Generation, expansion and functional analysis of endothelial cells and pericytes derived from human pluripotent stem cells

Human endothelial cells (ECs) and pericytes are of great interest for research on vascular development and disease, as well as for future therapy. This protocol describes the efficient generation of ECs and pericytes from human pluripotent stem cells (hPSCs) under defined conditions. Essential steps for hPSC culture, differentiation, isolation and functional characterization of ECs and pericytes are described. Substantial numbers of both cell types can be derived in only 2–3 weeks: this involves differentiation (10 d), isolation (1 d) and 4 or 10 d of expansion of ECs and pericytes, respectively. We also describe two assays for functional evaluation of hPSC-derived ECs: (i) primary vascular plexus formation upon coculture with hPSC-derived pericytes and (ii) incorporation in the vasculature of zebrafish xenografts in vivo. These assays can be used to test the quality and drug sensitivity of hPSC-derived ECs and model vascular diseases with patient-derived hPSCs.


IntroDuctIon
The discoveries of human embryonic stem cells (hESCs) and human induced PSCs (hiPSCs) were major breakthroughs for biomedical research [1][2][3] . The ability to generate many cell types of the human body made it possible to consider not only therapeutic applications for tissue regeneration but also disease modeling and drug discovery on human cells. Although initial challenges in deriving specific cell lineages through differentiation have been substantially addressed for neuronal, cardiac and endoderm cell types, including hepatic, lung and intestinal cells [4][5][6][7][8][9][10][11] , the derivation of large numbers of healthy and patient-specific ECs for use in pharmacological screening and tissue vascularization after transplantation has remained difficult. In addition, as dysfunctional ECs have been linked to multiple genetic and metabolic vascular disease conditions and the growth of tumors is promoted by neovascularization, new human cellular systems that model these conditions could prove crucial to moving these fields forward.
For decades, human umbilical vein ECs (HUVECs) have been the 'gold standard' in EC research because they are easily accessible from surplus umbilical cords. More recently, endothelial progenitor cells (EPCs) derived as endothelial colony-forming cells (ECFCs) from cultured cord blood or adult peripheral blood have also been used as a source of ECs 12,13 . Although cord blood banking is valuable for collecting hematopoietic stem cells, cord blood is not available from all individuals and it is not amenable to scaled-up production; thus, alternative sources for patientspecific ECs would be of value.
The importance of pericytes in the formation and function of the vasculature has been recognized for some time. Several studies have demonstrated their role in regulating EC 'sprouting', potentiating the formation of an EC barrier and determining blood capillary diameter, as well as their involvement in overall blood capillary stability 14 . Pericyte loss has been associated with multiple neurovascular diseases, including diabetic retinopathy, brain hemorrhage and some neurodegenerative disorders, and it is a target of some cardiotoxic drugs [15][16][17] . More recently, pericytes have also emerged as a possible source of adult multipotent mesenchymal stem cells (MSCs) 18 , which can contribute to the regeneration of not only blood vessels, but also bone, cartilage, fat and skeletal muscle 19,20 .
We have shown recently that it is possible to derive the principle vascular cell types robustly from hPSCs, and that the cells can be expanded in culture and cryopreserved and are functional in vitro and in vivo 21 . Here we describe these fully defined protocols for EC and pericyte derivation and functional analysis. hPSCs as used in these protocols are a renewable source of healthy and patient-specific vascular cells, enabling many of the applications mentioned above to be developed further. Figure 1 gives an overview of the entire procedure.

Overview of previous methods to derive ECs and pericytes from hPSCs
Several independent laboratories have published descriptions of the derivation of ECs and pericytes from hPSCs 22,23 . However, these have often been difficult to reproduce owing to neither the culture nor differentiation conditions for hPSCs being standardized using fully defined reagents. Multiple factors can affect the outcome of differentiation: the presence of feeder cells, FBS, BSA and the quality of stromal cell lines and growth factors. Most of the earliest protocols were developed based either on serumsupplemented medium or stromal cell lines, or on both [22][23][24][25][26][27][28] . Although defined conditions have recently been described, they result in very low efficiencies of EC differentiation (<1.5% ECs) 29 , and they are difficult to scale up because their methodology is based on aggregate (or embryoid body (EB)) formation [30][31][32] . More recently, an efficient forced aggregation (or so-called SpinEB) method for EC differentiation has been developed 33 , but this has the major disadvantage that an additional step is required in which hPSCs are adapted to single-cell passage. Establishing singlecell passage can be time-consuming and difficult to perform for multiple hPSC lines simultaneously. An alternative, technically more straightforward monolayer differentiation method has been used; however, neither the percentages of ECs in the differentiated cell populations nor evidence that the protocol was robust across multiple hPSC lines was provided 34 .
In addition to efforts to develop EC differentiation protocols, methods for the differentiation of pericytes and smooth muscle cells (SMCs, or mural cells) from hPSCs have been recently established 23,27,35 , although these do not describe simultaneous differentiation of ECs and mural cells in defined conditions that facilitate efficient purification and expansion of both cell populations in one procedure. Further, expansion and cryopreservation without loss of functionality is a key requirement for successful application of hPSC-derived ECs; our recent paper was the first in which this has been described 21 .
Experimental rationale for the derivation of EC and pericytes Source hPSCs and initial passage (Steps 1-12). The protocol described here has been optimized for hPSCs maintained in commercially available and well-defined mTeSR1 stem cell growth medium and passaged enzymatically. The differentiation protocol was tested and simplified for use on multiple hPSC lines, including various hESC lines (NL-HES4, HES3 (NKX2-5eGFP/w)), as well as hiPSC lines derived from either skin fibroblasts or blood outgrowth ECs (BOECs) by transduction with either retroviruses or lentiviruses. The differentiation efficiency was indistinguishable among more than 15 different hPSC lines tested in our laboratory (V.V.O., unpublished data and ref. 21). The differentiation was induced in BPEL medium (BSA polyvinylalcohol essential lipids) 36 , in which polyvinylalcohol is omitted (i.e., B(P)EL) or in the animal product-free equivalent (APEL), which is commercially available from StemCell Technologies.

Fixation
Step 32(xxv) Imaging Imaging Step 32B, zebrafish xenograft model Vascular specification Steps 14 and 15 Pause point (e,f) and vascular specification (g-l). (f,g,i-l) Emergence of mesenchymal cells can be observed on days 3 and 4 of differentiation (f,g) and the first islands of ECs appear between days 6 and 7 of differentiation (i,j); these are further expanded until days 9 and 10 (k,l). Scale bars, 500 µm. SpinEBs 8 . More specifically, hPSCs are maintained on Matrigelcoated plates in mTeSR1 PSC culture medium and they are routinely enzymatically passaged once per week (Fig. 2). On average, one hPSC colony can be divided into 8-10 pieces (Fig. 2a-c) and 5-8 pieces can be distributed over one well of a six-well plate before the induction of differentiation. One well of hPSCs is thus sufficient to make 2-3 six-well plates (12-18 wells) ready for differentiation. Differentiation is induced 4 d after passaging (day 0) (Fig. 2d). Mesoderm specification is induced by the addition of bone morphogenetic protein 4 (BMP4), activin A, small-molecule inhibitor of glycogen synthase kinase-3β (CHIR) 37 and vascular endothelial growth factor (VEGF). Upon mesoderm induction, pluripotent colonies continue to increase in size, and mesenchymal cells are observed at day 3 of differentiation ( Fig. 2e,f). Importantly, we found that the presence of activin A on days 0-3 results in more robust induction of ECs among different hPSC lines (V.V.O., unpublished data). Mesoderm-inductive factors are removed on day 3 of differentiation and are replaced with vascular specification medium supplemented with VEGF and the transforming growth factor-β (TGF-β) pathway small-molecule inhibitor SB431542. SB431542 specifically inhibits activin receptor-like kinase-4, activin receptor-like kinase-5 and activin receptor-like kinase-7 type I receptors (ALK-4, ALK-5 and ALK-7, respectively), and it supports the expansion of ECs by inhibiting the antiproliferative activities of cell-derived TGF-β-like factors that are also present in the culture. The earliest ECs can be observed on days 6 and 7 of differentiation ( Fig. 2i,j). Vascular specification medium is additionally refreshed on days 7 and 9 of differentiation, and isolation of ECs is performed on day 10. Day 10 of differentiation was chosen on the basis of morphological and immunophenotype examination of the differentiated cultures in order to identify the timing that results in the highest percentages and total cell yield of mature vascular endothelial (VE)-cadherin/platelet EC adhesion molecule (also termed CD31)-positive ECs (Fig. 3). Day 10 of differentiation is also more robust for the isolation of ECs should differentiation be delayed.

Isolation and expansion of ECs (Steps 16-29 and
Step 30A). ECs are isolated by using a simple procedure of immunomagnetic selection with anti-CD31 antibody-coupled magnetic beads (Dynabeads). Upon isolation, ECs are transferred to EC serumfree medium (EC-SFM) to which platelet-poor plasma serum    (1% vol/vol), VEGF and basic fibroblast growth factor (bFGF) have been added, as described previously for the derivation of blood-brain barrier ECs from hPSCs 34 . Immediately after isolation, ECs are highly proliferative and tend to reach confluence within 2 or 3 d of plating if they are seeded at a density of (5 × 10 3 ) -(10 × 10 3 ) cells per cm 2 .
Derivation of pericytes (Step 30B). If desired, pericytes can be derived from the CD31 − fraction simultaneously with the isolation of ECs. Pericytes thus derived are initially expanded in endothelial growth medium 2 (EGM-2), and then are transferred to pericyte differentiation medium consisting of DMEM-10% (vol/vol) FBS supplemented with TGF-β and platelet-derived growth factor (PDGF)-BB for 3 d. After 3 d, pericytes are switched to the maintenance medium consisting of DMEM-10% (vol/vol) FBS.

Characterization of ECs and pericytes
At this stage of the procedure, culture of the ECs can be continued by re-plating with a split ratio of 1:3-1:4. Immunotyping should be implemented or their phenotype should be confirmed (Step 31A and B), or cells can be cryopreserved (Step 31C). hPSCderived ECs should display typical endothelial morphology with junctional localization of VE-cadherin, CD31 and von Willebrand factor (vWF) expression and the absence of smooth muscle actin (SMA) (Fig. 4).
Pericytes are routinely maintained in DMEM-10% (vol/vol) FBS, and at this stage in the procedure they can also be additionally passaged, cryopreserved or characterized. At this stage, pericytes appear as a homogeneous population of cells (Fig. 4b).

Assessment of EC functionality
Although morphological and phenotypic characterization of ECs and pericytes is required to confirm the identity and purity of the isolated population (Step 31), this step does not address the functionality of the isolated cells. For many years, the Matrigel tube formation assay has been used as a standard for assessing EC functionality in vitro. However, this assay has a number of disadvantages, and it does not always confirm EC identity, because many other cell types, including mesenchymal cells, are capable of forming tubes on top of Matrigel. Another drawback of this assay and others is the inability to assess whether pericyte cell interactions are functional in vitro. In addition, although mouse models have been developed to evaluate EC, pericyte and SMC functionality in vivo, these are time-, cost-and laborconsuming models that are hard to adapt for drug screening and disease modeling applications. In addition, these methods are incompatible with implementing the 'three Rs' in animal experimentation (reduce, refine and replace).
We describe two independent functional assays that can be used to address these issues and to test vascular cell functionality. First, the hPSC-derived ECs can be cocultured with hPSC-derived pericytes, which can be used to test the functionality of both cell types in vitro. Second, a xenograft model in zebrafish can be used to test the functionality of hPSC-derived ECs in vivo. Coculture of hPSC-derived ECs with pericytes (Step 32A). The coculture assay is performed with ECs and pericytes derived from hPSCs. The assay is based on a previously published assay in which HUVECs were cocultured with pericytes [38][39][40] . We suggest using HUVECs as a positive control for ECs in the coculture experiment. In addition, human pulmonary artery vascular SMCs (PA-vSMCs) can be used as a positive control for pericytes, as suggested in the original report on coculture with HUVECs 39 . However, we noticed that hPSC-derived ECs perform better in the coculture assay than HUVECs. Upon coculture, hPSC-derived ECs first adhere to the substrate and form EC islands surrounded by pericytes that remodel into vascular-like structures. This recapitulates essential steps of primary vascular plexus formation. The formation of a vascular network is highly dependent on VEGF and TGF-β derived from mural cells 21,40 . We have demonstrated that inhibition of TGF-β receptor signaling pathway with SB431542 results in marked enhancement of EC proliferation and sprouting. The EC sprouts induce a contractile phenotype in the cocultured pericytes. We have further shown that this is Notch signaling-dependent 21 . This assay can be useful in studying EC remodeling and endothelialpericyte interactions.
ECs can be visualized in cocultures by immunocytochemical staining with either CD31− or VE-cadherin-specific antibodies; proliferative cells are visualized with the anti-Ki67 antibody, and contractile pericytes are visualized with anti-SM22 antibody.
Moreover, the EC network and the number of branches can be easily quantified with freely available software, such as AngioTool 41 ( Fig. 5a,b). In addition, the number of proliferative nuclei and area covered by SM22-positive pericytes can also be quantified with a custom-developed pipeline based on the CellProfiler software 42 (Fig. 5c,d).
Zebrafish xenograft as a model to assess mammalian EC functionality (Step 32B). Zebrafish xenotransplantation has recently emerged as a valuable assay for assessing tumor cell behavior in vivo, and it is now being widely used in human tumor studies, for example, to model cancer cell metastasis and in vivo cancer drug discovery [43][44][45] .
Zebrafish have multiple advantages: They are transparent so that the vasculature is easily visualized in transgenic strains in which all blood vessels are genetically marked to express green fluorescent protein (GFP) (Tg(fli1:GFP)) 46 . Transplantation of human cells into early embryos does not require immunosuppression, as the immune system only forms in late development. As a result, xenotransplantation has already been successful by using different mammalian and nontransgenic zebrafish cell types [47][48][49][50] .
Imaging can be performed either on live embryos or on fixed embryos. Injection of cells and imaging can also potentially be automated 51 .
One of the most crucial functional features of ECs is their ability to form lumenized blood and lymphatic vessels through which blood or fluid can flow. In addition, in the case of transplantation, it is important that any blood vessels that form can connect or anastomose with the host vasculature. The ability to visualize the formation of vasculature (preferably in a living organism) is then an important aspect of assessing full functionality of any prospective ECs. We have demonstrated that zebrafish xenograft meets these requirements and that it can be used successfully to study EC functionality in vivo 21 . In addition, we found that hPSCderived ECs perform better in zebrafish xenografts than HUVECs, so in principle HUVECs can be used as negative control. We have not observed the incorporation of pericytes into the zebrafish host vasculature.
The assay can be set up relatively easily in a lab that has access to a zebrafish facility, which can provide zebrafish embryos; most of the other steps can be adapted to be carried out in a standard cell culture laboratory with two incubators (set at 28 °C and 33 °C, respectively), a good-quality stereomicroscope, PicoPump and manual manipulator (Fig. 6). In addition, the assay requires relatively small Furthermore, excellent online resources are available for researchers with detailed descriptions of the zebrafish and embryo manipulation steps (https://wiki.zfin.org/display/prot/ ZFIN+Protocol+Wiki).
The zebrafish xenograft assay can be performed in 10 d (Step 32B), and includes the following substeps: Step 32B(i,ii): 24 h to set up breeding and to collect fertilized eggs.
Step 32B(iii-v): sorting of viable embryos at ~6 and 24 h post fertilization (h.p.f.); this can be performed manually or it can be adapted to automated sorting.
Step 32B(vi): sorting of embryos that express GFP 24 h.p.f.; this can be performed manually or it can be adapted to automated sorting.
Step 32B(xxiii-xxv): visualization of EC incorporation into the host vasculature 5 d post injection (Fig. 7). This step can be also adapted to automated imaging. Zebrafish embryos can be imaged live or as fixed preparations for prolonged storage or additional immunocytochemical analysis of cells.     Adjust the pH to ~7. Store aliquots at 20 °C indefinitely (no expiration date). Stock salt solution Dissolve 40 g of Instant Ocean sea salts in 1 liter of distilled water. Store the solution at RT. Egg water Add 1.5 ml of stock salt solution to 1 liter of distilled water with a small amount of methylene blue to change the water color to pale blue. Store the solution at RT indefinitely (no expiration date). mTeSR1 Prepare mTeSR1 medium as shown. Make sure that the lot no. of mTeSR1 basal medium and supplement medium end with the same letter when ordering. Store mTeSR1 for up to 2 weeks at 4 °C or prepare 40-ml aliquots in 50 ml-tubes, and store them at −20 °C for up to 1 year.

Composition
Volume (  Microwave it on high heat, stopping and swirling the mixture throughout the heating process until it is well dissolved (~5 min). Cool the solution until it is safe to handle, and pour it into a Petri dish until the gel is ~5 mm thick.

DMEM-10% (vol/vol) FBS
Cool it at RT until it is set, and then transfer it to a 4 °C refrigerator. Agarose plates can be reused if they are kept air-tight and stored at 4 °C. proceDure passaging of hpscs • tIMInG 3 h 1| Prewarm Matrigel-coated plates, cell culture medium mTeSR1, dispase solution and DMEM/F12 to RT.  crItIcal step All solutions have to be freshly prepared; see Reagent Setup for information regarding storage times.
2| Remove the differentiated parts of the hPSC colonies by scraping them with a 10-µl pipette tip.
3| Aspirate and discard mTeSR1 containing the differentiated parts of the culture.
4| Add 1 ml of dispase solution in DMEM/F12 (1 mg ml −1 ) and incubate it at 37 °C for 3 min.  crItIcal step The timing of the enzymatic treatment varies, so it has to be monitored carefully. The dispase solution should not be added for a prolonged period of time, as it might affect the viability of hPSCs and the outcome of the differentiation experiment.
6| Wash the wells three times with 2 ml of prewarmed DMEM/F12. Do not aspirate DMEM/F12 from the last wash. Concurrently perform Step 7.  crItIcal step The hPSC colonies should not be left in DMEM/F12 for a prolonged period of time, as continued incubation results in re-adhesion of colonies.

7|
Prepare culture plates by aspirating the remaining Matrigel solution from the plates prepared in Step 1 and adding 2 ml of mTeSR1.  crItIcal step Matrigel should be aspirated before the addition of medium in order to avoid drying of the gel layer.  crItIcal step This step should be performed at the same time as Step 6.
8| Scrape hPSC colonies off the plate with a cell scraper.  crItIcal step This step should be performed gently, without breaking colonies into pieces.
9| Collect the pieces with a 1-ml tip and transfer them to a new 15-ml conical tube.
10| Gently break the pieces with a 1-ml pipette tip into small clumps. The size of the pieces should be ~500-1,000 µm, on average (Fig. 2b).  crItIcal step This step is crucial for hPSC viability, as small cell clumps will result in increased death, and bigger cell clumps will have a negative effect on the differentiation efficiency.  crItIcal step Various hPSC lines behave differently, with some dissociating more easily than others.
? trouBlesHootInG 11| Distribute 5-8 pieces per well into the six-well Matrigel-coated plate from Step 7, and then place the plate in an incubator at 37 °C.  crItIcal step During this step, it is important not to put more than eight pieces in each well. It is also crucial to avoid moving the plate while the pieces attach, so that the pieces remain separated from each other.

expansion of hpscs before differentiation • tIMInG 3-4 d 12| 24 h after passaging the hPSC colonies from
Step 11, replace the medium with 4 ml of fresh mTeSR1 and incubate the colonies for a further 3-4 d (Fig. 2c,d).  crItIcal step Routinely, we do not observe much difference between incubating for a further 3 or 4 d before commencing differentiation. However, we would not advise expanding colonies for longer than 3-4 d. Timing is dictated primarily by the convenience of the differentiation starting point. As we routinely passage hPSCs on Thursday, differentiation starts on Monday.
Monolayer differentiation of hpscs • tIMInG 10 d 13| Mesoderm induction from hPSCs (day 0). Replace the mTeSR1 medium with 2 ml of mesoderm induction medium and incubate the cells for 2 d.  crItIcal step Add freshly made growth factors to the differentiation medium.
14| Vascular specification from hPSCs (day 3). Replace the mesoderm induction medium with 2 ml of vascular specification medium and incubate the plate for 4 d.

15|
Expansion of vascular cells (day 7, and day 9 or 10). Replace the medium with fresh vascular specification medium at day 7, and at day 9 or day 10.  crItIcal step EC islands can be observed on days 6 and 7 of differentiation (Fig. 2i,j). It is advisable to monitor the differentiation efficiency both phenotypically and by flow cytometry, using the same procedure as that described in Step 31A.  crItIcal step ECs can be isolated at days 10 and 11 of differentiation. However, we do advise refreshing the vascular specification medium 1 d before isolation of ECs.
17| Calculate the total volume of CD31-Dynabeads needed, on the basis of the following estimation: 7 µl of beads per well of a six-well plate.  crItIcal step We observed that ECs tend to tolerate 5-10 beads per cell well; thus, a higher number of beads per cell is not recommended. The number of CD31-Dynabeads per ml is listed in the product datasheet, and it is equal to 4 × 10 8 beads per ml. Average differentiation with ~20% CD31 + ECs will give ~2 × 10 5 ECs per one well of a six-well plate. Therefore, we propose adding ~7 µl of beads per one well of a six-well plate (or 2-3 × 10 6 beads per one well of a six-well plate), i.e., on average, ~10 beads per cell.
18| Wash the beads twice with 1 ml of DMEM-0.1% (wt/vol) BSA by pipetting them into a 5-ml FACS tube and placing the tube into a small DynaMag-5 magnet. Aspirate the medium with a glass Pasteur pipette and perform a second wash as stated above.
19| Resuspend the beads in approximately three times excess volume (~21 µl) of the initial volume of beads in DMEM-0.1% (wt/vol) BSA.
20| Aspirate the differentiation medium from the cells from Step 15, and wash the cells once with PBS (without Ca 2+ and Mg 2+ ). Add 2 ml of DMEM-0.1% (wt/vol) BSA to the cells and pipette ~21 µl of beads directly to the cells with gentle agitation.
21| Seal the six-well plates with Parafilm and place the plates on the rotating plate machine with gentle agitation (10 r.p.m.).
22| Incubate the cells with beads for 20-30 min at RT. ECs can be identified upon binding of magnetic beads to the cell surface and appearance of a brownish precipitate that also becomes visible by eye. 27| Place the cells on the magnet (DynaMag-5 or DynaMag-15). Collect the supernatant (CD31 − fraction) into a separate tube, put the CD31 − fraction aside and keep it at RT until the end of the isolation. The CD31 − fraction will be used to derive pericytes; see Step 30B.
28| Remove the tube with the beads from the magnet, and resuspend the beads in 2 ml (5-ml tube) or 5 ml (15-ml tube) of the FACSB-10. Place the tube back on the magnet and aspirate the supernatant. Repeat the washing step once more with FACSB-10 and twice more with DMEM-0.1% (wt/vol) BSA.
29| Retain the beads with cells bound, and proceed to Step 30A to use this CD31 + fraction to derive ECs.

expansion of isolated cells 30|
Expand the appropriate fraction from the earlier section to either ECs (option A) or pericytes (option B). The CD31 + cell fraction from the Dynabeads isolation (from Step 29) is expanded to ECs and the CD31 − cell fraction (from Step 27) is used to establish EC and pericyte cell cultures. (x) Place the suspension of fluorescent cells at RT for ~5 min before implantation, and during this incubation proceed with Step 32B(xi). (xi) Implantation procedure. Anesthetize dechorionized zebrafish embryos with 1× tricaine diluted in egg water in a Petri dish that is kept at RT. (xii) Collect 5-10 anesthetized embryos with a plastic transfer pipette and position them on a 10-cm Petri dish coated with 3% (wt/vol) agarose.  crItIcal step Excess water surrounding the embryos can be removed with a tissue. It is important to keep the embryos moist, but too much moisture is detrimental to the injection procedure. It is up to the researcher to find the right moisture balance.  crItIcal step It is best to inject the embryos in groups of ten, as this provides a number that is easily counted and enough embryos to process quickly without the agarose plate drying out. (xiii) Resuspend the cell suspension from Step 32B(ix,x) by repeatedly pipetting with a 200-µl pipette tip in order to break up any clumps of cells. (xiv) Break the tip of the borosilicate glass capillary needles with the stainless steel forceps.
 crItIcal step Note that the needle opening must be large enough for the cells to be implanted without damage, but thin enough so as not to cause harm to the zebrafish. The size of the needle opening must be optimized by the user.  crItIcal step The pressure and time limit can be altered until the correct number of cells is achieved.  crItIcal step Note that it is better to alter the timing of the injections rather than the pressure; zebrafish embryos cannot tolerate very high pressure. (xvii) Perform implantation by using a pneumatic PicoPump and use a manipulator according to the manufacturer's instructions. (xviii) Inject the cells ~60 µm above the ventral end of the DoC, where the DoC opens into the heart (Fig. 6d).
 crItIcal step It is important to recalibrate the number of cells per injection throughout the assay. Owing to gravity, the cells in the suspension will settle toward the tip of the needle and thus alter the number of cells per injection.  crItIcal step Visualize the DoC as a wide stream of blood cells entering the heart (blue dashed line) (Fig. 6d).
Inject ECs above the point that the DoC enters the heart (red circle and arrow) (Fig. 6d). (xix) Remove the embryos from the agarose plate and return them to the fresh egg water with the plastic transfer pipette.
 crItIcal step The zebrafish should recover from the tricaine within 1 min.  crItIcal step Too much tricaine is detrimental to the survival of the fish. Ensure that a 1× working dilution of tricaine is used. (xx) On the same day of injection, examine whether cells are present in the circulation by using fluorescence microscopy. (xxi) Maintain the correctly implanted embryos at 33 °C. (xxii) Keep the embryos in a Petri dish or smaller six-well plates, or place individual fish into a well of a 96-well plate.
Supply plentiful egg water. Refresh the egg water at least every second day. Monitor the embryo survival rate every day.  crItIcal step Experiments should be discarded when the survival rate of the control group is less than 80% over a 6-d period. (xxiii) Analysis on live embryos. Analyze live embryos with a stereo fluorescence microscope at any time throughout the experiment.  crItIcal step It is best to keep the embryos in egg water and to disturb them as little as possible. (xxiv) Fixation of zebrafish embryos. At day 5 post implantation, fix the embryos by overnight incubation in 4% (wt/vol) PFA at 4 °C.  pause poInt Samples can be stored in PBS for up to 6 months at 4 °C. (xxv) Analyze the fixed embryos by using, for example, confocal microscopy or immunohistochemistry.
? trouBlesHootInG Troubleshooting advice can be found in table 1.
• tIMInG See Figure 1 for timing details.

antIcIpateD results
The differentiation process can be monitored as morphological or phenotypic changes in the culture. First, the appearance of mesenchymal cells that migrate out from the hPSCs colonies can be observed on days 3 and 4 of differentiation (Fig. 2f,g). EC islands appear around days 6 and 7 of differentiation (Fig. 2i,j). In addition, the differentiation efficiency can be controlled by flow cytometry. Routinely, 20-30% of VE-cadherin/CD31 + cells can be expected during differentiation at days 7-10 (Fig. 3). The earliest VE-cadherin appears around days 5-7, and at day 10 most VE-cadherin + cells express CD31.
Freshly isolated ECs are highly proliferative. They normally reach confluence 2-3 d after isolation (Fig. 4a). Routinely, we achieve 98% EC purity after one round of selection; this can be observed morphologically (Fig. 4a,c) or by flow cytometric analysis 21 . The expansion of the CD31 − fraction of cells results in the establishment of a homogeneous population of pericytes (Fig. 4b,d); these are negative for EC markers and express PDGFR-β in addition to other pericyte markers 21 .
On average, from two six-well plates of initiating differentiation cultures or two wells of a six-well plate of hPSC culture, we can routinely derive and freeze down up to 2-3 × 10 6 ECs at passage 1, and 3-4 × 10 6 pericytes at passage 3. Cells can be efficiently cryopreserved with a viability >90% without affecting their functionality. ECs should be passaged upon reaching 90-100% confluence at an average ratio of 1:3 (at passage 2 and passage 3) and 1:2 (at passage 4). They can be propagated further for an additional three or four passages until passage 5. However, a slight decrease in proliferation rate may occur after passage 5. Pericytes should be passaged upon reaching ~80% confluence on average 1:3. Pericytes can be propagated until passage 15, although their ability to support EC sprouting tends to decrease after passage 10.
coculture Upon coculture, ECs readily remodel and form well-organized networks on top of the monolayer of pericytes. ECs can be visualized by green fluorescence (in general membrane labeling). We terminate the assay on day 7. ECs are additionally visualized with an anti-CD31 antibody (Fig. 5a), as this makes quantification easier, because general cytoplasmic dyes tend to give patchy staining results. Pericytes recruited to the EC sprouts can be visualized with the anti-SM22 maker (Fig. 5c). Proliferation can be assessed upon visualization with the Ki67 marker.
Quantification of the endothelial network can be done with freely available software AngioTool ( Fig. 5b; ref. 41). The area covered by SM22-positive cells or Ki67-positive proliferative ECs can also be quantified easily with a custom pipeline based on CellProfiler (Fig. 5d; refs. 21,42).

Zebrafish
Within the first few hours after transplantation, red fluorescent ECs will be circulating within the zebrafish's circulatory loop that can be visualized by GFP expressed in the zebrafish vasculature. Note that many cells may still be located within the yolk sac; it is almost impossible to inject all cells into the embryonic circulation. After 24 h, the fluorescent signal from the transplanted cells may fade and disappear completely in many of the zebrafish. This loss of fluorescence in zebrafish is Old EGM-2 medium High pericyte passage (above passage 10) Prepare fresh EGM-2 medium Start up low-passage pericytes from frozen stock similar to the loss of bioluminescence seen in human carcinoma cells injected into nude mice, described by a 2004 study by Rosol et al. 56 . However, as the cells begin to recover and survive in the new microenvironment, bright red fluorescent signal can be observed along the dorsal aorta and caudal vein (Fig. 7a). In some cases, cells can also incorporate into intersegmental vessels or small vessels in the heart and brain (Fig. 7b-d). In our hands, hPSC-derived ECs were able to integrate into embryonic vasculature of the zebrafish. Approximately 80% of injected zebrafish showed the presence of hPSC-derived ECs at day 5 after implantation, whereas only ~20% of zebrafish demonstrated the presence of HUVECs 21 .