Widespread lipoplex-mediated gene transfer to vascular endothelial cells and hemangioblasts in the vertebrate embryo

Authors

  • Karine Bollérot,

    1. UPMC, CNRS UMR7622, Laboratoire de Biologie du Développement; Bat C, 6ème étage, Case 24; 75252 Paris, Cedex 05, France
    Search for more papers by this author
  • Daisuke Sugiyama,

    1. UPMC, CNRS UMR7622, Laboratoire de Biologie du Développement; Bat C, 6ème étage, Case 24; 75252 Paris, Cedex 05, France
    Current affiliation:
    1. Department of Biochemistry, Dartmouth Medical School, Hanover, NH 03755
    Search for more papers by this author
  • Virginie Escriou,

    1. Unité de Pharmacologie Chimique et Génétique, INSERM U640/CNRS UMR 8151/Université René Descartes, Paris V/Ecole Nationale Supérieure de Chimie de Paris; Faculté de Pharmacie, 4, Paris, France
    Search for more papers by this author
  • Rodolphe Gautier,

    1. UPMC, CNRS UMR7622, Laboratoire de Biologie du Développement; Bat C, 6ème étage, Case 24; 75252 Paris, Cedex 05, France
    Search for more papers by this author
  • Samuel Tozer,

    1. UPMC, CNRS UMR7622, Laboratoire de Biologie du Développement; Bat C, 6ème étage, Case 24; 75252 Paris, Cedex 05, France
    Search for more papers by this author
  • Daniel Scherman,

    1. Unité de Pharmacologie Chimique et Génétique, INSERM U640/CNRS UMR 8151/Université René Descartes, Paris V/Ecole Nationale Supérieure de Chimie de Paris; Faculté de Pharmacie, 4, Paris, France
    Search for more papers by this author
  • Thierry Jaffredo

    Corresponding author
    1. UPMC, CNRS UMR7622, Laboratoire de Biologie du Développement; Bat C, 6ème étage, Case 24; 75252 Paris, Cedex 05, France
    • UPMC, CNRS UMR7622, Laboratoire de Biologie du Développement; Bat C, 6ème étage, Case 24; 75252 Paris, Cedex 05, France
    Search for more papers by this author

Abstract

We report here a method that allows fast, efficient, and low-cost screening for gene function in the vascular system of the vertebrate embryo. Through intracardiac delivery of nucleic acids optimally compacted by a specific cationic lipid, we are able to induce in vivo endothelial cell-specific gain-of-function during development of the vascular network in the chick embryo. When the nucleic acids are delivered during the period of intraembryonic hematopoiesis, aortic hemangioblasts, the forerunners of the hematopoietic stem cells known to derive from the aortic endothelium, are also labeled. Similarly, we show that siRNA could be used to induce loss-of-function in vascular endothelial cells. This gene transfer technique was also applied to the mouse embryo with a high efficiency. The present method allows large-scale analysis and may represent a new and versatile tool for functional genomics. Developmental Dynamics 235:105–114, 2006. © 2005 Wiley-Liss, Inc.

INTRODUCTION

Vessel formation in vertebrates is controlled by a cascade of tightly regulated cellular and molecular events starting from the specification of angioblasts from the mesoderm to the formation of a fully mature vessel surrounded by a mural cell tunicae (Carmeliet, 2003; Ema and Rossant, 2003; Gerhardt and Betsholtz, 2003). Endothelial cells (EC) also share a common progenitor with hematopoietic cells (HC) termed the hemangioblast (Murray, 1932; Lacaud et al., 2001; Choi, 2002). In particular, adult-type HC were shown to originate from a specific population of EC located in the floor of the aorta, revealing the existence of an aortic hemangioblast (Jaffredo et al., 1998; de Bruijn et al., 2002; Oberlin et al., 2002). Taking advantage of its large accessibility and the availability of the quail/chick system (Le Douarin and Barq, 1969), many aspects in the development of the vascular and hematopoietic systems have been established first in the avian embryo before they were confirmed in mammals. The use of this model, however, has been hampered by the lack of genetic manipulation applicable to the study of blood and vascular cells. The role(s) of genes involved in the different steps of vascular formation, hemangioblast specification, and eventually blood cell production is usually analyzed through loss- or gain-of-function experiments in the zebrafish and mouse models (Nishikawa, 2001; Ling and Dzierzak, 2002; Trede et al., 2004). However, due to their critical role in development, inactivation of these genes frequently leads to premature death. To circumvent this problem, sophisticated and time-consuming transgenic systems with conditional promoters allowing locally and temporally defined modifications of gene expression, have been designed. On one hand, given the cost and amount of time needed, the number of candidate molecules that can be tested is limited. On the other hand, given the increased availability of expressed sequence tags (EST) and expression pattern and genome databanks, a rapid and convenient test for gene function applicable to the developing vascular system of avian and, to some extent, mammalian embryos is needed. Such a test should allow the study of gene function in a short time window and ideally should be applicable during the entire embryonic life.

We have designed a lipofection-based method for specific gain- and loss-of-function to vascular EC and hemangioblasts of the chick embryo. Nucleic acids are compacted with cationic liposomes specifically designed for gene therapy applications. Lipid-based DNA transfection techniques have aroused much interest over the past 10 years and have been applied to target genes in the avian or mammalian embryos. Noncationic or cationic lipid-mediated gene transfer was reported to allow gene delivery to various regions of the embryo (Demeneix et al., 1994; Muramatsu et al., 1997; Yamada et al., 1997; Toy et al., 2000; Longmuir et al., 2001; Dickinson et al., 2002; Yamamoto et al., 2004) as well as to the adult organism (Zhu et al., 1993; Thurston et al., 1998; Kircheis et al., 2001; Uyechi et al., 2001). However, efficient gene transfer into EC of the embryo has somehow resisted previous attempts. In our system, lipoplexes, i.e., the nucleic acid(s) associated to liposomes, are delivered to the embryonic circulation by intracardiac injection, yielding very efficient EC transfection. Indeed most EC, including those of the aorta, are labeled. This delivery system can be applied to avian and mouse embryos with high efficiencies. Gain-of-function using the angiogenic molecule vascular endothelial growth factor (VEGF) but also loss-of-function mediated by siRNAs can be achieved. This method offers new possibilities to rapidly analyze the function of any gene involved in the formation and maintenance of the vascular and hematopoietic systems. Such an approach should also help discover new signaling pathways involved in normal or pathological situations and at the same time should provide opportunities for the identification of new therapeutic molecules. The simple design of the experiment combined with its low cost makes it possible to screen for the function of a large number of genes and gene variants in a short period of time. Moreover, because injections can be performed at any time during the first 6 days of chick development and possibly later, the present technology offers tight temporal control on the process of gene transfer and gene function analysis.

RESULTS AND DISCUSSION

Physicochemical Characteristics of 209120-Containing Lipoplexes

The lipid 209120 (see Experimental Procedures section for details) has been used previously by some of us to complex plasmid DNA in targeted (Hofland et al., 2002) or stealth lipoplexes (Nicolazzi et al., 2003) injected in mice. Here, the lipid 209120 was formulated as a cationic liposomal preparation by mixing with the neutral co-lipid DOPE (Fig. 1A). Optimal in vitro transfection efficiency is generally observed provided lipoplexes exhibit two characteristics. First, they should carry a net positive charge, and, therefore, an excess of cationic lipids. Because of the net negative charge born by the plasma membrane, lipoplexes are assumed to bind to the cell through ionic interactions. Second, one of us reported that lipoplexes, prepared either in carbonate buffer or in serum-free medium, form large complexes (approximately 1 μm in size) that provide a much better transfection in the presence of serum than the standard protocols that usually yield small particles (Escriou et al., 1998).

Figure 1.

Chemical structure of liposomes and in vitro transfection of human umbilical vascular endothelial cells (HUVEC). A: Molecular structure of the cationic and neutral lipids used in this study. Cationic lipid 209120 contains a globular spermine as cationic head-group linked to a double C14:0 anchor by means of a short spacer. DOPE is a widely used neutral lipid that is added to cationic lipids to form cationic liposomes. B: In vitro transfection of HUVEC. HUVEC were transfected for 24 hr with fluorescent (red) lipoplexes prepared at 6 nmol of cationic lipid/μg pCMS-EGFP. The presence of cationic lipid 209120 in the cells is evidenced by the red color, whereas the green fluorescent protein (GFP) expression gives a green color. The yellow color indicates that both products are present into the same cell. Scale bar = 50 μm in B.

Therefore, we have prepared large lipoplexes to transfer pGFP to human umbilical vascular endothelial cells (HUVEC), a cell type known to be refractory to transfection. Lipoplexes are formed in NaCl 150 mM at 6 nmol of cationic lipid per DNA μg and incubated with cells in the absence of serum. These conditions lead to large lipoplexes, as previously described (Escriou et al., 1998). A total of 21 ± 5% and 35 ± 7% of HUVEC cells are green fluorescent protein (GFP) labeled at, respectively, 24 and 48 hr after transfection (data not shown). Transfection is very efficient when compared with the data reported by others in HUVEC (Kaiser and Toborek, 2001; Zaric et al., 2004). Uptake by cells is monitored using rhodamine-conjugated lipoplexes and compared with vector expression. Of interest, most of the cells take up lipoplexes as testified by the red color, whereas 30% express GFP (Fig. 1B). Furthermore, lipoplexes prepared in carbonate buffer exhibit the same transfection efficiency even in the presence of serum (data not shown).

Taking advantage of this knowledge, we have selected two different physicochemical conditions for the formation of lipoplexes according to the stage of the embryo to be inoculated. The carbonate buffer, that allows large lipoplexes to be formed, is used when embryos earlier than Hamburger and Hamilton (HH; Hamburger and Hamilton, 1951) stages 16–17 (51 to 64 hr of development) have to be injected. NaCl buffer that forms smaller lipoplexes is retained when embryos from HH17 up to HH21 (52 to 84 hr) are injected (see next section for details).

Endothelium- and Hemangioblast-Specific Gene Transfer in the Avian Embryo

During the course of experiments, we have compared the viability of HH17 to HH21 embryos injected with the lipoplexes to the dilution reagent alone, i.e., NaCl. On the whole, a total of 320 embryos were injected with lipoplexes and 51 with NaCl. Respectively, 268 (84%) and 45 (88%) embryos survived to injection and developed normally. Only approximately 15% of the embryos die immediately after inoculation or display malformations that preclude their further development. This percentage is independent of the presence of lipoplexes, thus, is not mediated by a specific toxicity associated to the lipoplexes but rather is associated to the preparation of the embryos and the injection procedure.

In a first set of experiments, the distribution of fluorescent lipoplexes in the vascular tree was examined (n = 48). Rhodamine-conjugated lipoplexes are prepared using NaCl conditions (liposomes are available upon request, see also Experimental Procedures section for details) and injected as outlined in Figure 2A. This route of injection places reagents in close contact with EC lining the vessels and has been used previously to label EC with either human acetylated low density lipoproteins (Jaffredo et al., 1998; Sugiyama et al., 2003) or virus-borne reporter genes (Jaffredo et al., 2000; Hautmann et al., 2001). One hour after injection, rhodamine labeling is detected throughout the entire vascular tree both in embryonic and extraembryonic vessels (Fig. 2B,C).

Figure 2.

Distribution of the lipoplexes into the vascular system and green fluorescent protein (GFP) expression after transfection of chick and mouse embryos. A–J: chick. K–N: mouse. A: Route of injection. Lipoplexes are in green. The needle is introduced into the heart. The solution is flushed and distributes throughout the vascular system. B,C: Distribution of rhodamine-flagged lipoplexes 1 hr after injection. D–H: GFP expression 24 hr after lipoplexes delivery. D: GFP expression 24 hr after in ovo delivery of the lipoplexes. E: Twenty-four hours after delivery at embryonic day (E) 2 to an early chick embryo (Hamburger and Hamilton stage 11) cultured ex ovo. All the vessels express the transgene. F: GFP expression into the segmental arteries. Enlargement of the frame in D. G: YS vessel. H: GFP expression in a vessel. I,J: Double staining at the level of the aorta. GFP expression (arrowheads in I) does not impair the normal expression of c-myb (J). In situ hybridization. K: E10.5 mouse embryo. Whole-mount, rhodamine staining. The tagged lipoplexes distributed throughout the entire embryo. Cephalic vessels and segmental arteries are labeled (arrowheads). L: YS vascular tree filled with liposomes. M: GFP expression 12 hr after lipoplexes delivery. The capillary network of the trunk and the limb bud are labeled. N: Phase contrast of M. Al, allantois; Ao, aorta; E, encephalon; Ey, eye; H, heart; LB, leg bud; N, notochord; Pl, placenta; WB, wing bud; YS, yolk sac. Scale bars = 1 mm in B (applies to B,C), 500 μm in E, 100 μm in F, 50 μm in G, 70 μm in H, 70 μm in I (applies to I,J), 1 mm in K, 800 μm in L, 1 mm in M (applies to M,N).

Gene transfer to vascular EC has been monitored using pCMS-EGFP (n = 225), a cytomegalovirus (CMV) -driven expression vector encoding an enhanced GFP, in lipoplexes obtained in NaCl conditions (see Experimental Procedures section). As a control, delivery of naked DNA (n = 25) does not produce any detectable signal. All surviving embryos display GFP expression in their vascular system. GFP is first detected as early as 3 hr after injection, i.e., a time course similar to that reported after lipofection or plasmid electroporation (Muramatsu et al., 1997; Itasaki et al., 1999; Dickinson et al., 2002; Scaal et al., 2004). The signal peaks between 24 and 36 hr and disappears at approximately 72 hr after injection. However, when using a plasmid driven by FerL and FerH human ferritin promoters (pVIVO2), it is possible to detect a continuous GFP expression for up to 5 days after transfection (n = 47, data not shown). One day after transfection of pCMS-EGFP, GFP expression is detected in the yolk sac (YS) and the entire vascular tree of the embryo (Fig. 2D–H). YS arteries are usually much brighter than veins. This difference is likely to be because of the physicochemical properties of cationic lipoplexes prepared in NaCl conditions that readily adhere to the vessels. Because intracardiac injection distributes the solution into the arterial system, most of the lipoplexes adhere to arteries. When the solution reaches the venous system, lipoplex concentration is decreased and so is gene transfer. In the embryo proper, virtually all the vessels are labeled (Fig. 2D–F). On tissue sections, GFP expression is restricted to vascular EC both into the embryo proper and in the YS (Fig. 2H). In the most dramatic cases, most of the EC in a single vessel display a GFP signal.

When injection in somite-stage embryos is needed, however, intracardiac injection should be performed in culture according to Chapman et al. (2001). This approach is necessary because the heart is still positioned on the ventral midline and, thus, is not readily accessible. Because NaCl-diluted lipoplexes proved to be dramatically efficient on E3 embryos, they were applied to target vessels in E2 embryos. An intense labeling of the entire vascular system was observed at 24 hr after injection. However, 90% of the embryos died on the day after injection. NaCl-buffered lipoplexes readily adhered to the inner side of the vessels and formed obstacles that obliterate the vessels (data not shown). To avoid plug formation, the physicochemical properties of the lipoplexes were modified. Carbonate buffer was used instead of NaCl to form lipoplexes. Carbonate-buffered lipoplexes did not readily adhere to the vessels; instead, they circulated freely for a few hours before adhering to the endothelium. Embryos between HH11 to HH14 (40 to 53 hr) were injected (n = 150) and develop normally up to E3.5. The viability of embryos inoculated with carbonate-buffered lipoplexes reached 90% 1 day after inoculation. When injected with a GFP-bearing construct, embryos display a strong GFP expression in all their vascular structures similar to that seen in ovo with NaCl conditions (Fig. 2E). Embryos also displayed a strong GFP signal in circulating blood cells. This staining, rarely observed when embryos are injected at later stages, may result either from the transfection of hematopoietic precursors at the level of the developing YS blood islands when these structures become overtly opened to the blood flow or, alternatively, from the transfection of hematopoietic precursors after they were released into the circulation (see additional movie showing the intracardiac injection and the GFP expression in a living embryo 24 hr after injection; Supplementary Movie can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat).

Transfection of a GFP construct has no effect on the formation of the vascular system nor does it impair the capacity of aortic hemangioblasts to further differentiate into hematopoietic cells. In this respect, normal expression of molecular markers is not modified as testified by the presence of the hematopoiesis-specific genes c-myb (Fig. 2I,J), Pu-1, and CD45 (not shown). Because several forms of leukemias are now thought to be generated at an early HSC stage or even at the hemangioblast stage, gene transfer to hemangioblasts provides a unique opportunity to study the early molecular events involved in the initiation of these inherited blood disease (Tsuzuki et al., 2004).

Endothelium-Specific Gene Transfer in the Mouse Embryo

The efficiency of gene transfer into avian embryos has prompted us to perform similar experiments in cultured mouse embryos. Mouse embryos were isolated and inoculated according to Sugiyama et al. (2003). After injection, mouse embryos were maintained up to 12 hr in a whole embryo culture system according to Osumi and Inoue (2001) and Sugiyama et al. (2003). Embryos have been examined for the distribution of rhodamine-labeled lipoplexes and for GFP expression at, respectively, 2 and 12 hr. A total of 44 embryos were injected into the heart, either in carbonate or NaCl conditions and cultured for an additional period of 12 hr. The conclusions drawn for birds also apply to mouse embryos both for the distribution of rhodamine-labeled lipoplexes (Fig. 2K,L) and GFP expression (Fig. 2M,N). Twelve hours after injection, the distribution and intensity of the signal were similar to that found in the avian embryo after the same short period of time (data not shown for the avian embryo) but was slightly reduced when compared with the avian embryo 24 hr after plasmid delivery. This finding is likely due to the short culture time (12 hr) insufficient for complete vector expression. However, the recent improvements made in whole mammalian embryo culture allow sustained mouse embryo development for longer periods of time, up to 3 days. Thus, levels of expression similar to those obtained in the chick model should be expected. These improvements rely on the tight control of oxygen conditions, the size of the bottle wherein embryos are cultured, and the rotation rate of the bottle (Osumi and Inoue, 2001; Sugiyama et al., 2003). Delivery of lipoplexes could also benefit from the availability of ultrasound image-guided in utero injection (Gaiano et al., 1999).

Overexpression of VEGF122 in the Chicken Embryo

Overexpression of molecules to the vascular system was probed using the angiogenic factor VEGF122 (Sugishita et al., 2000). A plasmid DNA containing the backbone of the retrovirus (replication-competent retroviral vector [RCAS], Hughes et al., 1987) has been used, ensuring the long-term production of the VEGF122 protein. The plasmid DNA has been injected in ovo at E3. No anomaly has been detected with the control RCAS (n = 18). When RCAS-VEGF–containing lipoplexes are injected (n = 20), 100% of the embryos die at E6, with numerous hemorrhages and anomalies of the vascular system. In the YS, frequent vessel fusions were observed with the formation of vascular lacunae that obliterate circulation and cause death (Fig. 3A,B). In the embryo, vascular anomalies were also detected, tightly associated with a local increase of VEGF expression. In particular, when HIF2α (an EC-specific probe) was used, dorsal muscle groups displayed a dense vascular network, never observed in control embryos (Fig. 3C–F). This finding is in agreement with phenotypes reported after application of exogenous VEGF, i.e., aberrant patterning of segmental arteries or formation of vascular plexuses in areas not normally vascularized (Drake and Little, 1995; Flamme et al., 1995). However, with time, viral infection extends to additional cell types adjacent to EC (i.e., muscle cells or mesenchymal cells), a situation to avoid when EC-specific expression is needed. Using retroviral vectors within which the gene of interest is under the control of a tissue- or cell-specific promoter could circumvent this problem.

Figure 3.

Biological effects mediated by vascular endothelial growth factor (VEGF) after intravascular delivery of RCAS-VEGF–containing lipoplexes. A: Replication-competent retroviral vector (RCAS) control. Whole-mount of normal yolk sac (YS) vascular pattern. B: RCAS-VEGF. Note the enlarged, irregular vessels. Fused capillaries form vascular lacunae obliterating the circulation. C–F: Sections through the upper wing of embryonic day (E) 6 embryos injected with either with RCAS-VEGF (D,F) or control RCAS (C,E) at E3. CF: In situ hybridization with a chick VEGF probe (C,E) and with an HIF2α probe (D,F), a target of VEGF-signaling, specific for EC. Note the strong VEGF signal in the left side in RCAS-VEGF (D) compared with control RCAS (C). F: The VEGF-rich area (compared with D) is highly vascularized. The important vascular infiltration of the muscle mass in F is never observed in control embryos (E). Scale bar = 300 μm in B, 250 μm in F (applies to C–F).

Knockdown Effect Mediated by siRNA Delivery in the Chick Embryo

Loss-of-function using 21- to 23-bp double-strand RNA molecules (Hutvagner and Zamore, 2002; Lee and Roth, 2003) has already been implemented in the avian embryo to target genes to the neural tube using electroporation (Thurston et al., 1998; Dickinson et al., 2002; Pekarik et al., 2003). We show here, for the first time, that a similar knockdown approach could be applied in vivo to the vascular endothelium.

Distribution of lipoplex-associated siRNA has been monitored using a Cy3-labeled siRNA directed against the firefly luciferase LUC gene (siLUC-Cy3), an irrelevant target. From 24 to 48 hr after injection (n = 30), a strong Cy3 signal is visible in the entire vascular tree (Fig. 4A–C). On sections, Cy3 label is found in the cytoplasm of EC, showing that efficient cellular uptake had occurred. The majority of vascular EC display some labeling (Fig. 4D), and in the most dramatic cases, most of the cells do.

Figure 4.

Small inhibitory hairpin RNA (siRNA) in the avian embryo. A–C: Distribution of siRNA-Cy3 after injection. A,B: Vitelline arteries (arrowheads) in ultraviolet fluorescence (A) and phase contrast (B). C: siRNA-Cy3 distribution in the trunk. Aorta (AO, arrow) and segmental arteries (arrowheads) are strongly labeled. D: Distribution of fluorescent siRNA-Cy3 in the heart after intracardiac delivery. The fluorescent signal (yellow) is restricted to endothelial cells (arrowheads). EG: Representative heart samples 24 hr after injection. Cotransfection of green fluorescent protein (GFP, F) and red fluorescent protein (RFP, G) in the presence of irrelevant siRNA (luciferase). The two fluorescent patterns are similar. HJ: Cotransfection of GFP and RFP plasmids with GFP siRNA. RFP is strongly expressed, whereas GFP is almost absent. Scale bar = 500 μm in C, 100 μm in D, 500 μm in H–J.

Efficiency and selectivity of the knockdown have been tested by introducing pDSRed2 encoding the red fluorescent protein (RFP) protein (50 ng/μl) and pCMS-EGFP (75 ng/μl) with either siGFP (125 ng/μl, n = 6) or siLUC-Cy3 (125 ng/μl, n = 6). For convenience, fluorescent signals are analyzed in the heart, a site that consistently displays conspicuous expression. At 24 hr after injection, GFP and RFP signals are strong in control samples (Fig. 4E–G). In contrast, GFP is dramatically reduced when siGFP was co-injected, whereas RFP expression remains unchanged (Fig. 4H–J). These data indicate that a selective RNAi effect can be elicited in the vascular network by vessel delivery of lipoplexes without any toxicity for development. We might expect multiple genes (or a family of genes) to be knocked down at the same time by combining injection of multiple lipoplexes-associated siRNA. Indeed, the physicochemical principle of self-assembling cationic lipid/nucleic acid complex allows easy access to multiple siRNA-containing particles. Moreover, the RNAi silencing effect in the avian embryo does not necessarily require siRNA synthesis and may alternatively be elicited through either long double-strand RNA (Pekarik et al., 2003) or processed, short, interfering RNAs synthesized from partial cDNAs (Rao et al., 2004). By selecting the appropriate developmental stage, the timing of the knockdown could also be tightly controlled.

Comparing the Efficiency of Intravascular 209120 Lipoplex Delivery to Chicken Embryos to Other Techniques or Reagents

In the early aspect of this work, we have compared several techniques or reagents for their abilities to transfer genes into vascular EC. These attempts are summarized below.

Electroporation.

This technique has been reported as efficient at introducing genes into adult vascular EC (Nishi et al., 1996). Voltages of 35 V and plasmid concentration of 2.5 μg/μl in 0.1% phosphate buffered saline (PBS) /0.01% 2 M MgCl2 were used. A total of 114 embryos have been electroporated after naked GFP cDNA injection into the heart. More than 90% survive, and 25% exhibit staining at the level of the aorta. On section, GFP expression is restricted to one side of the aorta, as expected with the vectorial electric field used. Expression is not restricted to vascular EC but is also detected in tissues underlying the aorta. Decreasing the electric field results in a loss of gene transfer (data not shown). We thus show that electroporation-mediated gene transfer can be applied to the embryonic aorta but transgene expression is never restricted to EC, likely because plasmids moving into the electric field cross the thin EC barrier causing entry into non-desired abluminal cells.

Viral vectors.

Viral delivery has also been instrumental to tag the vascular tree of the embryo (Jaffredo et al., 2000; Hautmann et al., 2001). Viral approaches, however, have several limitations, among which, variations in tissue specificity according to the viral receptor, size constraints, and complex viral production (Miller, 1997). Virus-mediated gene expression takes between 12 to 15 hr to be expressed (Fields-Berry et al., 1992; Jaffredo et al., 2000), whereas plasmid expression occurs within 3 to 5 hr after liposome- or electroporation-mediated gene transfer, thus allowing early molecular events to be approached.

Commercial liposomes.

Three commercial reagents among seven have been selected according to their capacity to transfer genes into cultured avian cells in the presence of high serum concentration. Easy Fector (EquiBio), Superfect (Qiagen), and ExGen 500 (MBI Fermentas) have been selected. One of these lipids, selected in another study for its ability to mediate transfection into avian neural cells (Toy et al., 2000), is not efficient here; whereas another (DOTAP), retained for mediating germinal transgenesis and reported to give some EC and HC gene transfer, has not been retained for in vivo experiments (Watanabe et al., 1994).

Embryos were injected in ovo at E2.5. A total of 33 embryos were injected with Superfect, 28 with EasyFector, and 20 with ExGen 500 in three independent experimental series. A high mortality was observed with Superfect and ExGen 500 (60% of the embryos). None of the embryos displayed significant vessel labeling, although Superfect gave a conspicuous labeling restricted to the heart (not shown). Thus, none of the currently available commercial lipids appear efficient to transfer genes into embryonic EC.

Concluding Remarks

We present here a versatile and highly reproducible system that allows nucleic acids to be specifically delivered to vascular EC of the vertebrate embryo. The specific and novel formulation of the cationic lipid 209120 combined to intracardiac delivery results in an efficient and reproducible gene transfer into vascular EC and, in specific cases, into hemangioblasts. It constitutes a new tool for functional genomics in amniote embryos. Gain, but also loss-of-function, are obtained in a very simple manner. It should allow large-scale analysis, hence allowing specific molecules to be selected for further classic transgenesis approaches. Due to the accessibility of the chicken embryo for prolonged periods of development, lipoplexes could be delivered at different times of development, thus allowing several steps in the formation of the vascular system to be analyzed. As the vascular system is the target for many acquired and inherited diseases, such an approach should also provide opportunities for the discovery of new therapeutic molecules.

EXPERIMENTAL PROCEDURES

Liposomes

Easy Fector (EquiBio) associates a polycationic lipid; the neutral, nontransfecting lipid ExGen 500 (MBI Fermentas) is a linear polyethylenimine; and Superfect 5 (Qiagen) is an activated dendrimer. All have been used according to supplier's recommendations. Cationic lipid RPR209120 (Fig. 1A, available upon request) was synthesized as described (Byk et al., 2001; US Patent 6,171,612, January 9, 2001). RPR209120 2-{3-[Bis-(3-amino-propyl)-amino]-propylamino}-N-ditetradecylcarbamoylmethyl-acetamide and DOPE (dioleyl phosphatidyl ethanolamine; Avantis Polar Lipids) were dissolved in chloroform and mixed in equimolar amounts. The organic solvent was evaporated under vacuum at 20°C using a Heidolph rotating-dessicator to form a thin film. The dried film was then hydrated for 24 hr with sterile milliQ water at 20°C to produce large multilamellar vesicles. The suspension was finally sonicated at 4°C (115 V, 80 W, 50–60 Hz) with a G112SP1G model sonicator (Laboratory Supplies Co. Hicksville, NY) to obtain a homogeneous suspension of liposomes with a diameter of 80–100 nm as measured by light scattering (Zetasizer, Nanoseries, Malvern, UK) as described previously (Frisch et al., 2004). Fluorescent lipoplexes were obtained by incorporating rhodamine-labeled DOPE in liposomes (5 mol% of total lipid).

Preparation of Lipoplexes

An equal volume of cationic liposome suspension (at the appropriate concentration) was added to the DNA stock solution and rapidly mixed by a few aspiration/injection cycles with a pipette. Typically, 1 μg of DNA was complexed with 6 nmol of cationic lipid, in 150 mM NaCl (thereafter referred as NaCl conditions) or in 20 mM NaHCO3/150 mM NaCl (thereafter referred as carbonate conditions). Considering three positive charges per cationic lipid RPR209120 molecule at physiological pH, this method gives a positive amine over negative phosphate charge ratio of 6. Lipoplexes were allowed to form for 20 min at room temperature before use. Size particles and zeta potential, respectively, were estimated to be 169 ± 12 nm and 60 ± 3 mV with a final concentration of plasmid DNA of 8 μg/ml. For in vitro transfection and embryo injection, the final concentration of plasmid DNA was 5 μg/ml and 250 μg/ml, respectively.

In Vitro Transfection

HUVEC were obtained from Promocell (Heidelberg, Germany) and grown in endothelial cell growth medium (Promocell) in gelatin-coated culture flasks. The day before transfection, cells were seeded on 24-well culture plates, at 60,000 cells/well. Cells were rinsed once with MEM (Invitrogen). Lipoplexes prepared as described above were diluted to the tenth in MEM (1 ml/well, 0.5 μg DNA/well). This transfection medium was added to the cells, which were further incubated for 4 hr at 37°C in the presence of 5% CO2. Transfection medium was afterward replaced by fresh culture medium, and the cells were incubated for an additional period of 24 hr at 37°C. At the end of the incubation time, the cells were rinsed with PBS, fixed with 3% paraformaldehyde, stained with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI; 0.1 μg/μl in PBS), and mounted on slides in Mowiol. Examination was performed with a Zeiss Axiophot fluorescence microscope.

Embryos and Embryo Manipulations

Chicken (JA 57 or White Leghorn strain) eggs were incubated at 38 ± 1°C in a humid atmosphere until they reached the appropriate stages determined according to Hamburger and Hamilton, (1951). Intracardiac in ovo injections were performed according to previously described techniques (Jaffredo et al., 1998, 2000). Ex ovo cultures of chicken embryos were performed according to (Chapman et al., 2001). The OF1 pregnant mice were purchased from IFFA CREDO (France). Embryos at 10.5 days post coitum were isolated and cultured for 12 hr according to Sugiyama et al. (2003). Embryos were observed on an Olympus SZX12 binocular equipped with appropriate filters. Pictures were taken with a Nikon DMX1200 digital camera.

Chick and mouse embryos were injected into the heart with either plasmid- or siRNA-containing lipoplexes according to Jaffredo et al. (1998) and Sugiyama et al. (2003), respectively. Briefly, chick embryos were injected at E2 or E3 with 0.5 to 3 μl of lipoplexes, depending on the stage at injection. After lipoplexes delivery, embryos were incubated for an additional period of 1 hr to 3 days. Plasmids and siRNA were injected at the following concentration: 125 ng/μl of pCMS-EGFP and 125 ng/μl siGFP or siLUC-Cy3 as a negative control.

Mouse embryos were injected into the heart with 0.1–0.3 μl of either rhodamine-labeled lipoplexes or pCMS-GFP–containing lipoplexes. Embryo cultures were performed using the whole embryo culture system from BTC engineering (Cambridge, UK). Injected embryos were transferred immediately into culture bottles containing DMEM (Invitrogen, 6195-059) supplemented with 50% rat serum. The bottles were rotated at 37°C with a supply of gas mixture (60% O2 and 5% CO2 balanced with N2) for 12 hr. Pictures were taken with a Nikon stereomicroscope EM1200 equipped with appropriate filters.

Electroporation

Chicken embryos were injected into the heart with 0.1 to 0.5 μl of a 2.5 μg/μl of naked DNA plasmid solution diluted with 0.1% PBS (1 M) supplemented with 0.01% 2 M MgCl2. The solution was stained with 0.01% Fast Green fetal calf serum to visualize the inoculum. Electrodes were placed over the vitelline membrane on each side of the embryo at the mid-trunk level. Electrical stimulation with the following parameters: 35–40 V, 500 msec, 3 pulses 50 msec each, was applied when the green color reached the level of the electrodes.

Plasmid Preparation, Plasmids, and siRNAs

Plasmids were purified on Qiagen columns (Qiagen SA, Courtaboeuf, France) according to the manufacturer's recommendations. pDSRed2 and pCMS-EGFP were purchased from Clontech Laboratories (Palo Alto, CA), pVIVO2 containing the FerL and FerH human ferritin promoters was obtained from Invivogen, pCMV-Luc was described previously (Escriou et al., 2001), pCAβ-EGFP contains the EGFP under the control of the hybrid chicken β-actin promoter/CMV enhancer (Yaneza et al., 2002), and the replication-competent retrovirus RCAS-BP(A) was a kind gift from Dr. S. Hughes (Hughes et al., 1987). RCAS-BP(A)-VEGF122 was constructed by subcloning the chick VEGF122 cDNA (King gift from Dr. Takahashi, Tokyo; Sugishita et al., 2000) into the SLAX13 adaptor, using the NcoI and XbaI sites and inserting it in the sense orientation into the ClaI site of RCASBP(A) (Hughes et al., 1987; Logan and Tabin, 1998). Cy3 Luciferase GL2 duplex (138 ng/μl prepared in NaCl conditions) and GFP duplex siRNA were purchased from Dharmacon (Boulder, CO).

In Situ Hybridization and Immunohistochemistry

Embryos were isolated and fixed with 3.7% paraformaldehyde in PBS. After extensive washes in PBS, embryos were prepared for cryostat sectioning as previously described (Jaffredo et al., 1998). Sections (10 to 20μm thick) were deposited on Superfrost Plus slides (Menzel Gläser, Germany) and stored at −80°C until use. Gelatin was removed using a PBS bath at 37°C for 10 min and sections were mounted with Glycergel (DAKO Gmbh). Sections were photographed with a Nikon ECLIPSE E800 microscope equipped with Nomarski optics and appropriated filters for fluorescence. In situ hybridization was performed on frozen section according to (Ravassard et al., 1997). The VEGF122 probe was a kind gift from Dr. Takahashi (University of Tokyo). The HIF-2α probe was from J. Favier (Collège de France, Paris). The chick Myb probe was constructed by digesting the c-myb/CCC NEO (gift from Dr. S. Saule) with BspDI and subcloning the Myb fragment into pBSK-.

Acknowledgements

We thank F. Dieterlen, B. Péault, D. Duprez, C. Fournier-Thibault, and D. Dunon for helpful advices and critical reading of the manuscript. J. Collignon and A. Camus (Institut J. Monod, Paris) are acknowledged for sharing mouse embryo culture facilities and J. Nardelli for sharing experience with in situ hybridization on the mouse embryo. This work was supported by CNRS, INSERM, Universités Pierre et Marie Curie and René Descartes and ENSCP, by grants from the Association pour la Recherche Contre le Cancer 3312, from the Ligue Régionale contre le Cancer N° R04/75-158 and by an MERT ACI N° 22-2002-296. K.B. is a recipient fellowship from the MERT, ARC, Ligue Nationale contre le cancer and Société Française d'Hématologie. S.T. is financed by the MERT.

Ancillary