Design of a novel reporter of Notch activity
To generate a reporter system able to monitor Notch activity during neurogenesis, a promoter that responds positively to Notch signalling in the developing nervous system must be linked to a complementary deoxyribonucleic acid (cDNA) encoding a reporter protein whose detection reflects the promoter's transcriptional activity. The chicken genome contains three Hes5 genes clustered in a 20 Kbp region on chromosome 21, all of which respond to Notch signalling [14]. This gene cluster is close to the Pank4 gene, which also flanks the single Hes5 gene in the mouse and human genomes [14]. In the chick, Hes5-1 is the gene located nearest to Pank4, and has the highest homology to mouse Hes5 (Additional file 1. Fig S1A, B), being expressed in progenitors within the ventricular zone of the neural tube [14]. Moreover, its promoter is highly similar to the human and mouse Hes5 promoter (53% and 58% identity, respectively, over the most proximal 400 nucleotides), containing several putative CSL-binding sites (CBS), with two highly conserved CBS in the most proximal region (Additional file 1. Fig S1C). We therefore chose a DNA fragment from the region immediately upstream of the predicted transcription initiation site of Hes5-1 to be used as the promoter in the Notch reporter system.
To visualize transcriptional activity mediated by the Hes5-1 promoter, a two kilobase pair DNA fragment upstream of the Hes5-1 translation start site was linked to a cDNA encoding VNP [33]. The fusion protein contains the fluorescent protein Venus - a derivative of enhanced yellow fluorescent protein (EYFP) with fast maturation and increased brightness [34] -, a nuclear localization signal (NLS) to facilitate single-cell analysis and, finally, a PEST sequence from the mouse ornithine decarboxylase protein that confers fast degradation to the fusion protein [35]. We also included the 3' untranslated region (UTR) of the Hes5-1 mRNA in the reporter construct, because the presence of this element further decreases the time during which the reporter is active (see below). In addition, to ensure that a poly(A) tail is present in the mRNA, we included a rabbit β -globin polyadenylation signal downstream of the 3'UTR. The final reporter construct (P
Hes5-1
-VNP-3'UTR
Hes5-1
-poly(A)) will be referred in this paper as pHes5-VNP (Figure 1A). We built also a control construct in which the Hes5-1 promoter is replaced by the constitutively active cytomegalovirus-actin-globin (CAG) hybrid promoter [36] - this construct (pCAG-VNP-3'UTR
Hes5-1
-poly(A)) will be referred as pCAG-VNP (Figure 1B).
Expression of pHes5-VNP recapitulates the endogenous Hes5-1expression
To determine if reporter expression driven by the Hes5-1 promoter is able to mimic the endogenous pattern of Hes5-1 transcription, the pHes5-VNP and pCAG-VNP constructs were electroporated separately into the chick embryonic neural tube, and expression of VNP mRNA driven by each construct was compared with Hes5-1 mRNA expression. Embryos were harvested 48 h after electroporation and the presence of VNP and Hes5-1 mRNAs were analysed by in situ hybridization in sections (Figure 2A-C). These data show that VNP mRNA transcription driven by the Hes5-1 promoter is restricted to the ventricular zone and resembles the endogenous Hes5-1 expression pattern (Figure 2, compare A with C). In contrast, VNP mRNA transcription driven by a constitutive promoter (pCAG-VNP) occurs not only in the ventricular zone but also in the mantle layer (Figure 2B), showing that the specific activity of pHes5-VNP in neural progenitors is not an artefact of the electroporation procedure. To confirm further that pHes5-VNP expression occurs in cells that transcribe the Hes5-1 gene, double in situ hybridization for VNP and Hes5-1 (using a probe for the coding region and excluding the 3'UTR) was performed. Confocal analysis shows co-expression of the two mRNAs within the same cells, confirming that the reporter driven by the Hes5-1 promoter is active only in neural progenitors (Figure 2D, E). As observed for Hes5-1 mRNA (Figure 2F, G), VNP protein is expressed in proliferating cells during mitosis (Figure 2H) and during S-phase (labelled with a short 5-bromo-2'-deoxyuridine (BrdU) pulse, Figure 2I). Furthermore, at 24 h after electroporation 99% of VNP-expressing cells do not co-express the early neuronal marker HuC/D [37] (4086 VNP+ cells, 143 sections, 4 embryos) (Figure 2J), indicating that VNP does not perdure in neural progenitors once these cells commit to differentiation.
Taken together, these results indicate that VNP expression driven by the Hes5-1 promoter faithfully recapitulates the expression of the endogenous Hes5-1 gene in neural progenitors and may therefore report Notch activity in this context.
pHes5-VNP provides a read-out for Notch activity
To test the responsiveness of the pHes5-VNP reporter to Notch signalling, we monitored the appearance of VNP in conditions where the Notch pathway is ectopically activated in the chick neural tube. With this aim, a group of embryos was co-electroporated with 3 plasmids: pHes5-VNP, pCAG-NICD (a plasmid driving constitutive expression of the Notch intracellular domain) and pCAG-CherryNLS (encoding a nuclear form of a red fluorescent protein driven by the constitutive CAG promoter, allowing visualization of the nucleus of all electroporated cells, Figure 1E) (Figure 3A-F). As a control, another set of embryos was co-electroporated with pHes5-VNP and pCAG-CherryNLS, but without pCAG-NICD (Figure 3G-J).
After electroporation, the fluorescent signal in the neural tube was observed in ovo over time, starting at 4 hours. At this time, VNP reporter expression is detected only in embryos where NICD is ectopically expressed. VNP expression increases strongly with time and, at 24 h, embryos show very high levels of expression in the electroporated neural tube (Figure 3A-C), similar to what is observed when Hes5-1 mRNA expression is analysed under similar conditions [14]. In contrast, the group of embryos without ectopic Notch activation shows very weak VNP expression at 4 hours (data not shown), and the levels present at 24 h are much reduced in comparison with those elicited by NICD overexpression (Figure 3G).
To quantify the percentage of electroporated cells expressing the reporter in each condition, we compared the number of VNP- and CherryNLS-expressing cells in sectioned embryos, 24 h after electroporation. Without ectopic Notch activation, 31.7% ± 8.1% of the electroporated cells (CherryNLS+) co-express VNP (n = 806 cells, 19 sections, 3 embryos) (Figure 3H-K). This figure is similar to the number of cells that express endogenous Hes5-1 mRNA in the developing spinal cord of embryonic day 3 (E3) embryos (31.0% ± 6.1%; 20444 cells, 51 sections, 5 embryos), suggesting that in the absence of ectopic Notch activation, VNP expression reflects endogenous Notch activity on the Hes5-1 promoter present in the electroporated plasmid. By contrast, when pCAG-NICD is co-electroporated with pHes5-VNP, virtually all electroporated cells express the VNP reporter (101.40% ± 2.36%, n = 1125 cells, 17 sections, 3 embryos) (Figure 3D-F,3K). These results show that the pHes5-VNP reporter responds to ectopic Notch activity and suggests that in the absence of pCAG-NICD, reporter activity reflects endogenous Notch signalling. To confirm that reporter expression in the absence of ectopic Notch activity reflects endogenous Notch signalling, we repeated the previous experiment, but in this case quenching endogenous Notch signalling in all electroporated cells by co-electroporating pCAG-CSLDN (a plasmid driving constitutive expression of a dominant negative form of CSL). Misexpression of CSLDN has previously been shown to down-regulate endogenous Hes5-1 expression [14]. When Notch signalling is abolished in electroporated cells, no VNP reporter expression can be detected 24 hours post electroporation (14/14 embryos, 2 independent experiments) (Figure 4, compare H-J with A-C). In addition, pCAG-CSLDN expression results in reduced Hes5-1 transcription and increased Delta1 expression (Figure 4, compare K-N with D-G), as expected when Notch signalling is blocked in the neural tube [14]. Similar results are obtained when embryos are harvested 8 hours post electroporation, when all electroporated cells are still in the ventricular zone of the neural tube (12/12 embryos, 2 independent experiments) (data not shown). Altogether, these results confirm that the reporter system using the Hes5-1 promoter responds effectively and specifically to Notch signalling activity in the chick embryonic neural tube.
The presence of the 3'UTR of Hes5-1reduces expression of reporter protein
Several mechanisms may restrain the duration of Notch activity in neural progenitors, including post-transcriptional regulation of mRNA turnover by control elements in the 3'UTR, as shown for the modulation of xHairy2 expression during somitogenesis in the frog [17]. To evaluate if the 3'UTR of Hes5-1 mRNA contributes to the post-transcriptional regulation of reporter expression, we built a derivative of the pCAG-VNP vector without the 3'UTR of Hes5-1, named pCAG-VNPΔ3UTR (Figure 1C). Both plasmids contain the same constitutive promoter driving VNP expression in all electroporated cells. We then compared the levels of VNP expression elicited by the two vectors in the chick neural tube. Each vector was co-electroporated with pCAG-CherryNLS as a control for electroporation efficiency. Embryos were collected 24 hours later and both red and yellow fluorescence intensities were analysed in whole embryos and after sectioning.
Our results show a strong increase of VNP expression in embryos electroporated with pCAG-VNPΔ3UTR (Figure 5A-C), when compared to embryos electroporated with pCAG-VNP (Figure 5D-F). Quantification of fluorescence intensities in neural tube sections shows that electroporation with pCAG-VNPΔ3UTR results in a 99.1% increase with respect to pCAG-VNP (p-value = 0.0001, t-test, 32 sections, total of 4 embryos in 2 independent experiments) (Figure 5J). These results show that the presence of the Hes5-1 3'UTR in the pCAG-VNP vector strongly reduces the levels of reporter expression, most likely by decreasing the half-life of the VNP mRNA or by preventing translation of the reporter protein.
Sequence alignment of the 3'UTRs of mouse Hes5, human Hes5 and chick Hes5-1 shows only very few conserved regions, the longest spanning 20 bp (Additional file 2. Fig S2). This region contains a sequence (cTATGATa) that resembles a K-box sequence present in Drosophila Enhancer of Split gene transcripts (consensus sequence cTGTGATa), which has been reported to be a binding site for specific microRNAs [38, 39]. To test if this conserved region is responsible for the post-transcriptional regulatory activity of the Hes5-1 3'UTR, we deleted it from pCAG-VNP and compared the levels of VNP expression elicited by this construct (pCAG-VNPΔ20 (Figure 1D)) to that of pCAG-VNP and pCAG-VNPΔ3UTR. The pCAG-CherryNLS vector was again used as a control for electroporation efficiency and embryos were harvested 24 h after electroporation (Figure 5G-I). Fluorescence intensities were quantified in cryostat sections. Comparison between cells electroporated with pCAG-VNPΔ20 or pCAG-VNP reveals that removal of the 20 bp sequence from the Hes5-1 3'UTR leads to an eight-fold increase of fluorescence levels (p-value = 0.0016, t-test, 32 sections, total of 4 embryos in 2 independent experiments), although still reduced in comparison to the fluorescence intensity of embryos electroporated with pCAG-VNPΔ3UTR (Figure 5J). This implies that the 20 bp conserved sequence has a significant contribution to the activity of the Hes5-1 3'UTR in the post-transcriptional regulation of linked mRNAs, although other regulatory elements must be present in the Hes5-1 3'UTR and contribute to the same end. Together, these results reveal that the Hes5-1 3'UTR is important to modulate the expression of the Notch reporter and its inclusion in the final construct is therefore crucial to provide a more faithful read-out of the dynamic nature of Notch signalling.
Live-imaging of pHes5-VNP reveals distinct behaviours of Notch activity in sibling cells
The rigorous validation experiments described above establish that the Hes5-1 promoter provides a highly accurate read-out of Notch signalling activity, and that the fluorescent VNP protein is a suitable reporter of promoter activation. We next used this reporter system to monitor Notch activity in embryonic neural progenitors, using slices of electroporated embryos cultured in chemically defined serum-free conditions for up to 72 hours [40]. Using these conditions, neural progenitors proliferate, show clear interkinetic nuclear movement and display the ability to give rise to differentiated progeny [40]. Embryos were co-electroporated with the pHes5-VNP and pCAG-CherryNLS vectors, so that electroporated cells could be permanently traced by red fluorescence, while VNP expression reflected the dynamics of Notch signalling. In separate control experiments, another group of embryos was co-electroporated with pCAG-VNP and pCAG-CherryNLS plasmids. In this case, expression of VNP and CherryNLS is driven by the same constitutive promoter. After electroporation, slices from both groups of embryos were prepared and cultured together in the same Petri dish. Fluorescent images were taken at 7 or 10 minute intervals, using a wide-field DeltaVision imaging system as described previously [40]. Slices were incubated for up to 48 hours and imaging started at either 4 or 12 hours after electroporation. Images were processed and analysed as described in Methods. From 36 embryos electroporated with pHes5-VNP in 12 independent experiments, 704 VNP-expressing cells were analysed in 36 embryonic slices. We did prospective and retrospective studies in individual cells aiming at identifying cell divisions and thereby defining cell lineages. From the 704 VNP-expressing cells, we have defined 175 lineages, each one containing only one of the 704 VNP-expressing cells, with VNP expression occurring before and/or after mitosis. The remaining 529 VNP-expressing cells were excluded from further analyses since we could not identify the previous cell division or the next, as they became out of focus or the time-lapse imaging ended before mitosis could be observed. Thus, from this analysis, we could identify 175 lineages of VNP-expressing cells and monitor their behaviour for different periods of time.
To correlate the timing of the onset of Notch activity in relation to the cell cycle, we analysed lineages where de novo pHes5-VNP expression could be clearly monitored (Figure 6 and 7). From the 175 VNP-expressing lineages, we could not observe de novo pHes5-VNP expression in 145 lineages, as these moved into the imaging planes already expressing VNP, or expressed it since the beginning of the imaging due to Notch activation soon after electroporation. We analysed instead 30 cell lineages where the timing of the onset of Notch activity could be clearly determined.
When correlated with cell cycle phase, these 30 lineages could be classified into two groups: one where Notch activation occurs before mitosis and another where activation occurs after cell division. In some cases, two mitotic events were observed within the same cellular lineage and each was analysed as a separate lineage.
In the first group, we observed that the onset of Notch activity occurs in a wide range of times before mitosis, from 1 to 27 h (24 lineages, Figure 6A, A' and Figure 7A. See also Additional file 3. Fig S3 and Additional file 4. Movie S4). In the majority of these cells (21/24), the VNP reporter is still detected in both daughter cells after cytokinesis and VNP expression is similar between siblings, suggesting that both cells experience Notch activity (Figure 6A, A'). In four of these lineages, we could observe that the two daughter cells divided again during the observation period (data not shown), revealing their progenitor character.
In the second group, containing cells in which the onset on Notch activity occurs after mitosis (11 lineages), we observed that in the majority of cases (10/11), Notch is activated in only one of the two sibling cells, an event that may occur at different times after cytokinesis, from 4 to 12 h (Figure 6B, B' and Figure 7A. See also Additional file 5. Fig S5 and Additional file 6. Movie S6). However, in one cell lineage we could detect Notch activation in the two daughter cells after mitosis, although with different onset times: one cell activating Notch 6 h after mitosis and the other 7 h later (Figure 6C, C' and Figure 7A. See also Additional file 7. Fig S7 and Additional file 8. Movie S8). These findings suggest that progenitors in which Notch is inactive during mitosis give rise to daughter cells with different potential to activate Notch signalling, or that activation of Notch signalling in this context is random (depending on their chances to contact a ligand-expressing cell). Overall, given that the average cell cycle length in chick neural progenitors is 16 hours [40], our results show that Notch signalling can be activated in different phases of the cell cycle and at a wide range of times.
We have also analysed Notch activity patterns generated by this reporter to elucidate the duration of signalling and its termination. A not infrequent event following transfection of chick embryos is plasmid loss [41, 42] and this might limit the use of the reporter as an indicator of Notch activity downregulation in neural progenitors, as cessation of promoter activity cannot be easily distinguished from plasmid loss. We assessed this by following electroporated cells where CherryNLS and VNP are driven by the CAG constitutive promoter. Our results show that while the stable CherryNLS protein persists for a long time in electroporated cells, VNP levels driven by the same constitutive promoter decrease more rapidly, after the initial post-electroporation increase (Figure 7B, B'. See also Additional file 9. Fig S9 and Additional file 10. Movie S10). Such decrease in VNP expression must be due to plasmid dilution and loss from electroporated cells, followed by degradation of the remaining unstable VNP mRNA and VNP protein, while the more stable CherryNLS perdures. This finding therefore precludes the routine use of plasmid electroporation as an assay to monitor the termination of Notch signalling using our reporter system. However, we did observe some cells where reduction of Notch activity is followed by a new transcriptional activation of the reporter, allowing us to exclude that the previous downregulation was due to plasmid loss. These cells were included in our analysis of Notch activation events and reveal that Notch signalling dynamics can be captured using this reporter. Altogether, these results show that the electroporation-based reporter assay is suitable for the analysis of patterns of Notch activation in neural progenitors, and that the timing of onset of this activity can be correlated to the behaviour of neural progenitors after division.