Although macrophages are very important for vascular remodeling in the adult, very little is known about their role during vascular development. In the adult, macrophages are involved in angiogenesis, secreting chemoattractant proteins such as vascular endothelial growth factor (VEGF) and producing proteases that degrade the extracellular matrix (Yu and Rak, 2003). Proper vascular development is dependent on the initiation of blood flow (Lucitti et al., 2007). In the adult, changing blood flow patterns results in an increase in monocyte/macrophage recruitment (Schaper et al., 1976). If macrophages are depleted in the adult, remodeling of the vasculature does not occur in response to altered blood flow patterns (Bakker et al., 2008; Nuki et al., 2009; Freidja et al., 2011). These results strongly suggest that macrophages would be essential to vascular development.
Macrophages are the first leukocytes to form during embryonic development and are present as early as embryonic day (E) 8.5 in the mouse (Godin et al., 1995) and Hamburger Hamilton (HH) stage 13 in the chick embryo (Cuadros et al., 1992). The earliest macrophages are referred to as “primitive macrophages.” These cells skip the monocyte stage and differentiate directly into macrophages (Naito, 1993). The second wave of macrophage development begins at E10.5 in the mouse and results in the formation of “definitive macrophages” (Naito et al., 1990). The gene expression of mature and primitive macrophages differs significantly. Mouse primitive macrophages do not express “typical” macrophage markers, such as F4/80, LysM, or PU.1, but they do express Csf1r, the macrophage mannose receptor, and CD11b (Lichanska and Hume, 2000). Because the gene expression profile of primitive and definitive macrophages differs so dramatically, no genetic models have successfully ablated both cell populations (Lichanska and Hume, 2000). Because no null models exist, it is essential to develop tools to modify and study the role of macrophages in early vascular development.
We have developed time-lapse techniques to simultaneously image endothelial cells and macrophages that allow us to study the interaction of macrophages with blood vessels during development. Traditional in vivo time-lapse microscopy relies on transgenic animals in which a fluorescent protein is expressed in a subset of cells. Although transgenics are a powerful tool, the requirement to breed the fluorescent transgene has limited its application in genetic ablation studies. Furthermore, crossing one fluorescent transgene onto a knockout background is a lengthy process such that models with two fluorescent transgenes are rarely made. We have developed techniques to label both endothelial cells and phagocytic cells during vascular development in an avian model. Our technique involves the injection of fluorescently labeled acetylated low-density lipoprotein (AcLDL), which is specific to endothelial cells and macrophages, and used a red fluorescent phagocytic cell labeling dye, PKH26-PCL, to label macrophages. AcLDL is taken up by scavenger receptors on endothelial cells and macrophages (Brown et al., 1980). Injection of AcLDL has previously been used to label endothelial cells during avian development (Hallmann et al., 1987). PKH26-PCL is a vital dye made of aggregate particles such that only phagocytic cells take up the dye (Bellingan et al., 1996). Although our time-lapse technique was developed in an avian model, it is equally applicable to other models, such as mouse.
Labeling of Endothelial Cells by Alexa Fluor AcLDL and Macrophages by Alexa Fluor AcLDL and PKH26-PCL in Live Avian Embryos
Endothelial cells have receptors that are capable of taking up AcLDL (Stein and Stein, 1980). We first sought to confirm that AcLDL covalently labeled with Alexa Fluor (AF) fluorescent dyes can label the entire vascular endothelium after intravascular injection of the quail embryo. We injected red AcLDL into the sinus terminalis of Tg(tie1:H2B-eYFP) quail embryos. In these transgenic embryos, the H2B tag localizes yellow fluorescent protein (YFP) to the nucleus of Tie1 (an endothelial marker) -positive cells (Sato et al., 2010). During early stages of vascular development, cells labeled with red AcLDL were consistently observed to have a nucleus that expressed YFP in the entire extra-embryonic yolk sac (Fig. 1A). We also observed red AcLDL showing the extension of a sprouting endothelial cell (Fig. 1B), which could not always be observed in the Tg(tie1:H2B-eYFP) quail embryos (Fig. 1C).
Because AcLDL can also be taken up by macrophages (Brown et al., 1980), we wanted to see if we could label macrophages separately using a second dye, PKH26-PCL, that is used to label phagocytic cells. Although PKH26-PCL is an established marker of phagocytic cells in full-grown animal systems (Maus et al., 2001; Bellingan et al., 2002), to our knowledge, it has never been used on avian embryos at the early stages of embryonic vascular development. When the PKH dye is mixed with Diluent B, it forms into small aggregates that are engulfed by phagocytic cells while inhibiting their uptake by nonphagocytic cells. Aggregates that are not engulfed by phagocytic cells remain fluorescent and in circulation. To ensure that the dye was actually being taken up by cells, we harvested cells from whole HH22 embryos injected with PKH26-PCL at HH18, a stage where the existence of numerous cells with macrophage-like characteristics has been previously established (Cuadros et al., 1992). These cells were also labeled with green AcLDL. From this cell suspension, we isolated cells using Dynabeads that were coated with the QH1 antibody (Fig. 1D,E). QH1 is an antibody that is specific to quail endothelial and hematopoietic cells such as macrophages (Pardanaud et al., 1987; Cuadros et al., 1992, 1993). Many of the QH1-positive cells were labeled with green AcLDL (Fig. 1D). We observed that some of these cells were also labeled by PKH26-PCL (Fig. 1E), confirming that the aggregates of dye were being taken up by QH1+/AcLDL+ cells in the embryo. We repeated this experiment and imaged cells from HH18 quail embryos that were injected with PKH26-PCL at HH12–HH13, a stage just before the observed appearance of macrophage-like cells (Cuadros et al., 1992). A similar labeling pattern of cells was seen at this younger stage (Fig. 1F,G). We also imaged the cells that were QH1-negative (cells that did not attach to the antibody coated beads). A few of these cells were labeled with only AcLDL and rarely, a cell was observed to be double-labeled by both AcLDL and PKH26-PCL. However, very few labeled cells not attached to the beads were seen (data not shown). Because AcLDL is known to bind to receptors on endothelial cells or macrophages, which are QH1-positive cell types, we can only attribute this to the fact that Dynabeads are not 100% efficient (Gomm et al., 1995).
The only leukocytes known to be present at the stages of development that we investigated are macrophages and monocytes (Cormier et al., 1986). Granulocytes are first detected at HH36 (or 10 days of incubation) in chicken embryos and lymphocytes are first present at HH40 (or 14 days of incubation) (Nicolas-Bolnet et al., 1991). We confirmed, nonetheless, that AcLDL labeled only macrophages and not other leukocytes such as neutrophils. We could not collect enough embryonic blood and, therefore, tested the ability of AcLDL to stain leukocytes in adult chicken blood. We isolated mononuclear cells and stained them with AcLDL. In the mononuclear cell isolation, 7.4% of cells were labeled by AcLDL. By hematoxylin and eosin staining, we confirmed that macrophages were present in the isolated cell population. The morphological staining could not be performed on the same cells as were stained with fluorescent AcLDL because hematoxylin, eosin, and Wright's are themselves very fluorescent. We then performed a Ficoll separation to isolate neutrophils and stained this sample as well with AcLDL. One cell (out of 204 counted cells) was labeled with AcLDL after Ficoll separation. The absence of macrophages in this cell separation was confirmed by hematoxylin and eosin staining.
We also confirmed by immunohistochemistry that PKH26-PCL-labeled cells were macrophages. We used antibodies directed against Sialyl Lewis X (also called CD15s) that labels granulocytes and macrophages (Terstappen et al., 1992). We found that PKH26-PCL-labeled cells also stained positive for CD15s (white arrow, Fig. 1H,I). Occasionally, a PKH26-PCL was stained negative for CD15s (orange arrowhead, Fig. 1H,I). Not all CD15s-positive cells, however, were PKH26-PCL labeled (blue arrowhead, Fig. 1H). LEP100 and AcPase staining are commonly used to identify macrophages in the literature. The fixation protocol for both requires the use of alcohols that quench fluorescence and, therefore, could not be used. We attempted to modify the fixation and staining protocol to work with fresh or paraformaldehyde-fixed cryosections; these stains were not successful. We also injected DiO-labeled liposomes encapsulated with phosphate buffered saline (PBS). We observed cells labeled singly by PKH26 and by DiO liposomes (data not shown), as well as double-labeled cells (Fig. 1J). The DiO-liposome and CD15s staining indicate that PKH26-PCL-labeled cells are positive for macrophage markers but only a subset of the macrophage population is labeled.
We also injected embryos with PKH2-PCL (Sigma), the green fluorescent equivalent of PKH26-PCL, to see if we could label and image macrophages in conjunction with red AcLDL. We found that cells labeled using the green PKH dye were not as bright, as previously reported (Oh et al., 1999), and would bleach and lose fluorescence after only a few minutes of imaging.
Phagocytic Cells Are Present in the Vessel Wall and Express Tie1
After confirming that PKH26-PCL was labeling QH1+, AcLDL+, CD15s+ cells in the avian embryo, we next imaged the vasculature of the embryo double-labeled with green Ac-LDL and PKH26-PCL. We injected quail embryos at stage HH12–HH13 with a 1:1: mixture of AcLDL and PKH26-PCL and imaged the extra-embryonic yolk sac at HH18 (Fig. 2A). We observed double-labeled cells in the lumen of blood vessels that appeared adherent to the endothelium (white arrow). Many of these cells were circular in shape. PKH26-PCL also appeared to label cells that were integrated into the vessel wall (Fig. 2A, blue arrowhead). These elongated labeled cells were the most common observed phenotype of imaged macrophages. In addition, we noticed some circular shaped cells attached to the endothelium that were only labeled with green AcLDL (blue arrow).
The PKH26-PCL-labeled cell in the wall of blood vessels was an unexpected observation. We used the tie1:H2B-eYFP transgenic embryos to establish whether these cells expressed endothelial cell markers. Most circulating PKH-PCL26-labeled cells were Tie1-negative (white arrow, Fig. 2B). We did, however, observe one double-labeled circulating cells during our time-lapse microscopy (n = 5 for time-lapses in Tg(tie1:H2B-eYFP)). Most mural PKH26-PCL-labeled cells were Tie1-positive (Fig. 2C,D). Occasionally, PKH26-PCL-labeled mural cells appeared to be Tie1-negative (Fig. 2D, orange arrow). Because of the density of endothelial cells in the vessel wall, we were never fully convinced that PKH26-PCL labeling was not associated with a proximal Tie1-positive nucleus.
We next time-lapsed the extra-embryonic yolk sac of HH18–HH19 quail embryos that were injected with a mixture of green AcLDL and PKH26-PCL at HH12–HH13 and studied the behavior of the double-labeled cells (n = 11). We focused on the regions near the vitelline arteries close to the embryo proper. We noticed consistent behavior patterns of macrophages with respect to the vasculature (Figs. 3, 4; Supp. Movie S1, S2; which are available online). Double-labeled cells were often observed attached to the endothelium. Some of these cells were rolling along the endothelium in the direction of circulation (Fig. 3B,C). Some of these cells would disappear suddenly from one frame to the other (Fig. 3D,E, white arrow). These macrophages appeared to detach themselves from the endothelium and re-enter circulation. In only one time-lapse (of a total of 11 with AcLDL) did we observe double-labeled cells localized to a region in which a new blood vessel was forming by angiogenic sprouting. Many macrophages were immobilized in the vessel wall throughout the entire time-lapse (Fig. 3D,E, blue arrowhead). We also observed a rare occurrence of a macrophage migrating out of the vessel and entering the nonvascular region of the yolk sac. Within the time of imaging, this cell remained, hovering around the vessel (Fig. 4B–E). These macrophage interactions of attachment, rolling, firm adhesion, and transmigration with the endothelium were reminiscent of extravasation behavior during an inflammatory response (Middleton et al., 2002).
We also frequently observed brightly fluorescent circulating cells that were labeled with AcLDL but not with PKH26-PCL. The cells were plentiful in circulation and exhibited similar behavior to the double-labeled cells. Some circulated freely, whereas a subset adhered to the endothelium and transmigrated. Because PKH26-PCL labels only a subset of macrophages, we cannot preclude the possibility that these are unlabeled macrophages. They may also represent circulating endothelial cells.
Phagocytic Mural Cells Can Be Recruited to Sites of Vascular Remodeling
Double-labeled cells that were integrated in the vessel wall were always immobile in our time-lapse movies. We investigated whether the mural PKH26-PCL-labeled cells could be recruited to sites of vascular remodeling. Both macrophages and circulating endothelial cells are known to be home to regions of injury and remodeling (Schaper et al., 1976; Pardanaud and Eichmann, 2006). We adapted a previously published technique to induce vascular remodeling in the capillary plexus of avian embryos (le Noble et al., 2004). We obstructed flow in one of the two vitelline arteries of the yolk sac by placing a small metal wire over the vessel 16 hr after injection of PKH26-PCL. Control embryos were injected with PKH26-PCL, but flow was not obstructed. We imaged the vasculature of the embryos on both sides of the yolk sac 9 hr later. We found that there was an increase in the number of PKH26-PCL-labeled cells on the side of the yolk where we had induced remodeling and an equivalent decrease in the number of labeled cells on the contralateral side as compared to control (Fig. 5; n = 10 for control, n = 6 for obstructed).
In this study, we have developed a method of imaging vessels and macrophages during early vascular development through the use of two fluorescent vital dyes: covalently labeled AF488-AcLDL and PKH26-PCL. Our method, which involves intravascular injection of fluorescent dyes, is much easier, less expensive, and less time consuming than previously used techniques. Our method can easily be applied to image and study vascular development in knockout animals without needing to crossbreed them with a fluorescent transgenic line. In addition, we show that this technique labels not only existing vessels but also sprouting endothelial cells. The co-label with PKH26-PCL, allows us to identify macrophages during development and study their behavior. This method can also be applied to other vertebrate models such as mice.
Live imaging of macrophages in an embryonic vertebrate model system has so far only been done on zebrafish embryos. Through live imaging, it was observed that macrophages developed before circulation is established and before the erythroblasts arise. These macrophages were seen in circulation, sometimes phagocytizing erythrocytes. In the same study, it was shown that these macrophages responded to bacterial infection after both intravascular and nonvascular injection of bacteria (Herbomel et al., 1999). In another experiment, transgenic zebrafish whose endothelial cells produced green fluorescent protein (GFP) were crossbred with a transgenic whose macrophages produce red fluorescent protein (RFP). Live imaging of tissue macrophages showed that macrophages were involved in anastomosis of growing sprouts (Fantin et al., 2010). We rarely observed an association of macrophages with vascular sprouting. Previously published results looked at mature macrophages that represent a later stage of development. Furthermore, we are not labeling all macrophages. Therefore, we do not believe that our results negate a role for macrophages in angiogenesis during development. In fact, our results showing that these cells are recruited when flow through the vitelline artery is ablated indicate that they are homing to sites of hypoxia.
In the avian embryo, the first observation of phagocytic cells during early development was seen using QH1 and MB1 antibodies, markers of quail hemangioblastic lineage (Peault et al., 1983; Pardanaud et al., 1987), and acid phosphatase (AcPase) labeling. AcPase is an enzyme found in many cell types that is involved in the hydrolysis of orthophosphate monoesters under acidic conditions. AcPase in phagocytic cells like macrophages is tartrate resistant (Bull et al., 2002), and thus sodium tartrate can be used to distinguish these cells. In both chick and quail embryos, labeled free cells are identified as early as 20 somites (HH13–HH14), whereas cells with macrophage morphology are only observed after HH15 (Cuadros et al., 1992). While many of these cells are seen in regions of cell death such as the limb buds and tail regions, their presence in the developing embryo is observed before cell death processes occur. In the developing avian eye, macrophage invasion in the fibrous matrix of the cornea is seen at HH22–HH23 before the emergence of endothelial cells and is believed to play a role in conditioning the matrix and facilitate blood vessel formation (Bard et al., 1975). Our data agree with previously published results showing the presence of macrophages early in vascular development. We find, however, that the majority of circulating macrophages remain in the vessel wall or in the perivascular area, alluding to a largely vascular function in the yolk sac at this stage.
During arteriogenesis in the adult, mononuclear cells accumulate in the tunica adventitia (Schierling et al., 2009). Macrophages degrade the extracellular matrix, allowing the vessel to expand in diameter. Although the vast majority of PKH26-PCL-labeled cells remain near or part of the vessel wall in our experiments (Figs. 1H, I, 2, 3, 4), few of them are actually in the adventitia around blood vessels. High magnification images (Fig. 2A, blue arrowhead) show that some cells are incorporated in the wall of the vessel similar to endothelial cells. Similar results have been observed in the developing nervous system of quail embryos at HH17 using AcPase to stain for macrophages (Cuadros et al., 1993). In the adult liver, a type of macrophage called a Kupffer cells is present in the blood vessel wall. These cells send out processes into the vascular lumen and function to clean circulating bacterial and foreign proteins from the blood. Both because of the location of these cells (i.e., the vitelline artery) and the morphology of PKH26-PCL-labeled cells in the embryo, we do not believe that these are Kupffer cells. During embryonic development, Tie1/Tie2 are expressed by hematopoietic stems cells in the AGM and fetal liver (Takakura et al., 1998; Hamaguchi et al., 1999). The vascular endothelium in the embryo has been shown to have hematogenic potential (Chen et al., 2009), and these cells could represent a component of the hemogenic endothelium. An alternative possibility is that PKH26-PCL cells are of myeloid origin but have trans-differentiated into endothelial or endothelial-like cells. When CD31−/CD14+ cells derived of peripheral blood are cultured in “pro-angiogenic” media, they begin to express endothelial cell markers as well as display endothelial cell behaviors, such as the ability to form tube-like structures when cultured in matrigel. This finding has led to the hypothesis that monocytes have endothelial progenitor potential (Rehman et al., 2003). Because hematopoietic stem cells are not known to be phagocytic, we favor this latter theory. Our results with vascular injury show that these cells can be recruited and, therefore, retain the ability to re-enter circulation. We cannot, however, negate the possibility that hematopoietic stem cells/circulating endothelial cells have phagocytic capacity, and it is known that both these cell types are recruited to sites of vascular injury and remodeling as well in the embryo (Pardanaud and Eichmann, 2006).
Our results show that we can successfully identify and time-lapse macrophages during early vascular development. We show that these cells adhere, roll, and extravasate from the blood vessels during development. The majority of cells appear to be integrated into the blood vessel wall and have a morphology similar to adjacent endothelial cells. Although these cells are largely immobile during normal vascular development, they can be recruited when vascular remodeling is induced, as is seen for both macrophages and circulating endothelial cells (Pardanaud and Eichmann, 2006). We believe that this may indicate the ability of circulating macrophages to assume an endothelial-like phenotype, however, we cannot rule out that these cells represent hematopoietic stem cells or circulating endothelial cells.
Intravascular Injection of Fluorescent Labels
Fertilized quail eggs (Coturnix japonica) were incubated at 36–38°C at approximately 60% humidity for 48 or 72 hr, as noted. Both normal and Tie1 transgenic (tie1:H2B-eYFP) quail embryos were used (Sato et al., 2010). At the desired stage, 1 ml of albumin was removed from the egg and a small window was cut out of the top of the shell to uncover the embryo. Embryos were then injected in the sinus terminalis with fluorescent dye solutions using pulled quartz needles attached to a PICOSPRITZER III micro-injector (Fig. 6A). PKH26-PCL stock dye (Sigma-Aldrich, 1,000 μM) was diluted to 100 μM with ethanol to form the working stock solution. Just before injection, PKH26-PCL was freshly prepared by combining 1 μl of working stock with 99 μl of Diluent B (Sigma-Aldrich), a reagent included in the phagocytic cell-labeling kit. The mixture was vortexed, centrifuged, and then left to rest on the lab bench for 30–60 min before it was injected into the embryo. Both green (Alexa Fluor 488) and red (Alexa Fluor 594) AcLDL were used as noted (Invitrogen) and injected as provided in the commercial stock. When both green AcLDL and PKH26-PCL were injected, a 1:1 solution was made. For all injections, a small amount of blue food dye was added to each dye mixture to enable us to detect successful intravascular injection. After injection, embryos were rehydrated with Ringer's solution, then the eggs were sealed with either Parafilm or scotch tape and were returned to the incubator.
Isolation of QH1+ Cells
Embryos injected with PKH26-PCL alone were dissected and broken down into a cell suspension by treating individual embryos with a 1 mg/ml solution of collagenase for 1 hr at 37°C. Between 5 and 10 μl of green AcLDL was then gently mixed with the cell suspension for 10 min. Cells were isolated using QH1 coated Dynabeads (Iowa Developmental Hybridoma Banks, Invitrogen) and imaged immediately by confocal microscopy.
In Vivo Fluorescent Labeling of Whole-Mount Embryos
Embryos injected with fluorescent dyes were dissected the next day and transferred to 35-mm glass bottom Petri dishes where they were imaged whole-mount under the confocal microscope.
Embryos injected with only PKH26-PCL were re-injected the next morning with DiO-labeled liposomes-encapsulated PBS (Dr. Nico van Rooijen). After 4–6 hr, the embryos were dissected and transferred to 35-mm glass bottom Petri dishes for imaging.
Whole chicken blood isolated with EDTA was obtained from Lampire Biological Lab. A total of 5 ml of blood was centrifuged at 2,400 rpm for 15 min in CPT tubes (BD Bioscience). Mononuclear cells, which separate as a buffy layer with these tubes, were collected and stained with 5 μl of stock AF488-AcLDL per ml of cells for 15 min. Cells were washed twice with 1 ml of PBS and then imaged with a 20× objective lens for both fluorescence (AcLDL labeling) and white light (total cell count).
Neutrophils were isolated by Ficoll separation. In a 50-ml Falcon tube, we layered 12 ml of Ficoll 1119, covered by 12 ml of Ficoll 1077, which was covered by 24 ml of avian blood. We then centrifuged the sample at 600 g for 30 min. The layer just above the packed red blood cell layer was collected and stained with AcLDL as we did for mononuclear cells. Cells were imaged at the same settings as for mononuclear cells.
Embryos were injected with PKH26-PCL at HH12–HH13 as described above and dissected at HH18. A small piece of the yolk sac surrounding the vitelline artery was dissected (∼20 mm2) and fixed for 15 min in 4% paraformaldehyde. The tissue was placed in optimal cutting temperature (OCT) compound and frozen immediately. Seven-micrometer sections were cut using a cryostat. The sections were then blocked with Tris-NaCl-Blocking Reagent (Roche) for 1 hr. The sections were incubated with primary antibodies diluted in blocking against CD15s (BD Bioscience) at a 1:100 dilution. The sections were washed 5 times for 5 min with Tris-NaCl-Tween and then reblocked for 1 hr. AF488-labeled secondary antibodies (Invitrogen) were used at a 1:400 dilution. Sections were washed 5 times for 5 min with Tris-NaCl-Tween. Sections were then imaged on a Zeiss LSM Exciter with either 488 nm or 543 nm excitation separately to verify no cross-talk between the signals.
Time-Lapse Live Imaging
Embryos were injected with the PKH26-PCL/AcLDL fluorescent dye mixture at 48 hr of incubation. The next day, they were removed from their shell by carefully pouring the entire contents of the egg into a 60-mm Petri dish, ensuring the embryo remained on top and the yolk was left intact (Fig. 6B). Embryos were rehydrated with Ringers both before and after the transfer. A window large enough to expose the embryo was formed in the lid of the Petri dish. A small hole, the size of a syringe needle, was also made next to the window. The window was covered with a thin Teflon membrane (YSI Incorporated) that was sealed to the insides of the lid by framing the edge of the cut out window with high vacuum grease (Dow Corning) (Fig. 6C). The Teflon membrane allows air to flow through, but is impermeable to liquid, thus preventing the embryo from drying out during imaging (Kulesa et al., 2010). The Teflon covered lid was then placed on top of the Petri dish, aligning the window with the position of the embryo. The lid was then sealed to the dish by carefully wrapping the dish with Parafilm (Fig. 6D). Preheated albumin (37°C) was added into the Petri dish using a syringe through the small hole next to the window in the lid until the embryo surfaced to the Teflon membrane, at which point the hole was quickly sealed with grease. Embryos were then moved to a temperature-controlled, upright fluorescence microscope and imaged once every 10 min for a maximum period of 10 hr (Fig. 6E).
Vascular Injury Model
Eggs were incubated for approximately 48 hr until the embryos had reached HH12–HH13. The eggs were opened and embryos were injected with PKH26-PCL as described above. The eggs were resealed and returned to the incubator for 16 hr. The embryos were then poured into Petri dishes. A wire (0.5-mm diameter) was fixed to lid of the Petri dish in such a way that it formed a loop. Once the lid was positioned on the Petri dish, this wire loop caused an indentation in the yolk sac, inhibiting blood flow through one of the two vitelline arteries. The dishes were then placed in a Tupperware that contained 2–3 cm of water and wicks to increase ambient humidity. The Tupperware was placed in an egg incubator at 37°C. Control embryos were injected with PKH26-PCL, poured into a Petri dish, and placed in the Tupperware in the absence of the wire loop. Embryos were allowed to develop for 9 hr and then two fluorescent images were taken (one image of each of the vitelline arteries). From these images, the number of fluorescent cells was counted manually.
S.A.R. was supported by the Eugenie Lamothe Fellowship.