A-Gene transfer into early precursors from 7 dpc YS-blood islands
I-Determination of suitable stages for mesodermal/pre-hematopoietic cell targeting
To precisely identify the stage when blood islands are enriched in immature mesodermal or pre-hematopoietic cells, we used lmo2 mRNA expression as a marker of immature blood islands precursors. Indeed, the LIM-zinc finger transcriptional co-regulator lmo2 is required for the generation of hematopoietic precursors of both extra- and intra-embryonic origin [14,15]. Lmo2 is expressed in the extra-embryonic mesoderm that gives rise to hematopoietic precursors [16,17], but also by their erythroid progeny. The differentiation of blood islands mesoderm into mature erythroid cells occurs extremely fast. We thus used β-H1 mRNA expression, as the earliest marker of differentiating erythroid cells, namely primitive erythroblasts. We compared the temporal evolution of the expression of both markers, during early blood islands formation (7.25 to 8 dpc), to select the time window when lmo2 is expressed and β-H1 not yet/poorly expressed.
Lmo2 transcripts are first detected at the OB stage in the whole extra-embryonic mesoderm (Fig. 1A) and are subsequently restricted, at the EB stage (Fig. 1B), to a subset of mesodermal cells, which might correspond to blood islands precursors. Upon proliferation of these precursors, lmo2 expression expends to encompass all hematopoietic derivatives of the blood islands, from the LB to LHF stages (Fig. 1C, D) .
The first β-H1 transcripts are detected at the OB stage in a limited number of cells located within the lmo2-expressing ring (Fig. 1E). The number of cell clusters expressing β-H1 gradually increases, leading to the formation, at the LB stage, of a ring which expression overlaps that of lmo2 (Fig. 1F, G). During the following stages (EHF/LHF), this correlation of lmo2 and β-H1 expression is maintained (Fig. 1H). The correlation of these two expression patterns strongly suggest that mesodermal and/or pre-hematopoietic precursors are enriched in the OB to EB YS blood islands, even if mesodermal cells and native hematopoietic precursors may be still present in the blood islands at later stages. Targeting mesodermal cells was thus conducted at the OB to EB stages of development (Fig. 1 Lower panel).
II-Developmental constraints imposed on the choice of methodology
At these developmental stages, the determination and further proliferation/differentiation of hematopoietic precursors strictly depend on endoderm/mesoderm interaction. This implies that a method allowing the maintenance of tissue structure during transduction will be required for suitable development of a hematopoietic progeny from transduced cells. Accordingly, when we nevertheless tested the feasibility of dissociated cells transduction at the OB-EB stages using lipofection (using ExGen500) or retroviral transduction (see paragraph B2), we found that, in both instance, the transduction levels were naught, and cell viability drastically affected (Data not shown).
III-Viral-mediated transduction and in situ electroporation of 7dpc YS-blood islands
Given these constraints, two alternative approaches could be undertaken to transfer genes into the YS while keeping its tri-dimensional structure, viral-mediated transduction  and in situ electroporation. Indeed, both methods have already been used to transfer genes in other embryonic tissues and stages.
We first tested the ability of dissected YS to be transduced in toto by exposure to retroviral supernatant. OB-EB YS-explants were cultured in medium containing viral particles for 24 hours (for the method, see B). Explants were subsequently cultured in fresh medium for two days before assessing GFP levels and hematopoietic cell production. At the outset of the organ culture step, the morphology of transduced YS appeared affected by the process. Cytometry analysis of these YS explants showed that only 0.1% of viable cells expressed GFP. These cells were in addition unable to produce a hematopoietic progeny (data not shown).
We thus investigated in situ electroporation as a second approach.
1-General scheme of the in situ electroporation protocol
To optimize this method, we used the peGFP-C1 plasmid (Clontech), which brings on a ubiquitous expression of the GFP reporter gene driven by the CMV promoter. After injection of the plasmid into the YS cavity and in situ electroporation, the YS is dissected from the embryo and kept in organ culture for three days.
The evolution of electroporated YS is systematically compared to non-electroporated control and/or control YS electroporated without plasmid. The effect of in situ electroporation on the development and viability of YS-explants is assessed through morphology examination after the organ culture step, as well as by the detection of endothelial and erythroid cells differentiation within the explants. GFP levels allow the monitoring of transduction efficiency. Finally, the characterization of the hematopoietic progeny of GFP+ sorted cells, obtained upon culture on OP9 stromal cells, which normally allows the hematopoietic differentiation of mesodermal precursors , is used to typify the targeted precursors.
2-In situ electroporation parameters adapted to YS-mesoderm transduction
Previous studies involving in situ electroporation stressed the importance of several parameters for an efficient DNA delivery as well a proper preservation of cell viability, and an accurate tissue targeting [20,21].
The electroporation procedure (voltage (V), pulse duration (ms), number of pulses applied and intervals (ms) between two pulses) is critical for the maintenance of tissue integrity and effectiveness of construct delivery.
The choice of the electrode type used, the distance between them and their position regarding the injected site clearly constitute the most important parameters influencing cell targeting.
Targeting the YS-mesodermal/pre-hematopoietic precursors, while avoiding endoderm transduction, requires the injection of the construct into the YS cavity, since the mesoderm layer is facing this cavity while endoderm is exposed to the external environment. The embryos are freed from the decidua and Reichert's membrane (Fig. 2A, B) to allow injection. A capillary, back-filled with 1 μg/μl peGFP-C1 plasmid DNA solution is inserted into the YS cavity from the node region and through the amnios. Fast Green (0.01%) is added to the solution to visualise construct delivery. This injection mode, designed to limit plasmid release from the YS cavity, leads to the collapse of the amnion towards the ectoplacental cavity. Upon mouth driven-release of the plasmid, the amnion returns to its former position and an accurate injection can thus be visualized by the fast green filling of the YS cavity (Fig. 2C, D).
b-Electrode positioning and electroporation parameters
Electroporation allows gene delivery into the specified cell subset provided that the path of the negatively charged DNA towards the positive electrode (anode) is correctly adjusted. Careful examination of wholemount β H1 and lmo2 in situ hybridization patterns allocated the blood islands to the median third of the distance extending from the floor to the roof of the YS-cavity. Following injection, the two electrodes (gold genetrodes 512 from BTX) are thus positioned at each side of the YS (Fig. 2E), parallel to the presumptive blood islands ring.
We used a square-wave pulse generator (ECM830, BTX) to deliver the construct at a low voltage, thus preserving tissue integrity. The best results (efficient electroporation, preservation of explants morphology and recovery of a normal hematopoietic progeny from transfected precursors, see below) are obtained with five 50 ms pulses of 30 V at 500 ms intervals, with a 4 mm distance between the electrodes (Fig. 3). In theory, with such settings, a maximum 40–50% of the blood islands cells might take up the plasmid. In order to increase the targeted area, pulses were applied twice, with an inverted field, using the same parameter set-up. Unfortunately, the gain in the number of transduced cells was annihilated by a decrease in cell viability (Data not shown).
3-Development of electroporated YS-explants during organ culture
When placed in organ culture, YS-explants, both non-electroporated control and electroporated (with or without plasmid), rapidly form a bubble-like (Fig. 4A) structure that mimics YS development. It contains cell clusters similar to those observed in control blood islands (Fig. 4B). Moreover, differentiation along the erythroid (Fig. 4A, B) and endothelial (Fig. 4C, D) lineages normally occurs. After 3 days in organ culture, the explants have progressively organized into a bipolar structure with most mesoderm slightly adhering to the plastic dish, while the endoderm is exposed to the external environment, as shown by the expression of α-foeto-protein (AFP) transcripts, as a visceral endoderm marker  (Fig. 4E–H). Within the mesodermal layer, β-H1-expressing erythroid cells are mostly located at the adhering site (Fig. 4I, J see also Fig. 4A, L), while lmo2-expressing cells are more largely distributed in the YS-explants (Fig. 4K).
GFP can be visualized in electroporated YS after on average three hours and is still detected after the 3 days of organ culture. In YS explants, GFP+ cells are mostly located within the blood islands-like clusters and the adhering site (Fig. 4L–O), which provides a first indication that the prospective blood islands has been accurately targeted.
4-Hematopoietic recovery from electroporated YS-explants kept 3 days in organ culture
a-Characterization of the hematopoietic progeny of transduced cells
We next analysed by flow cytometry the presence of transduced hematopoietic cells in YS-explants immediately after the three days in organ culture (thereafter referred to as OrgD3-YS). The percentage of GFP positive cells represents on average 0,5–1% of the whole OrgD3-YS population (corresponding to 50–100 cells), a percentage classically obtained using unilateral electroporation procedure in mouse embryonic tissues [9,23]. The presence of both c-kit and CD45 positive cells within the GFP+ subset (Fig. 5A) confirms that mesodermal/pre-hematopoietic cells, capable to give rise to a hematopoietic progeny, were successfully transduced.
In order to assess whether the targeted mesodermal/pre-hematopoietic precursors keep a normal differentiation pathway, we next analysed the differentiation potential of sorted GFP+ cells after a 5 days culture on OP9 stromal cells. These cells gave rise to the typical YS progeny, namely Mac-1+/CD45+ macrophages and Ter119+ erythroid cells, and few c-kit+/CD45+ precursors (Fig. 5B), indicating that the GFP plasmid was indeed transferred to mesodermal/pre-hematopoietic precursors. Interestingly, GFP is not expressed by their hematopoietic progeny (data not shown), reflecting the non-integrated status, and hence the transient expression, of the transgene.
b-Viability of electroporated hematopoietic precursors
The frequency of hematopoietic precursors present in control OrgD3-YS, as well as in OrgD3-YS electroporated without plasmid, was quantified by culture in limiting dilution on OP9 stromal cells for 5 days. In control OrgD3-YS, an average 1 out of 480 cells gives rise to a hematopoietic progeny, which corresponds to about 10–11 precursors per YS (Fig. 5C, Table 1). A similar frequency (1 hematopoietic precursor out of 453 cells) is obtained from OrgD3-YS electroporated without plasmid (Table 1), indicating that our electroporation protocol does not affect the recovery/viability of hematopoietic precursors.
c-Efficiency of the gene transfer into mesodermal/pre-hematopoietic precursors
We next quantified the recovery of hematopoietic precursors derived from transduced mesodermal cells. As stated above, our unilateral electroporation procedure only delivers the construct to at the best 40–50% of the blood islands, corresponding to 5–6 precursors per YS.
Sorted GFP+ cells were distributed at 10 or 30 cells per wells on OP9 stromal cells and scored for the presence of a hematopoietic progeny five days later. In five independent experiments, on average 1/150 GFP+ cell displayed a hematopoietic potential (Table 1). This value corresponds to a 3-fold enrichment in hematopoietic precursors compared to control OrgD3-YS. This enrichment indicates the presumptive blood islands were accurately targeted during the in situ electroporation step. However, the absolute number of GFP+ hematopoietic precursors (0.56 ± 0.18 obtained per 0.5 YS-equivalent) remains low (Table 1), even though a hematopoietic progeny is systematically obtained from transduced mesodermal/pre-hematopoietic precursors.
B-Transduction of 8 dpc YS-hematopoietic precursors
I-In situ electroporation
We investigated whether in situ electroporation is also efficient to target hematopoietic precursors at later development stages (From EHF stage to the 5-S stage). The same settings were applied to transfer the plasmid to the blood islands. Electroporated YS were kept in organ culture for one day (instead of the 3 days in organ culture for 7 dpc YS, required for the differentiation of mesodermal/pre-hematopoietic precursors) to allow for GFP expression prior to sorting. Flow cytometry analyses, performed immediately after the organ culture step, showed that GFP-expressing cells again represented 0.5–1% of the whole population. Upon sorting and culture on OP9 stromal cells for 5 days, these cells produced a hematopoietic progeny comparable to that obtained from whole control (non-electroporated) YS. These data indicate that in situ electroporation may be used to transduce hematopoietic precursors at 8 dpc.
However, at this stage, the blood islands have developed into vessels filled with hematopoietic cells, so that the hematopoietic precursors are diluted amongst erythroid cells (Figure 1D, H). Thus electroporation appears less convenient at this stage while a method ensuring a high efficiency of transduction (with respect to the whole population) would be more adapted. In addition, at 8 dpc, contrary to 7 dpc, blood islands precursors can be directly obtained from dissociated YS. We took advantage of these features to examine whether retroviral transduction would be adapted to efficiently transfer genes into hematopoietic precursors from 8 dpc YS.
1-General scheme of the protocol
Transduction using ecotropic retrovirus has previously been used to transfer HoxB4 into 8 day YS hematopoietic precursors . However, neither quantitative data nor optimization of the various infection parameters was provided.
As a first step, the following protocol was designed to target gene to hematopoietic precursors from 8 dpc (0–4 somite stages) YS. In this study, transduction of hematopoietic precursors was obtained using a MSCV-based retroviral vector called MPI. This retroviral vector harbours, as a FACS-selectable reporter, the eGFP cDNA downstream a IRES element .
We first compared the transduction efficiency attained using freshly isolated YS or YS explants dissociated after one day in organ culture (thereafter referred to as OrgD1-YS). In both conditions, we systematically analysed: 1-GFP level at day 1 post-infection (i.e. 24 hours after exposure to the viral supernatant), 2- the persistence of GFP expression in YS-transduced cells during in vitro hematopoietic differentiation, 3- the viability and phenotype of the hematopoietic progeny of sorted GFP+ cells, 4- the number and the type of hematopoietic cells produced upon in vitro culture on OP9 stromal cells or in methylcellulose assay. In both instances (8 dpc YS freshly explanted or maintained in organ culture for one day), all these analyses were performed after GFP+ cell sorting at day 1 post-infection.
2-Retroviral transduction of 8 dpc YS on freshly isolated cells or after organ culture
In our first tests, we used the MPI retroviral vector to transduce 8 dpc YS directly upon dissection. After dissociation, YS cells are seeded in a 48-well plate in 1 ml medium supplemented with cytokines and 4 μg polybrene (see Materials and Methods). After addition of the retroviral supernatant (MOI = 1), the culture is maintained at 37°C, 5% CO2 for 12 hours and re-plated in fresh medium and either seeded on OP9 stromal cells or analysed for clonogenic potential in methylcellulose assay. In these conditions, the initial efficiency of transduction is reproducible and quantitatively high (60–70% GFP+ cells). However, the GFP+ population rapidly decreases in culture to completely vanish after 4–5 days (Fig. 6A). The 8 dpc YS is enriched in terminally differentiated hematopoietic cells and immature erythro-myeloid precursors only appear at the 2–5 somite-stage [25,26]. Accordingly, we suspect that the rapid exhaustion of the GFP+ population during the 4 days culture arises as a consequence of the relative maturity of the transduced cells, which therefore rapidly complete differentiation and disappear.
A mean to obtain a larger number of targeted hematopoietic precursors and still avoid cross-contamination by intra-embryonic-derived hematopoietic precursors, is to place the YS, dissected before the 4 somite-stage for one day in organ culture (OrgD1), as previously described [3,4]. Thus, dissociated OrgD1-YS cells were exposed to the retroviral supernatant with the same MOI and cultured as described above. Under these conditions, transduction efficiency is also highly reproducible, but lower (25–30%) than in the former conditions. However, in sharp contrast with the former conditions, the absolute number of GFP+ cells increases between 4 and 8 days post-infection (Fig. 6A). Moreover, the percentage of GFP+ cells also increases during the culture to reach a maximum (50–60%) at day 4 post-infection (Fig. 6B). Since about 15–20% of GFP-negative cells sorted at day 1 post-infection are shown to acquire GFP expression over the following days in culture (Fig. 6B), the increase in GFP+ population between day 1 and day 4 post-infection may result, at least in part, from de novo GFP expression. Interestingly, when observed in semi-solid culture, the colonies appear as entirely composed of strongly GFP positive cells, which indicates that the whole progeny of transduced hematopoietic precursors keeps the expression of the transgene (Fig. 6C). The threefold increase in the number of monopotent as well as multipotent clonogenic precursors among the GFP+ cells at day 4 compared to that obtained from 8 dpc YS transduced immediately after explantation (Fig. 6D), confirms that inclusion of a organ culture step leads to a transduction of a higher number of hematopoietic precursors. The persistence of GFP expressing cells, when transduced cells derive from OrgD1-YS, thus reflects that the organ culture step indeed improves the targeting of immature precursors.
We therefore focused on the protocol including the organ culture step for further analyses of the hematopoietic phenotype and differentiation potential of transduced cells.
3-Hematopoietic development from YS transduced cells
Flow cytometry analysis of the percentage of 7AAD labelled cells shows no difference in cell viability between infected or non infected YS-cells after one or four days in culture on OP9 stromal cells (Data not shown). In our conditions, the proliferation rate is perfectly equal between transduced and non-transduced cells (Data not shown).
The phenotype of hematopoietic cells recovered from OrgD1-YS at 24 hours post-infection is similar to that of control OrgD1-YS (data not shown). Moreover, the various hematopoietic subsets, both myeloid (CD45+) and erythroid (c-kit+/Ter 119+) are similarly distributed within the GFP+ and GFP- populations (Fig. 7A). The fact that the retrovirus-mediated transduction protocol does not disturb hematopoietic differentiation from OrgD1-YS is also evidenced upon phenotype analyses performed at day 4 after culture on OP9 stromal cells, since the relative percentage of erythroid and myeloid cells is not modified in transduced cells compared to normal YS (Fig. 7B).
We coupled the phenotype analyses to in vitro clonogenic assay to verify that the differentiation potential of transduced precursors was not modified by the whole procedure. As shown in Fig. 7C, in three independent experiments, the clonogenic potential obtained from sorted GFP+ does not significantly differ from that obtained from similar number of OrgD1 control (normal) YS, both at days 1 and 4 post-infection. The clonogenic potential thus correlates with the phenotype analyses performed at the same time points (Fig. 7A,B).
In conclusion, there is no phenotypic as well as differentiation potential bias between GFP+ and GFP- cells recovered from Org-D1 transduced YS. Moreover, our results (to be published elsewhere) obtained following retroviral transduction of a candidate gene into Org-D1 YS indicates that insertion of a 2 kb cDNA into the MPI vector: 1- does reduce neither the GFP level nor the percentage of transduced cells, 2- allows the identification of the exogenous protein through Western Blot. This procedure is thus adapted to analyse the effect of candidate gene on YS-hematopoietic cells development.