM-PE antibody staining demonstrated successful purification. (Fig. 4N). Western blot analysis of the PECAM + cells then further proved eGFP expression of PECAM + cells in PECAM/ eGFP animals while there was no signal in WT animals confirming specifity of the eGFP antibody (Fig. 4O).
Next, we also investigated the eGFP expression in large arteries and veins. While stereomicroscopy revealed that in our previously generated flt-1/eGFP mouse model eGFP expression was absent in large arteries 4 (Fig. S4A-E), in the PECAM/eGFP mouse line we detected prominent eGFP expression (Fig. S4A) in the endothelium of aorta (Fig. S4F-I) and carotid arteries. Similarly, also in large veins such as the portal vein we found a complete overlap of eGFP expression and PECAM staining (Fig. S4J-L).

PECAM/eGFP mice for aortic ring assay and experimental myocardial cryoinfarction.
To determine if the PECAM/eGFP mouse line can be also used in order to facilitate analysis of standard angiogenesis assays we isolated aortas and performed an aortic ring sprouting assay. For this purpose PECAM/eGFP aortic rings were embedded in a collagen matrix and sprouts could be easily visualized by native eGFP fluorescence at d8 (Fig. 6A,B) and d10 (Fig. 6C). Immunohistochemical analysis with Griffonia Simplicifolia lectin (GSL, Fig. 6D-F) and PECAM (Fig. 6G-I) staining confirmed the endothelial character of these eGFP + sprouts. Additionally, we also tested the PECAM/eGFP mouse line in an injury model in vivo, because in particular the processing and analysis of revascularization of infarcted heart tissue can be difficult. This is due to staining artefacts, as antibody staining in the infarct area often lacks specificity 24 . Therefore, myocardial cryoinfarctions were generated in PECAM/eGFP mice and vascularization was analyzed on d 3. In the infarction area background fluorescence was strongly decreased (area surrounded by dotted line, Fig. 6J) and in the overview many eGFP + elongated structures could be identified. Immunostainings of cryosections revealed that there was a good overlap of eGFP fluorescence (green) and PECAM staining (red) in the border zone (upper rectangle Fig. 6J, Fig. 6K-M) and also in the center of the cryoinfarction (lower rectangle Fig. 6J, Fig. 6N-P). Thus, the PECAM/eGFP mouse line enables straightforward identification of endothelial structures in ex vivo angiogenesis assays and injury models in vivo without any additional processing and staining procedures.

Discussion
Herein, we have generated a PECAM/eGFP murine ES cell and a transgenic mouse model for live monitoring of endothelial cells in small and large vessels in vitro, ex and in vivo.
For the generation of transgenic models we have chosen the PECAM promotor driving the reporter gene eGFP. During development the intraembryonic PECAM expression already starts at 7.8 dpc 25 and persists throughout adulthood in the entire vascular endothelium 26 . PECAM/eGFP plasmids had been used for in vitro transfection of ES cells [27][28][29] and HUVECs 30 before, however, the generation of a PECAM/eGFP transgenic mouse has not been reported so far. In our PECAM/eGFP ES cell and mouse lines we have found endothelial-specific expression of eGFP at all stages and in vessels of all calibers. This expression pattern, which recapitulates the endogenous PECAM expression, is probably due to the inclusion of all regulatory sequences in the PECAM BAC that was used as transgenic vector. Apart from endothelial cells, PECAM expression can be also found on the surface of platelets and leukocytes, however, in these cells expression is described to be much lower 31 . This is in accordance with our data, as only a small number of eGFP + /PECAM + hematopoietic cells was found in the bone marrow and this is most likely due to the overall low transgene expression in these cells. All our experimental data underscore the high specificity of the transgene expression for endothelial and also hematopoietic cells.
We are convinced that the high specificity of the labeling is very helpful for the analysis of state of the art ex vivo angiogenesis assays, such as the aortic ring assay 32 . In fact fluorescence microscopy enabled us to clearly distinguish between eGFP + endothelial cells and other types of sprouting cells (i.e. fibroblasts and smooth muscle cells),  thereby avoiding additional immunostainings, which may also facilitate the quantification of results 33 . Moreover, the PECAM/eGFP mouse model is helpful for the monitoring of revascularization processes upon injury (e.g. myocardial infarction). Currently, fixation and antibody staining is required, which adds a confounding factor, as unspecific stainings in injured tissues make it difficult to discern real signals from artefacts 24

Data Availability
Data are available upon request.