Hematopoiesis is the dynamic process whereby blood cells are continuously produced in an organism. Blood cell production is sustained by a population of self-renewing multipotent hematopoietic stem cells (HSCs) throughout the life of an organism. In early amniote embryos, the first functional definitive HSCs emerge from a splanchnopleura-derived region called AGM (Aorta-Gonads-Mesonephros), which includes the dorsal aorta floor, the mesonephros, mesentery, gonads, and the mesenchyme in between these structures (Cumano et al.,1996; Medvinsky and Dzierzak,1996; Muller et al.,1994). The first definitive HSCs produced in the embryo are thought to occur in dense clusters of hematopoietic cells budding from the floor of the aorta and also present in the vitelline and umbilical arteries. These intra-aortic clusters have been found in all vertebrate species studied so far (reviewed in Hartenstein,2006). Furthermore, transcription factors involved in HSC ontogeny are evolutionarily well conserved among vertebrates (for review, see Hsia and Zon,2005).
Transplantation experiments in birds have demonstrated that embryonic endothelial cells (ECs) originate from two different mesodermal lineages (Pardanaud et al., 1996). The aortic endothelium has a dual origin, roof and sides being contributed by somite-derived ECs and floor by splanchnopleura-derived ECs. As only splanchnopleura-born ECs display hemogenic capacities, intra-aortic clusters are restricted to the ventral aspect of the aorta. As hematopoiesis proceeds, the hemogenic endothelium also disappears from the aortic floor and is replaced by somitic ECs. Thus, the aortic floor appears as a transitory structure spent out in producing blood cells and replaced (Pouget et al.,2006). These experiments confirm that the dorsal aorta ventral endothelium has hemogenic ability, as proposed by the hemogenic endothelium hypothesis (Jordan,1916). This hypothesis has been supported as well by recent lineage tracing experiments in the mouse (Zovein et al.,2008). That close relationship of ECs and HCs led Murray to coin the term “hemangioblast” to refer to the components of the cell aggregates appearing in yolk sac and precursors of blood islands (Murray,1932). Today such hemangioblast is understood as a common precursor for both endothelial and hematopoietic lineages and its existence is still controversial.
Characterization of HSC is mainly done by bone marrow transplantation and relies mostly on the isolation of HSC populations from different sources according to specific surface markers (Sanchez et al.,1996). Actually, few attempts have been made to analyze HSC biology in their endogenous environment. The development of Cre transgenic lines in the mouse has approached the labeling of intra-aortic hematopoietic clusters, which contain functional HSCs based in transplantation experiments; however, up to now they failed to provide AGM-specific hemogenic labeling (reviewed in Yoshimoto et al.,2008). Further dissection of the regulatory sequences conferring HSC-specific expression would be required to achieve a more specific gene transfer strategy.
The accessibility of the chicken embryo has allowed the development of a number of manipulation techniques, including direct gene transfer, surgical manipulation, and time-lapse observation. Presently, gene transfer into chick cells is performed by three major systems: lipofection (which is based on the ability of liposomes to fuse with the cell membrane), electroporation (which uses electric pulses to make small holes in the cell membrane through which naked DNA molecules can enter the cells), and virus-mediated transfer (which allows integration of foreign genes into the chromosomes of host cells and, hence, the prolonged expression of the genes). Electroporation is the most widely used because it is highly efficient, fast, and requires only small amounts of DNA. Lipofection instead needs large amounts of DNA, whereas virus-mediated transfer requires a long period of time, to construct viral vectors, establish clones of producer cells, and assess viral titers. Also, the electroporation transfer has the advantage that naked plasmid DNA produces little antigenicity in the host (Gilkeson et al.,1991; Jiao et al.,1992). For these reasons, electroporation in ovo is today a routine tool for functional modification in the avian model (Muramatsu et al.,1997; reviewed in Itasaki et al.,1999).
This technique is efficient at introducing genes into adult vascular EC (Nishi et al.,1996) but efforts to electroporate DNA into the embryonic vascular EC have been unsuccessful, since labeling was not restricted to vascular EC but was also detected in tissues surrounding the vessel (Bollerot et al.,2006). Instead, lipoplex-mediated gene transfer was set up to modify embryonic vascular cells, with the drawback that modifications are driven to all the embryonic vasculature and not only to the hemogenic endothelium (Bollerot et al.,2006).
We present here a new protocol for gene transfer by electroporation into embryonic ECs and show that foreign DNA expression is local and constrained to the endothelial cells, not spreading initially to the surrounding tissues. By transferring expression constructs into the dorsal aorta ECs, we were able to direct gene expression to the hemogenic endothelium at the critical stages for HSC generation in the AGM. Moreover, gain-of-function using expression constructs for chick Lmo2 and Tal1 genes was achieved.
This method provides a fast and efficient means of generating gain- or loss-of-function models to investigate the function of genes involved in blood cell generation from hemogenic endothelium, including HSC birth. Furthermore, the simple experimental design and low cost make this an attractive tool for rapid large-scale screening of gene function.
RESULTS AND DISCUSSION
Setting Up Conditions for Endothelium-Specific Gene Transfer by Electroporation in the Avian Embryo
Embryos were electroporated between stages HH14 and HH18, to target gene expression during the period just prior to intra-aortic hematopoietic clusters generation from the hemogenic endothelium in the AGM. Intraembryonic hematopoiesis in the chick is not evident during dorsal aorta fusion (HH16), but at HH18 ventral ECs start to down-regulate EC-specific markers and to up-regulate HC-specific markers. The first active hematopoietic sites are observable at HH20, with hematopoietic cells budding off from the endothelial wall into the aortic lumen. Taking advantage of the tubular shape of the dorsal aorta, we injected DNA directly into its lumen (Fig. 1). The inner wall of the dorsal aorta at these stages consists of a single layer of endothelial cells, all exposed to the DNA confined to the lumen. Previous attempts to electroporate aortic endothelium involved intra-cardiac DNA injection, exploiting the heartbeat to distribute the DNA through the arterial tree. In contrast, we slowed the heart rate to ∼1 beat/minute, allowing us to inject the DNA directly into one of the paired dorsal aortic branches and apply the electroporation pulses between beats. This allowed the DNA to concentrate in the area of interest, increasing the spatial specificity and efficiency of electroporation. Moreover, injection before fusion of the paired aortae automatically generated electroporated and control sides within the fused aorta.
Electroporation parameters, including voltage; pulse length, interval, and number; and electrode design and position, were optimized as depicted in Figure 1. In summary, the most suitable combination of parameters consisted of 5 pulses (50 ms, 18V with a 100-ms interval between pulses), with the electrodes placed above and below the embryo and parallel to the A-P axis of the embryo (Fig. 2). Gene transfer and expression dynamics were monitored using pCAGGS/SE-EGFP reporter construct. Seventy percent of specimens were specifically labelled, with GFP signal detectable as early as 8 hours-post-electroporation (hpe) and reaching a maximum at nearly 24 hpe. Beyond this time, and especially after 30 hpe, lethality increased sharply and in parallel EGFP labelling decreased (Fig. 2B). Viability of electroporated embryos was compared to that of control embryos of three kinds: embryos counterstained with neutral red, embryos counterstained with neutral red and cooled to slow heart beating, and embryos electroporated without injecting DNA constructs. In the two first groups, 100% (6 out of 6 and 5 out of 5, respectively) embryos survived to treatment and developed normally. In the third group, 60% (3 out of 5) of the embryos died immediately after electroporation. This percentage is due to the electric field that is applied to the embryo or to the placement of the electrodes.
Regarding the stage of the electroporated specimens (Fig. 2C, D), we observed a different behavior when monitoring the embryos at 14 or 20 hpe. Younger embryos seemed to recover poorly from manipulation, since 35% of them were found dead either at 14 or 20 hpe. HH16 and HH17 embryos instead seemed to recover better, as lethality at 20 hpe was less than 15%. Efficiency peaked at HH15 14 hpe and HH17 20 hpe; however, the percentage of non-labeled or un-specifically labeled specimens was higher for HH17 embryos (around 30%), which may be due to resistance to the establishment of the electric field across the specimen exerted by more developed tissues.
Efficient Expression of Reporter Genes Restricted to Endothelial Cells
On tissue sections, GFP expression was detected in vascular EC of the dorsal aorta and vitelline artery (Fig. 1H–J). To check that gene transfer was restricted to the EC lining of the aorta, lectin-labelling of EC was used. Lens culinaris agglutinin (LCA) is useful for visualization of cells lining the vessels of early chick embryos (Jilani et al.,2003), including the intra-aortic hematopoietic clusters (Fig. 3B–C).
Expression of the CFP encoded by the reporter construct was largely confined to the endothelium (Fig. 3C–E). Furthermore, positive endothelial cells were detected in small populations lateral to the neural tube incorporated into vessels that branch from the roof of the aorta (Fig. 3G).
Occasionally, CFP was detected in circulating blood cells (data not shown), which might result from the labeling of hematopoietic precursors lining the wall of the aorta (hemogenic endothelium) or the labeling of hematopoietic precursors after they were released into the circulation. In addition, occasional cells of round appearance were found below the aorta (Fig. 3D), which is in accordance with previous reports of ingression of hematopoietic cells from the hemogenic endothelium into the subjacent mesenchyme (Bollerot et al.,2006).
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 hematopoietic-specific gene Lmo2 (Fig. 3H–I).
Functional Overexpression of Transcription Factors Implicated in HSC Birth
To test the suitability of our approach for functional overexpression, we determined its ability to trigger a known pathway in the vascular endothelium. Mouse Runx1 encodes a transcription factor of the Core Binding Protein family whose expression associates with HSC production in the hemogenic endothelium (North et al.,1999,2002; Chen et al.,2009).
The Runx1 avian homolog Runtb2, is expressed at HH18 in EC of the ventral aortic floor: expression begins when the aortae are still paired and persists upon fusion of the vessels. At the time of hematopoietic cluster production, Runx1 becomes restricted to HC bulging into the aorta lumen (Jaffredo et al.,2005).
A transcription factor complex involved in Runx1 activation has been described bound to the + 23 Runx1 enhancer in vivo, so that it regulates Runx1-mediated HSC emergence (Nottingham et al.,2007). This complex includes Gata2, the Ets factors Fli-1, Elf-1, Pu.1, and a Scl/Lmo2/Ldb1 complex. On the other side, Scl/Tal1 is a direct regulator of Runx1 during hematopoiesis, according to studies in zebrafish (Patterson et al.,2007) and mouse (Landry et al.,2008). To test the effectiveness of our system for functional in vivo genetic manipulation of the hemogenic endothelium, we monitored the expression of Runtb2 after simultaneous gene transfer of Lmo2 and Tal1 by electroporation at HH16, before the start of endogenous Runtb2 expression in hemogenic endothelium. We found that the combined expression of these two factors is sufficient to induce premature and ectopic expression of Runtb2 in the endothelium of the aorta-vitelline artery connection, which suggests the induction of the HSC-specific pathway (Fig. 4C–F). This effect is not observed when only the reporter construct encoding GFP is applied (Fig. 4A–B).
Here, a new in vivo approach to genetically manipulate the hemogenic endothelium was reported. Taking advantage of the susceptibility of chick embryos to manipulation and the wide use of gene transfer by electroporation in these embryos, we developed a system for gene transfer into the hemogenic endothelium of the dorsal aorta in chick embryos just prior to the appearance of intra-aortic HSC clusters. Gene transfer by electroporation, which until now had been applied to central nervous system, somites, limb mesenchyme, lens, and surface ectoderm, can also be applied to the endothelial lining of the dorsal aorta in the avian embryo. This technique had been reported as efficient at introducing genes into adult vascular EC (Nishi et al.,1996), but not in the vertebrate embryo. Modifications in the injection procedure and a fine-tuning of electroporation parameters were instrumental to allow the specific delivery to the embryonic aortic ECs avoiding plasmid entry into non-desired abluminal cells. By transferring expression constructs into the dorsal aorta ECs, we were able to direct gene expression to the hemogenic endothelium at the critical stages for HSC generation in the AGM. Functional overexpression of HSC-signature transcription factors resulted in the induction of HSC markers in the hemogenic endothelium, indicating the suitability of the approach for functional analysis of HSC generation. Considering that transcription factors for the ontogeny of HSCs are evolutionarily well conserved among vertebrates (Hsia and Zon,2005) and that they are implicated in the induction of leukemias and other tumors, our approach of gene transfer to the hemogenic endothelium constitutes a powerful tool to dissect the early molecular events involved in HSC generation and in the initiation of leukemic transformation, and to describe accurately the hierarchy among the transcription factors implicated in these two processes. Its simple experimental design and low cost make this an attractive tool for rapid large-scale screening of gene function.
Electroporation of DNA Constructs Into Dorsal Aortic Endothelium in the Chick Embryo
White Leghorn chick embryos (Granja Santa Isabel, Cordoba, Spain) were staged according to Hamburger and Hamilton (HH) (1951). Embryos were visualized by injecting neutral red solution (1/10 in distilled water) into the subgerminal cavity. Circulation was slowed by 30-min incubation at 4°C. Chick Lmo2 cDNA (full-length sequence amplified with primers 5′ GATATCCCATCCCTGCCAATGTCAT C 3′ and 5′ CATGGCGGCCGCGCCC CTCTTTGCTATATCATTCC 3′) was cloned into the expression vector pCAGGS/SE (a β-actin promoter- and cytomegalovirus (CMV) enhancer-driven expression vector; Stuhmer et al.,2002). Chick Tal1 full-length cDNA was cloned from a pBSKII+ plasmid (kindly provided by T. Jaffredo) into pCAGGS/SE EcoRI XhoI sites.
Both constructs were diluted in RNase-free water (3–4 μg/μl) at a 6:1 ratio with a pCAGGS/SE reporter construct encoding enhanced green or cyan fluorescent protein (EGFP or ECFP; Momose et al.,1999). Fast green (Sigma-Aldrich, St. Louis, MO) was added to a final concentration of 0.36 μg/μl. The DNA mix was administered with a borosilicate glass needle pulled in a micropipette puller into a tip of approximately 7–10 mm in length, and was inoculated into the intersection of the vitelline artery and the right branch of the dorsal aorta at HH16. Platinum electrodes (Sonidel Ltd, Dublin, Ireland) were placed in parallel above and below the embryo along the anteroposterior axis and 5 pulses (50 ms, 18V, 100 ms interpulse) were immediately applied. Optimal conditions for efficiency and viability need to be calibrated for each set of electrodes.
Intravital Lectin Injection
Before recovery, some embryos were injected with 0.2 μg fluorescein isothiocyanate (FITC)-conjugated Lens culinaris agglutinin (LCA) (Vector Laboratories, Burlingame, CA). FITC-LCA (50 ng/μl in PBS) was injected into the vitelline vein. Perfusion was achieved by closing the main yolk arteries with forceps and allowing solutions to circulate for 3 min.
Analysis of Electroporated Embryos
Recovered embryos were fixed overnight in 4% paraformaldehyde in PBS (PFA) in the dark. Some embryos were cryopreserved, then 8-μm sections were obtained, air-dried (1 hr in the dark), and mounted in Vectashield (Vector Laboratories). Other embryos were embedded in 4% low melting temperature agarose, 50-μm vibratome sections were obtained, dried (45–60 min in the dark), counterstained with ToPro3 (Invitrogen Corp, Carlsbad, CA), and mounted in Vectashield. Others were paraffin embedded, sectioned, and in situ hybridised for chick Lmo2 (with a probe targeting a 3′UTR sequence amplified with primers 5′ CACGATGTGATAACGAATGCTC 3′ and 5′ AATAACCCCCACCCTCTTCCCC 3′), according Mallo et al.,2000. These sections were then incubated sequentially with anti-GFP antibody (Clontech Laboratories, Inc, Mountain View, CA), streptavidin ABC HRP complex, and the substrate DAB (3,3′-diaminobenzidine; Vector Laboratories). Sections were counterstained with nuclear fast red solution (Sigma-Aldrich) and mounted in Aquatex (Sigma-Aldrich). Whole-mount in situ hybridization was performed according to Wilkinson (1995) using a probe against Runtb2 (Jaffredo et al.,2005).
Stained sections were observed with a microscope (Nikon, Eclipse 90i), and images were captured with a digital camera (Nikon DXM1200C) controlled by NIS-Elements D 2.30. Alternatively, a confocal microscope (Leica, Confocal microscope TCS SP5) was used. Digital images were adjusted with Adobe Photoshop CS4.
We thank A. Morales for the pCAGGS/SE construct, T. Jaffredo for the pBS·Tal1 construct and for the Runtb2 probe, S. Bartlett for English editing, and members of the lab for encouragement and discussion, especially Carlos G. Arques for a fruitful contribution to this work. The CNIC is supported by the Spanish Ministry of Science and Innovation and the Pro-CNIC Foundation.