Effective EGFP knock-down by dsRNA electroporation
We first determined the most efficient conditions for the uptake of dsRNA between the 8-cell and blastocyst stages using labeled GFP dsRNA (540 bp) and were able to develop parameters for electroporation of dsRNA that did not require thinning of the zona pellucida (see Methods). Our final method permitted uptake of dsRNA into 100% of the electroporated embryos (n = 120). Optical sectioning by confocal microscopy at the blastocyst stage indicated that greater than 90% of the cells in each embryo were targeted by the labeled dsRNA on average (minimum 70%) (Fig. 1A).
To assess the relative efficiency of knock-down in the different cell lineages at both pre- and post-implantation stages, we used a transgenic mouse line expressing the fluorescent fusion reporter histone (H2B)-EGFP [11]. Intact embryos carrying the H2B-EGFP transgene, which is ubiquitously expressed in a high-level constitutive manner, were electroporated with GFP dsRNA (and LacZ dsRNA, as a control) at the early 8-cell and blastocyst stages. Confocal fluorescence microscopy following in vitro culture for 24 h showed that the delivery of the GFP dsRNA had been equally effective in reducing the levels of H2B-EGFP fluorescence in both trophectoderm and ICM lineages (Fig 1B,C). Quantification of the fluorescence intensity revealed a 73% overall reduction of fluorescence in the GFP dsRNA-treated embryos compared to the LacZ dsRNA-electroporated controls (Fig. 1C). The relative reduction of the fluorescence levels in the ICM and trophectoderm cells was calculated to be of similar degree. The overall reduction of H2B-GFP expression correlates well with the strong knock-down at the mRNA level (80%) as determined by RT-PCR (Fig. 1D).
Electroporated embryos were also transferred into the uteri of pseudopregnant females and recovered and analyzed post-implantation at E4.3 and E5.5. As before, the levels of histone-GFP fluorescence were much reduced in GFP dsRNA-treated embryos and to a similar extent in the different lineages of the pre-streak stage embryo. The epiblast, although of overall higher GFP fluorescence intensity than the surrounding lineages, had its relative fluorescence reduced to the same degree as the VE and ExE lineages: an average 65% and 52% reduction at E4.3 and E5.5, respectively (Fig. 1B,C). The GFP mRNA levels were still strongly depleted at these stages despite the differences of overall fluorescence intensity being less pronounced between the GFP dsRNA-treated embryos and controls (Fig 1D). This could be explained by the stability of the EGFP fusion protein, which remains bound to the target even during cell division when the nuclear envelope has broken down [11]. The reduction of H2B-GFP fluorescence levels induced by GFP dsRNA electroporation was actually surprisingly high for a highly expressed protein product that starts being expressed at the 4-cell stage and, because of its inherent stability, persists in the cells for long periods of time.
Bmp4knock-down recapitulates null phenotype and reveals new Bmp4 function
To further study the role of key signaling pathways in peri-implantation development, we chose to target the endogenous mouse gene Bmp4, a member of the TGF-β superfamily of secreted signaling molecules [12]. We wished to determine whether electroporation of pre-implantation embryos with Bmp4 dsRNA led to similar phenotypes, specifically in events around the time of gastrulation, as reported for the genetic loss of its function [13–15]. Bmp4 is first expressed in the ICM and polar trophectoderm of the embryonic day 3.5 (E3.5) blastocyst. As the embryo develops to E6.5 its expression becomes restricted to the ExE and is highest in the cells abutting the epiblast close to the boundary between the embryonic and extraembryonic regions of the conceptus [16]. After gastrulation, Bmp4 is expressed in epiblast-derived tissues, including the extraembryonic mesoderm [15].
We introduced Bmp4 dsRNA into intact E3.5 blastocysts by electroporation and first determined the efficiency of knock-down by measuring the Bmp4 expression levels 24 h and 48 h later by RT-PCR. Bmp4 mRNA was reduced in Bmp4 dsRNA-electroporated embryos at both 24 h (Fig 2A,B) and 48 h after electroporation (not shown), compared to the GPF dsRNA-electroporated experimental controls. Levels of Gap3dh (Fig 2A) and Stat3 mRNA (not shown) were unaffected, suggesting that the Bmp4 dsRNA electroporation resulted in sequence-specific knock-down of Bmp4 expression. Electroporated embryos were transferred into the uteri of pseudopregnant females and then recovered and analyzed at several stages post-implantation. Bmp4 expression was reduced by as much as 70% at E5.25 and 44% at E5.75, as determined by RT-PCR (Fig 2A,B). The Bmp4 dsRNA-electroporated embryos displayed a range of phenotypes characteristic of that of Bmp4 null embryos many of which arrest in development shortly after implantation, but a proportion showing some development to later stages [13]. Between E6.5 and E7.5, and at some later stages, 46% (n = 255) of Bmp4 dsRNA-electroporated embryos were reduced in size by at least 30%. They were also rounded, disorganized or necrotic, and had underdeveloped epiblasts (Fig 2C). This contrasted with the 1% (n = 202) of control embryos that were slightly delayed in development, but morphologically normal, demonstrating that the long dsRNA has a specific effect and that, in accordance to other reports, does not elicit non-specific responses at these stages [7, 9]. At E7.5 Bmp4 dsRNA-electroporated embryos also lacked or had a reduced mesoderm-derived allantois (Fig 2C, panel g).
Bmp4 is required for epiblast cell proliferation, mesoderm formation [13, 14] and has been suggested to promote differentiation of the VE [16]. We therefore examined the expression of a set of molecular markers of the prospective mesoderm (Brachyury, Eomes, Lhx1) and visceral endoderm (Cer-l, Lefty-1, Lhx1) at E6.5 and E7.5 following Bmp4 RNAi. Brachyury, Eomes and posterior Lhx1 expression was absent in all defective embryos. It was also absent or significantly reduced in over 52% of the remaining group of Bmp4 dsRNA-electroporated embryos that appeared morphologically normal, whether delayed in development or not (Fig 2E, panels a'-c', g', h', and Additional File 1). This reduced expression of mesoderm markers suggests that in some embryos primitive streak formation and the initiation of gastrulation probably occurred but failed to progress further. It is consistent with the reported phenotypes of Bmp4-/- embryos [13].
The consequences for the expression of markers of the anterior VE (AVE) were less expected. Before gastrulation, by E5.5, Cer-l and Lefty-1 transcripts are normally restricted to the distal tip of the VE (DVE). These DVE cells then move towards the prospective anterior of the embryo and so by E6.0 become the AVE that plays a key role in anterior patterning [17, 18]. We found that the expression of Cer-l was abnormal in Bmp4 dsRNA-electroporated embryos already at E5.5 and E5.75 (Fig 2E and Additional File 1). In a small number of embryos (3/16) we were unable to detect Cer-l transcripts at these stages (Fig 2E, panel q'). Rather than being slightly shifted towards the prospective anterior at E5.5 [18], we found that in 50% of Bmp4 dsRNA-treated embryos (n = 8/16), the Cer-l expressing domain was expanded symmetrically around the distal tip towards both the anterior and posterior regions of the VE (Fig 2E, panels k', p'). Between E6.5 and E7.5, Cer-l expression was either abolished or localized to the distal tip in all morphologically defective embryos. Moreover, in 52% (n = 16/31) of the remaining group of dsBmp4-treated embryos of normal morphology we observed an extensive diffuse region of ectopic Cer-l expression around the DVE and in the lateral VE (Fig 2E, panels d'-e'). Lefty-1 expression had a similar pattern to that of Cer-l in all defective embryos at these stages, as well as in 55% (6/11) of the remaining Bmp4 dsRNA-treated embryos, although no ectopic transcripts were observed (see Additional File 1). Lhx1, which was expressed throughout the AVE in the control embryos, was downregulated in 55% of the Bmp4 dsRNA-electroporated embryos (n = 6/11), thus confirming a disruption of AVE patterning.
Thus, our results indicate that down-regulation of Bmp4 expression changes the AVE expression domain. The expression of DVE markers either remains distal (Cer-l and Lefty-1) or spreads laterally (Cer-l). This suggests that Bmp4 signals that mainly originate from the ExE at these and earlier stages can regulate the restriction of the AVE expression domain and its specific positioning. This pattern of regulation is in agreement with a report that showed that BMP signaling promotes differentiation of VE in embryoid bodies [16]. It is also consistent with the phenotype of the BMP antagonists Chordin and Noggin double mutants, which fail to maintain expression of Cer-l at E7.5 [19], supporting the role of Bmp4 in regulating Cer-l.
Our results also indicate that the effects of Bmp4 in patterning the VE occur in a time window between E3.5 and E5.5. We observed that a proportion of embryos can "recover" from the transient nature of the knock-down during the course of the experiment, as determined by RT-PCR at E5.75 (Fig 2B) and at E6.5 by in situ hybridization (Fig 2E, panel i'). The transient nature of the RNAi effect can be seen as an advantage in a given experimental context, as it enables the period of gene expression required for events relating to embryo patterning to be pin-pointed. The definition of time intervals in which essential signaling molecules are required is critical for the understanding of the AVE specification and subsequent migration. Thus the approach we propose here might prove very valuable in addressing AVE function as it allows modulating gene expression just after embryo implantation.
Aside from these novel findings, the other phenotypes following Bmp4RNAi are strikingly similar to those reported for null mutant embryos both in form and variability. Approximately 20% of embryos developed to E9.5, albeit with severe defects [13]. In all, over 70% of the embryos were either morphologically abnormal and/or defective in the expression of essential mesoderm and VE markers following Bmp4 RNAi. It cannot be excluded that some part of the observed pleiotropy may be attributed to differential uptake of Bmp4 dsRNA, although our experiments with the H2B-GFP transgenic embryos indicate that all cell lineages are targeted by the dsRNA to a very similar extent.
Simultaneous knock-down of the three murine Dishevelledgenes
We next wished to determine whether this approach could enable functional redundancy between genes to be examined. To this end, we simultaneously targeted the three murine Dishevelled (Dvl1–3) genes [20–22]. The Dishevelled proteins are members of the Wnt developmental pathway, which regulates cell fate and subsequent cell behaviour in metazoans [23]. Analysis of the genetic null mutants (single and double mutants) has suggested redundancy of function among these genes [24–26]. To our knowledge a triple genetic knockout of these genes has not been reported, perhaps due to the difficulty of generating such a mutant. Although the three isoforms are expressed throughout pre-implantation development, a significant increase in expression is seen from the morula stage onwards, which is shared by a number of Wnt family members [27]. Thus, to ensure effective knock-down of all three Dvl genes we subjected embryos to co-electroporation with the corresponding three dsRNAs at two successive stages. The first treatment was given to 8-cell stage embryos and was followed by a second treatment when the embryos had developed to the blastocyst stage. A significant knock-down of the Dvl mRNA levels (average 70% Dvl1, 76% Dvl2 and 73% Dvl3) was observed 24 h (Fig 3A) and 48 h (not shown) after each electroporation treatment. We also electroporated each Dvl dsRNA independently and found that the mRNA knock-down achieved was similar to that observed upon co-electroporation of the three Dvl simultaneously (n = 10 for each Dvl).
Following a period of development in utero, the embryos were recovered and analyzed at E6.5 and E7.5 or E8.0. While the embryos treated with a single Dvl dsRNA did not shown any morphological abnormalities at these stages (for each Dvl: n = 14, E6.5 and n = 15, E7.5+), nearly half (42/89) of the Dvl1–3 dsRNA-electroporated embryos were severely retarded in development or showed morphological abnormalities (Fig 3B,C). Forty percent of the affected embryos had not gastrulated as assessed by morphological analyses (Fig. 3B, panel g). In contrast, only 2 of 72 dsGFP-electroporated control embryos were developmentally delayed but showed no apparent defects. Interestingly, approximately 12% (n = 65) of the Dvl1–3 dsRNA-treated embryos examined at E8.0 showed a grossly distorted anterior midline and neural fold abnormalities. These were developmentally retarded and had a sinuous or bent anterior axial mesendoderm, which seemed to be connected anteriorly to a single neural fold of either side (Fig 3B, panel c-e). These defects appear to represent an earlier onset of the later neural tube closure phenotype observed in Dvl1-/-; Dvl2-/- double mutants [25]. Unexpectedly, neither the expression of mesodermal markers Brachyury, Eomes, and Lhx1, the VE markers Cer-l and Lhx1, nor the epiblast marker Oct4 appeared to be abnormal in a set of developmentally delayed embryos following dsDvl1–3 dsRNAi (Fig 3D). This indicates that while the knock-down of Dvl was effective in the "delayed" embryos – and enough to cause developmental delays – it was not sufficient to prevent gastrulation, such as observed for the group of morphologically defective embryos. The failure to gastrulate observed in the latter set of embryos upon Dvl1–3 RNAi is in agreement with the role of the Wnt signaling pathway in primitive streak formation, as typified by the Wnt3 mutants and Lrp5-/-;Lrp6-/- double mutants [28, 29].
It is possible that the observed pleiotropy of the phenotype following Dvl down-regulation can be attributed to differential uptake of dsRNA. Based on the relative kinetics of EGFP and Bmp4 knock-down, it seems likely that the efficacy of knock-down and the rate of mRNA recovery are target dependent. Thus, factors such as the level of expression, turnover rate of mRNA species, and protein product stability/degradation might dictate the extent of the RNAi phenotype.
Our results have shown novel phenotypes following down-regulation of specific genes and so demonstrate that electroporation of pre-implantation embryos with single or multiple dsRNAs directed against genes encoding related proteins is effective.