Fli + etsrp + Hemato-Vascular Progenitor Cells Proliferate at the Lateral Plate Mesoderm during Vasculogenesis in Zebrafish

Background Vasculogenesis, the de novo formation of blood vessels from precursor cells is critical for a developing embryo. However, the signals and events that dictate the formation of primary axial vessels remain poorly understood. Methodology/Principal Findings In this study, we use ets-related protein-1 (etsrp), which is essential for vascular development, to analyze the early stages of vasculogenesis in zebrafish. We found etsrp + cells of the head, trunk and tail follow distinct developmental sequences. Using a combination of genetic, molecular and chemical approaches, we demonstrate that fli + etsrp + hemato-vascular progenitors (FEVPs) are proliferating at the lateral plate mesoderm (LPM). The Shh-VEGF-Notch-Hey2 signaling pathway controls the proliferation process, and experimental modulation of single components of this pathway alters etsrp + cell numbers at the LPM. Conclusions/Significance This study for the first time defines factors controlling proliferation, and cell numbers of pre-migratory FEVPs in zebrafish.


Introduction
In vertebrate embryonic development the blood and endothelial cells emerge side-by-side from a hypothesized common precursor termed, ''hemangioblast [1,2].'' In zebrafish and xenopus, it has been established that transcription factor fli1 is required for both primitive and definite hemangioblast formation [3], and acts at the top of the transcriptional network dictating blood and endothelial cell development. Recently, in zebrafish, a transcription factor of the ets family, ets1-related protein (etsrp) has been identified as a key regulator of vasculogenesis [4]. Further, etsrp was shown to be required for the formation of myeloid but not erythroid cells [5]. Both fli1 and etsrp are expressed at zebrafish trunk LPM at 10 som when the putative angioblasts are specified from mesoderm cells, and also at 26 som when the axial vessel is formed at the midline [3,4]. Sufficient numbers of hemangioblasts or angioblasts are necessary to generate a primary axial vessel. Cell proliferation is one means of attaining the required cell number. However, in zebrafish this mechanism has been considered unlikely to contribute substantially to the precursor pool because of the short temporal window in embryonic vasculogenesis (,7 hours), and proliferation process especially of differentiated cells takes 16-24 h for one cell cycle. Because fli1 acts at the top of the transcriptional network driving blood and endothelial cell development, and is sufficient to induce expression of key hemangioblast genes such as etsrp [3], we hypothesized that fli + etsrp + hemato-vascular progenitors (FEVPs) at the 10 som LPM proliferate prior to migration to the midline. To investigate this hypothesis, we performed detailed in situ hybridization (ISH) analysis for etsrp in zebrafish embryos of early somite stages, and counted etsrp cells between 4 and 8 som (,2 h window), which showed a two-fold increase in number of cells. Next, we performed pulse chase, chemical treatment and genetic mutant experiments to investigate whether FEVPs proliferation occurred at the LPM. Classic bromodeoxyuridine (BrdU) incorporation analysis, phospho-histone3 immunostaining and hydroxyurea (HU) (S-phase inhibitor) treatment showed that FEVPs are proliferating at the LPM. Further, we observed reduced numbers of FEVPs in gridlock (grl) knockdown embryo, which can partly be attributed to reduction in cyclin protein levels. Grl/hey2, a Notch target gene was modulated by a variety of approaches, which lead to the expected changes in FEVPs numbers at the LPM. In vitro, grl knockdown (KD) in differentiated venous endothelial cells (ECs) resulted in minimal changes in cell cycle. Finally, we determined that modulating each component of the Shh-VEGF-Notch-Hey2 signaling axis results in altered numbers of FEVPs indicating that these signals control precursor cell number. Also, we present evidence suggesting that the hypochord, a transient midline structure is the source of VEGF. Together, this study identifies that the canonical Shh-VEGF-Notch-Hey2 signaling axis as responsible for FEVPs behavior, and generation of appropriate numbers through proliferation in a developing vertebrate embryo.

Results
Distinct etsrp + cell behavior in head, trunk and tail of embryonic zebrafish To investigate vasculogenesis in vivo, we performed a detailed ISH for ets1-related protein (etsrp), a marker for vascular and some hematopoietic precursors in vertebrates [4]. We performed etsrp ISH on embryos fixed at approximately every two hours in development starting at 4 somites (som) until 30 hours post fertilization (hpf) stage ( Fig. 1). We noticed three distinct events in the head, trunk and tail regions of the developing embryonic zebrafish vasculature with two groups within the trunk zone.
First, in the head two lateral stripes of etsrp + cells (Fig. 1A, 4 som, yellow asterisk) are connected in the anterior region with a bridge of cells (Fig. 1A, 4 som). At 12 som, two populations of etsrp + cells one from the anterior bridge (Fig. 1A, 8 or 12 som, green asterisk), and another from the middle part of the lateral stripes (Fig. 1A, 12 som, white asterisk), showed an inward trajectory resulting in four cephalic patches of etsrp + cells (Fig. 1A, 14 som, red arrows) at 14 som. The anterior most cells do not undergo substantial changes as the embryo develops until 26 som, but cells in the posterior patches become consolidated into a confined cluster, and presumably contribute to the head vasculature.
Second, in the trunk (Figs. 1B-C, 4 and 8 som), we noticed two lateral stripes of etsrp + cells evenly distributed on either side of the midline. At 8 som, an asymmetric break is noticed (Fig. 1B, 8  In the tail tip region, a third population of etsrp + cells (Fig. 1D, 16 and 18 som, asterisk) gathers, which occurs around 16-18 som stages. It is noteworthy that the trunk etsrp + populations (Fig. 1C, 14 som, pink arrow) observed at the midline in 14 som are improperly assembled at the midline, and at this time point the tail populations are not congregated at the tip as observed in later time points of 16 and 18 som (Fig. 1D, 16 and 18 som, pink and green arrows). At 18 som, when the tail etsrp + populations are aggregated, the etsrp + cells at midline appear to properly assemble in a linear fashion along the anterior-posterior axis, the future site of axial vessels. At prim-6 ( Fig. 1, prim-6, yellow arrow) and 30 hpf (Fig. 1,  30 hpf, yellow arrow), we notice remnants of the tail etsrp + population (Fig. 1, prim-6 and 30 hpf, asterisks). Lateral and dorsal views of the entire process are captured in Figs. 1E and 1F respectively. The dorsal view also illustrates that the head and trunk etsrp + population do not fuse throughout development.

FEVPs proliferate at the LPM and midline
The observations on embryonic expression of etsrp suggest that etsrp + cells in the head, trunk and tail zones show distinct and coordinated behaviors. We furthermore noted a sizeable increase in etsrp + cell numbers in between 4 and 12 som ( Fig. 2A-C). To quantify this observation, we counted the number of etsrp + cells at 4 and 8 som ( Figs. 2A & B, inset red dots). These counts showed a 2fold increase (Fig. 2D, red bar) in absolute numbers of etsrp + cells. We further quantitated whole embryo etsrp mRNA level using qPCR. Between 3 to 10 som, we observed a 22-fold increase of etsrp + levels in the embryo (Fig. 2D, blue bar). To examine if the developmental increase in etsrp + pools requires proliferation, we performed bromodeoxyuridine (BrdU) incorporation assay to detect cells undergoing DNA synthesis. Embryos incubated with BrdU from 1 to 10 som showed robust BrdU incorporation at the midline (Fig. 2E), and a weaker signal overlying the presumptive LPM (red arrows, Figs. 2F and 2G). Developing somites serve as an internal BrdU positive control at the midline (black arrow, Fig. 2G).
Since Fli is upstream of etsrp in the transcriptional hierarchy, and both participate in hematopoiesis and vascular development, we refer to the population of cells in this study as Fli + etsrp + hematovascular progenitor cells (abbreviated as FEVPs). These cells are vascular precursor populations, which retain some hematopoietic potential. To identify the proliferating cells in LPM observed in 10 som embryos (Figs. 2F & 2G), we switched from BrdU to fluorescent-conjugated ethynyl deoxyuridine (EdU) because it provides the best option for efficient detection of incorporated EdU. EdU allows for coupling reaction with fluorescent azide dyes, and also allows for use of milder conditions (no DNA denaturation process). We performed IF for green fluorescent protein (GFP) (Fig. 2I) and fluorescent-conjugated EdU (Fig. 2J), incorporation [6] on sections from 10 som transgenic embryos that carries friend leukemia inhibitor (fli-1a) protein promoter driving nuclear localized enhanced GFP (nEGFP) Tg(fli1a:nEGFP) [7] in the vasculature. As shown in Fig. 2I, roughly 2-4 GFP + cells are visible in the same location as etsrp + cells in the 10 som ISH embryos (Fig. 2H). Importantly, 1-2 cells out of the GFP + population are EdU + (Fig. 2J, white arrows) suggesting that a sub-population of FEVPs were proliferative. To conclusively determine if the proliferative cell was indeed an endothelial precursor cell, we performed co-staining for phospho-histone H3 (p-H3) and GFP protein in Tg(fli1a: EGFP) 10 som section (Fig. 2K). We confirmed these results using double staining for p-H3 and GFP protein on the same Tg(fli1a: EGFP) 10 som section ( Fig. 2K-L, arrows).
Etsrp WISH observations, qPCR message and cell quantitation measurements suggest a rapid increase in this precursor population at the same time that multiple mitotic markers are detected on Fli expressing vascular cells. Taken together, these data suggest that FEVPs proliferate at the LPM.

Expression of the Shh-VEGF-Notch-Hey2 pathway components is consistent with their action at LPM during vasculogenesis
The sonic hedgehog (Shh)-VEGF-Notch-Hey2 signaling pathway is well established for artery vs. vein (A/V) specification during embryonic vascular development in vertebrates [8,9]. We investigated whether this signaling pathway may be active during vasculogenesis at the LPM. We first investigated whether components of this pathway are expressed appropriately to influence vascular development at the LPM. To determine endogenous hey2/grl expression, we performed WISH study of grl expression during embryonic zebrafish development ( Figure S1). Grl expression is observed only at the LPM as shown previously [10], and at the midline ( Figure S1). Notch1 and Notch4 are the endothelial specific Notch [11]. In zebrafish, notch1b expression was intriguingly similar to etsrp expression at 10 som in the LPM [12]. ISH for endogenous shh ( Figure S2A-F) showed that at bud stage ( Figure S2C & F) and 3 som ( Figure S2B & E), diffuse midline expression was observed, suggesting that presumptive notochord cells express shh. At 10 som, shh was expressed in midline notochord structures as shown ( Figure S2A & D), and demonstrated in previous reports [13]. ISH for vegf (both isoforms) showed expression in somites and endoderm at 10 som ( Figure  S2G, inset & H) [14]. IHC with VEGF antibody clearly showed VEGF protein expression adjacent to hypochord in ventral somites ( Figure S2I, yellow arrow). Taken together, all components of the Shh-VEGF-Notch-Hey2/grl pathway are expressed temporally, and spatially at the LPM or in tissues adjoining it during stages in which endothelial precursor cell numbers are generated.
Gridlock (hey2) KD embryos show fewer FEVPs at the LPM We next focused on addressing whether the Shh-VEGF-Notch-hey2/grl pathway influenced the FEVPs population at the LPM, beginning with the most downstream component, hey2/grl.
Using morpholinos (MOs) and a dominant negative approach, we investigated whether FEVPs cell numbers were affected by loss of grl/hey2 function. We first injected grl MOs into Tg(fli1a:EGFP) embryos [15] and observed fli + cells at 16 som (Fig. 3A), and found significantly fewer EGFP + cells compared to uninjected embryos (Fig. 3B). Time-lapse photography of FEVPs migration revealed fewer EGFP cells in the grl MO-injected embryo (Movie S2) comparison to uninjected embryos (Movie S1). We also performed p-H3 IF in 10 som control MO (Fig. 3C9), grl MO ( Fig. 3D), and uninjected embryos (Fig. 3C). We observed no difference in p-H3 + cell numbers at the midline, but a decrease in p-H3 + cell number is noted at LPM suggesting that grl loss of function selectively affects proliferation at the LPM (Fig. 3E).
We next interfered with grl expression using a dominant negative Suppressor of Hairless DNA binding mutant [Su(H) DBM ] construct, which interferes with endogenous Su(H) activity by sequestering Notch intracellular domain (notchICD) without DNA binding. [16]. Su(H) DBM -injected embryos (Fig. 3G) showed fewer etsrp + cells compared to control MO-injected embryos (Fig. 3F). Taken together, these experiments argue that grl function affects proliferation and numbers of FEVPs cells at the LPM.
Notch signaling pathway is critical for FEVPs cell development at the LPM We next investigated Notch, an upstream positive regulator of Grl. Since loss of grl function resulted in fewer etsrp + cells, we hypothesized that constitutive activation of Notch dependent transcription by injecting notch-intracellular domain (notchICD) mRNA should increase numbers of these cells. Indeed, notchICD mRNAinjected ( Fig. 3H) embryos showed more etsrp + cells compared to control MO-injected embryos (Fig. 3F). Quantitative determination of etsrp transcript levels in notchICD mRNA-injected embryo showed a 1.75 fold increase in etsrp transcript levels (Fig. 3I, NICD panel). As expected, grl-MO and Su(H) DBM -injected embryos showed a 0.3 to 0.6 fold reduction compared to WT embryos (Fig. 3I, blue bars). We also checked for changes in vascular endothelial growth factor (vegf) (Fig. 3I, red bars), and a reference transcript actin (Fig. 3I, green bars), and observed no changes in the levels of these transcripts in treated embryos compared to controls. Finally, we inhibited notch ICD generation by chemical inhibition of c-secretases using DAPT. Compared to DMSO treated embryos (Figs. 3J & 3M), DAPT-treated embryos from 10-14 hpf (Figs. 3K & 3N) showed an overall expansion of scl + cells, and decrease of etsrp + cells most prominently in the posterior LPM resulting in a decreased gap between anterior and posterior LPM (bracket region). Similar results were observed for scl + cells in embryos treated with a second c-secretase inhibitor, WPE-III-31C- [17] (Fig. 3L) suggesting that etsrp + scl + double positive cells in this region are sensitive to notch inhibition, which is consistent with a recent study that shows co-expression of both markers in this cell population [18]. The whole embryo images of c-secretase inhibitor-treated embryos are provided in Figure S4A To dissect the temporal requirement of Notch signaling pathway in controlling etsrp cell numbers, we utilized the Notch gain-of-function (GOF) transgenic embryo system in which a heat shock-inducible hsp70:Gal4 transgene drives a Gal4-responsive upstream activating sequence (UAS) UAS:notch1aICD allele [19]. Embryos were transferred to 40 uC for 30 min at 1 som, and analyzed using etsrp ISH at 10 som. Heat-shock induced embryos showed ectopic induction of etsrp + cells in head (

Shh and VEGF are both involved in increasing etsrp cell numbers at LPM
We next investigated the final two upstream elements of the signaling axis, shh and VEGF. We performed shh and VEGF GOF experiments followed by etsrp ISH at 10 som. Shh overexpressing embryos (Fig. 4J) showed an increase in etsrp + cells at LPM compared to age-matched controls (Fig. 4I). We individually injected VEGF 165 , capable of heparin binding [20,21], and VEGF 121 mRNA, incapable of heparin binding [22] into 1-cell embryos. Both VEGF 165 (Fig. 4K) and VEGF 121 (Fig. 4L) overexpressing embryos showed increased etsrp + cells relative to controls, but distinct cell populations are affected by the different isoforms. To confirm the embryos stages of the injected embryos, an ISH for somite marker myod was performed, and is shown in Figure S3A To explore if temporal and spatial availability of VEGF influenced angioblast numbers at the midline, we injected deltaA (dlA) (a zebrafish delta homologue/notch ligand) mRNA into 1-cell embryo. DlA mRNA has been shown previously to increase floorplate and hypochord cells, sources of VEGF, at the expense of notochord [23]. DlA mRNA-injected embryos showed expected increase in col2a1 expression (data not shown), a marker of floorplate and hypochord cells. In dlA mRNA-injected embryos, we noticed a distinct increase ( Figure S4G-I) of etsrp + cells compared to control uninjected embryos ( Figure S3D-F). We also performed western blots for VEGF protein levels in dlA mRNA, grl MO and uninjected embryos from 10 som lysates (Fig. 5A). We found that dlA mRNA injected embryos showed a slight increase (Fig. 5A, Table) in the faster migrating VEGF 121 isoform, while grl MO showed a decrease in VEGF 121 isoform (Fig. 5A, lower band missing, asterisk) compared to uninjected embryos. Interestingly, the QPCR data in Fig. 3I showed no change in total vegf mRNA (both isoforms) in embryos injected with grl MO, while the western blot in Fig. 5A shows changes in VEGF 121 protein isoform. This suggests that increased stability or presence of vegf 165 mRNA isoform compensates for loss of vegf 121 mRNA in those embryos. Taking the grl/hey2, Notch, VEGF and Shh data together, it implies that the Shh-VEGF-Notch-Hey2 signaling pathway temporally and spatially regulates etsrp + cell proliferation during embryonic vascular development.

Grl/hey2 participation in proliferation process in differentiated endothelial cells in vitro is minimal
To investigate loss-of-grl/hey2 function in vitro, we designed small interfering (siRNA) using the dicer method [24]. Efficacy of hey2 siRNA was demonstrated by RT-PCR for hey2 gene, and showed targeting when compared to control (lacZ) siRNA-transfected cells ( Figure S5, hey2 gel). We investigated proliferation of hey2 silenced ECs initially using human pulmonary artery ECs (HPAECs) because Hey2 functions in artery development [25]. Because HPAECs were difficult to contact inhibit, subsequent cell cycle analysis using FACS was performed on a synchronous population of human umbilical vein endothelial cells (HUVECs). When HUVECs were released from contact inhibition (CI), a robust increase from 4.3 to 21.7% was noticed in S-phase ( Figure S5, red bar, CI to CR) along with a concomitant reduction in G0/G1 phase (82.1 to 62.8%, blue bar, CI to CR). In hey2 siRNA transfected HUVECs ( Figure S5, hey2), we noticed an increase in G0/G1 phase (73.3%) compared to lacZ siRNA transfected HUVECs (68.6%) and a concomitant decrease in S-phase cells (16.2%-hey2 siRNA, 17.1%-lacZ siRNA). However, statistical analysis showed no significance of this difference across samples. These results argue that grl functions in cell cycle events are distinct in endothelial precursor and venous differentiated endothelial cells.

Grl/Hey2 affects cell cyclin protein levels in vivo, which is reflected in cell phase specific inhibition by chemical inhibitors
To investigate how Grl influences proliferation, we analyzed cell cycle protein levels in zebrafish embryos. We investigated cyclin B1, E and D1 protein levels in grl LOF and dlA GOF embryos at 10 som (Fig. 5B). Grl LOF represents loss of VEGF 121 and dlA mRNA injected embryos represent indirect gain of VEGF function because it is known to increase floorplate and hypochord cells, sources of VEGF. Compared to uninjected WT embryos, grl MO embryos showed a decrease in cyclin B1 levels (Fig. 5B, red bar, cyclin B1). In contrast dlA mRNA injected embryos showed an increase in cyclin B1 levels (Fig. 5B, green bar, cyclin B1). Little change in cyclin E or cyclin D1 is observed among sample groups.
We next investigated whether chemical inhibition of proliferation would reduce FEVP numbers at stages in which we observed increased etsrp expression. Based on published literature for cell cycle inhibitors in zebrafish [26], we treated zebrafish embryos   Figure S4 and Figure S6). The numbers on the top right panel of all ISH embryos in this figure indicate number of embryos out of total number displaying that particular phenotype. doi:10.1371/journal.pone.0014732.g005 in dramatic changes in etsrp + cell populations in embryos. Taken together, these results indicate that global inhibition of proliferation interferes with the increase in etsrp + cells observed in control staged embryos, suggesting a requirement for proliferation in generation of the appropriately sized FEVP pool.

Discussion
A recent flurry of publication emphasizes the anatomical and developmental inter-relationship between hematopoietic and vascular cells in vertebrates [28,29]. Recent studies have shown that that endothelial cells can give rise to hematopoietic precursors [28,29], and conversely, hematopoietic precursors (HSC) have vascular potential [30]. The shared origin within the LPM for blood and vascular precursors perhaps contributes to this phenomenon [31,32]. Depending on context, precursor cell populations with vascular and hematopoietic potential can be referred to as hemangioblast or hemogenic endothelium [33]. Because fli1 is the most upstream transcription factor in the hierarchy of factors that drive specification of blood and vascular precursor pools [3], and Scl [34,35], and more recently etsrp [4,5] transcription factors (downstream of fli1) operate in both vasculogenic and hematopoietic lineages, we refer in this study to fli + etsrp + hemato-vascular progenitor cells (abbreviated as FEVPs) as vascular precursor populations, retaining some hematopoietic potential. The identification of etsrp as a marker required for vasculogenesis in zebrafish [4] and Xenopus [36], has allowed the investigation of the early events in vasculogenesis in a developing embryo. Here, we have taken advantage of this marker analysis in developing embryonic vasculature in zebrafish.
In this study, we have demonstrated that FEVPs proliferate at the LPM in vivo, and that cellular proliferation is critical for generation of correct numbers of vascular precursors. Several salient features can be surmised from this study. First, etsrp + populations in the head, trunk and tail follow distinct developmental and morphogenetic programs in response to local cues. Second, the Notch signaling pathway regulates FEVPs proliferation at the LPM both temporally and spatially revealing a novel contribution for this pathway in addition to its previously understood role in A/V specification. Third, hypochord/endoderm may be the critical tissue source for VEGF driving FEVPs proliferation at the LPM. Finally, cell cycle protein levels are altered in embryos where Notch signaling is attenuated, suggesting a mechanistic basis for FEVPs proliferation at the LPM.
Etsrp expression provides an unprecedented window into the early events of vasculogenesis, as etsrp + populations in distinct regions of the embryo appear to respond to local cues in different manners and generate a pattern that is specific for that region. First, injection of different vegf mRNA isoforms vegf 165 vs. vegf 121 results in differential effects on etsrp + cells in the trunk and head. Importantly, other groups [37] have reported similar observations at a later time point (30 hpf). Second, HU-treated embryos showed different outcomes in the head and trunk etsrp + cells. Thus the major populations of etsrp + cells along the anterior-posterior axis of the embryo show distinct developmental effects and differential sensitivity to perturbations.
The Shh-VEGF-Notch-Hey2 signaling pathway has previously been identified as a central player in embryonic vascular development in zebrafish [14,38,39]. The expression patterns of Shh-VEGF-Notch-Hey2 are consistent with a role in FEVPs proliferation at the LPM. Shh expression is detected as early as bud stage, and continues until 10 som in midline notochord structure as shown by others [13] and us. Vegf expression in somite is not seen until 10 som [14], and is observed along the intermediate mesoderm between the LPM and midline at 18-20 som [40]. At 6 som, the vegf expression is located in regions adjacent to yolk, and is likely emerging from the transient hypochord structure, which is in close association with the notochord [41]. At 9-10 som, we hypothesize that VEGF from the hypochord (VEGF 121 ) diffuses laterally and comes in contact with the LPM cells, which are continuous with the endoderm at this time point. Consistent with this interpretation is our vegf ISH and IF data at 10 som. In addition, data in Xenopus where diffusible form of VEGF expressed by the hypochord is responsible for angioblast migration [42], and the dlA mRNA injected embryo showing more etsrp + cells at the LPM support this hypothesis. Further, notch activation increases hypochord cells at expense of floor plate cells in zebrafish [43], thereby causing an increase in VEGF source at hypochord similar to dlA, and thus explains the putative increase in FEVPs in notchICD-injected and heat-shocked Tg(hsp70:Gal4;uas:nICD) embryos. The flk + cells are also found at 10 som adjacent to vegf + cells at the LPM [14] suggesting that VEGF-Flk interaction may induce Notch-hey2 signaling cascade resulting in FEVPs proliferation at the LPM. Indeed, of the duplicated Notch1 in zebrafish, notch1b expression is intriguingly similar to etsrp expression at 10 som in the zebrafish LPM [12] suggesting that Notch1b could transactivate hey2 promoter [44], and induce a proliferative signal in FEVPs at the LPM. Importantly, our heat shock experiments indicate that precocious activation of Notch at 1 som lead to enhanced etsrp + cell population at 10 som suggesting that pathway components are already in place at the LPM as early as 5 som, assuming that it takes 1-2 h for the production of NotchICD. Finally, Notch target: hey2/ grl is expressed also at the LPM at 5 som [10] and 10 som indicating that temporally and spatially the Shh-VEGF-Notch-Hey2 signaling axis is likely involved in FEVPs proliferation at the LPM from 3-10 som. A model is presented (Fig. 6) to explain this pathway in the context of FEVPs development from 3 som to 10 som.
Chemical genetic studies with two known c-secretase notch pathway inhibitors show that scl + etsrp + cell populations are sensitive to notch inhibition thus expanding the junction gap between the anterior and posterior LPMs. These data contradict the recent report from Lee et al, which concluded that Notch signaling functions as a cell-fate switch between endothelial and hematopoietic lineages. This group [45] treated embryos with DAPT at 10-14 hpf, and showed an increase in endothelial cells and decrease in hematopoietic cells at 18-20 hpf. However, this work has several limitations. The interval spanning the drug treatment and final analysis encompass several discrete developmental events at the LPM and midline (Fig. 1). Secondly, the majority of the analysis in this study relies on absolute cell numbers taken from confocal sections. It is clearly not possible to match the exact same location of the section each time across different embryos. Thirdly, taking the first two limitations together, observations comparing absolute numbers from section to section for a process that is undergoing dynamic changes in a 10 h window are necessarily correlative. Our study used two distinct c-secretase inhibitor treatments, and a host of distinct complementary manipulations to conclude that Notch signaling is involved in proliferation of vascular precursor cells. Combining present work with those presented in Lee et al's, we hypothesize that modulating Notch by DAPT alters the absolute number of precursors at 12 som, thereby changing the numbers of differentiated cells of the blood and vascular lineage observed at 18 hpf. However, this scenario does not account for the selective up regulation of one lineage over another, which is unclear for now.
Although cell proliferation is one mechanism of increasing cell numbers, a second mechanism of cell specification can also achieve the same effect. In this alternate hypothesis, ''newly speci-fied endothelial precursor cell in the LPM is the cause for increased cell numbers of the vascular lineage.'' At present, we have no evidence to refute this hypothesis. However, the hypothesis in itself does not refute the evidence that is provided here, which shows that endothelial precursor (fli + , etsrp + ) cells are proliferating at the LPM. Therefore, although increased specification vs. proliferation could be reasoned for increasing cell numbers, there is clear evidence here that suggests that once a cell is specified, i.e., fli + etsrp + (FEVP), this cell undergoes proliferation. Importantly, the Notch signaling pathway, which we imply here in proliferation shows precedence for playing a functional role in proliferation of midline cells in zebrafish [43], and in zebrafish hematopoiesis [46]. Moreover, Notch also participates in cell proliferation events in Drosophila eye [47][48] and wing development [49]. The zebrafish notch study [43] in particular is highly relevant because the authors convincingly show that elevation of notch signaling caused expansion of hypochord at expense of notochord without affecting floor plate cells, and notch inhibition via DAPT caused loss of hypochord cells. They also demonstrated that notch signaling is involved in the proliferation of midline floor plate progenitors similar to FEVPs reported here.
Prior to the mid-blastula transition (MBT), cells of zebrafish embryos cycle without G1 and G2 phases [50,51]. Duffy and colleagues have reported that key cell cycle regulators involved in S to M phase transition include ccnb1 (cyclin B1), ccnb2 (cyclin B2), and ccne (cyclin E). Here, we identify that cyclin B1 levels are lower in grl LOF 10 som embryos. Enhancement of VEGF production (indirectly by DeltaA) gave the opposite result, which suggests that VEGF-Hey2 signaling axis at 10 som is capable of controlling FEVPs proliferation via modulating cyclin proteins.
This study demonstrates FEVP cell proliferation at the LPM in a developing vertebrate embryo, and implicates the Shh-VEGF-Notch-Hey2 signaling pathway in this process. Whether this process, and the signaling pathway utilized to accomplish precursor cell proliferation are conserved in mammalian development is not known. The data presented here continues the theme of signaling pathways redundantly utilized at multiple stages in a developing embryo, as is often the case in embryonic development [52].

Zebrafish husbandry and Microinjection
All zebrafish studies were performed according to MCW animal protocol guidelines under protocol # 312-06-2, which were approved by the MCW Institutional Animal Care and Use Committee (IACUC). WT TuAB, Tg(fli1a:EGFP) and Tg(fli1a:-nEGFP) were obtained from ZFIN. All DNA templates were linearized with appropriate enzymes and transcribed by mMessenger mMachine kit (Ambion). Microinjection was performed according to previously established protocols [53]. Seventy-five picograms of each capped mRNA were injected into 1-cell stage embryos. Heat-shock experiment was performed on a clutch of embryos from mating of adult Tg(uas:notch1a-intra) and Tg(hsp70:gal4) by transferring embryos at 1 and 5 som to 37 uC for 30 min in a water bath [46]. Subsequently, embryos were cultured at 28.5 uC until 10 som, fixed with 4% PFA, followed by etsrp ISH.
Whole mount in situ hybridization, immunohistochemistry, and section Whole mount in situ hybridization (WISH) was performed as described previously [53]. Hybridized embryos were photographed using a Leica Stereomicroscope equipped with a Qimaging camera and images assembled in Adobe Photoshop CS3 Extended. For immunohistochemistry, Tg(fli1a:nEGFP) embryo was fixed at 10 som and frozen in TFM. Microm HM 550 (Thermo Scientific, Inc.) was used to generate serial 10 mm cryosections embedded in tissue freezing medium (TFM) (Triangle Biomedical Science, Inc.), which was mounted onto glass slide. Immunohistochemistry was performed with rabbit anti-GFP antibody (Cell Signaling, #2555) as primary and goat anti-rabbit IgG-HRP-conjugated as secondary (Cell Signaling, #7074). The signal was detected using tyramide labeled fluorescein (NEN Life Science Products, Inc.) followed by company's protocol. For phospho-histone3 immunostaining, Tg(fli1a:EGFP) embryos was fixed at 10 som and processed as described above followed by immunohistochemistry with rabbit anti-GFP antibody (Cell Figure 6. A model depicting the Shh-VEGF-Notch-Hey2 pathway in LPM angioblast proliferation. This model is based on data from here and from other groups [12,14,37,38,40,41,55]. Nomenclature for the various structures, molecules and cell types are indicated. Starting at 3 som, sonic hedgehog (shh) or other signal from the notochord stimulate VEGF 121 expression from adjoining hypochord, which is associated with endoderm cells lining the yolk boundary. At 4-8 som, VEGF 121 diffuses via intermediate mesoderm and contacts angioblast cells that are positive for etsrp and flk marker. At 9 som, a second source of VEGF from somite is detected, which is likely induced by shh from Notochord. Whether the two sources of VEGF have similar or different function is unknown. At approximately 10 som, the VEGF binding to its receptor (Flk?) triggers Notch activation at the LPM [12], which results in activation of downstream target genes such as hey2/gridlock. We hypothesize that this signal activation triggers angioblast proliferation at the LPM. A second feedback loop of hey2-VEGF between hey2 angioblast and VEGF in somite or in hypochord exists, but the temporal and spatial function of this loop is yet to be determined. NO: notochord, NT: neural tube, S: somite. doi:10.1371/journal.pone.0014732.g006 Signaling, #2555) and mouse monoclonal anti-phospho-histone3 (Sigma, H6409) as primary and anti-rabbit IgG-Alexa 488 (Invitrogen, A-21204) and anti-mouse IgG-cy3 (Jackson Immu-noResearch, 715-165-151) conjugated as secondary. The Signals were imaged by Observer Z1 inverted microscope (Carl Zeiss). The details of additional methods used in this study are provided in the supplemental text file (Text S1).

Statistical analysis
Statistical analysis was performed using the Student's t-test with Graph Pad Prism (GraphPad Software, La Jolla, CA) and Microsoft Office Excel 2010 software package. All data are presented as mean 6 SEM (n and p-value are provided in each figure).

Supporting Information
Text S1. This file contains methods used in the study but not described in the main text file.