|Home | About | Journals | Submit | Contact Us | Français|
Dbx homeodomain proteins are important for the production of multiple spinal cord cell types. To examine the regulation of Dbx genes in more detail, we have generated transgenic zebrafish in which fluorescent protein expression is driven by predicted dbx1a enhancers. We identified three areas of sequence conservation upstream of the dbx1a coding sequence and generated fluorescent reporter constructs driven by these predicted enhancer elements and the endogenous dbx1a promoter. In multiple stable insertions of a 3.5kb enhancer fragment, we observed that there was additional reporter expression in the dorsal spinal cord not normally observed by dbx1a in situ hybridization. In addition, these lines exhibited only transient reporter expression, unlike the endogenous gene. Surprisingly, a single insertion line expressed the reporter in the endogenous pattern, indicating that other local regulatory elements modulate gene expression through the 3.5kb enhancer.
In the vertebrate spinal cord, a set of homeodomain transcription factors defines specific populations of progenitor cells at defined dorsal-ventral positions (Jessell, 2000). Hedgehog (Hh) signals from the floor plate and notochord, as well as Bone Morphogenetic Proteins (BMPs) and Wnts from the dorsal spinal cord, act to regulate the expression of these homeodomain proteins (Jessell, 2000). One class of these proteins, encoded by Dbx genes, does not appear to require any of these signals for positive regulation of expression. Instead, both dorsal and ventral signaling molecules are required to restrict the expression of Dbx genes to a narrow stripe at an intermediate dorsal-ventral position (Timmer et al., 2002; Wijgerde et al., 2002; Alvarez-Medina et al., 2008; Bonner et al., 2008). While retinoic acid (RA) has been suggested to induce Dbx expression (Pierani et al., 1999; Novitch et al., 2003), it is not clear whether this activity is direct, or is instead required to antagonize inhibitory dorsal signals (Gribble et al., 2007). The identity of specific positive and negative factors acting on Dbx expression, and the mechanism of their activity, therefore remain undetermined.
The two amniote Dbx genes, Dbx1 and Dbx2, are required for the generation of V0 and V1 ventral interneurons respectively (Pierani et al., 2001), indicating that they play a role in specifying particular neuronal fates. Furthermore, fate-mapping of the Dbx1-expressing lineage in mouse showed that these progenitors also give rise to radial glia, astrocytes and oligodendrocytes (Fogarty et al., 2005). Because of this, we were interested in examining the genomic elements that regulate Dbx1 expression in the vertebrate spinal cord, using zebrafish as a model.
In zebrafish, two Dbx1 orthologs, dbx1a and dbx1b, are expressed and regulated similarly to their amniote counterparts (Gribble et al., 2007). In particular, dbx1a expression begins at the earliest stage of spinal progenitor specification, and is negatively regulated by Hh signaling (Gribble et al., 2007). In this study, we identified evolutionarily conserved genomic elements upstream of the dbx1a coding region, and asked whether they could drive reporter transgene expression in a similar manner to the wild-type gene. We found that a 3.5kb fragment immediately 5' to the mRNA was capable of driving GFP expression in the spinal cord, but in the majority of transgenic insertion lines this expression did not mimic the endogenous pattern. Instead we found ectopic transgene expression in the dorsal spinal cord, and premature silencing of the transgene in progenitor cells. However, a single insertion line expressed the transgene in the appropriate endogenous pattern, indicating that the local chromatin environment may be important in regulating dbx1 expression.
Adult wild type (AB-1) fish were bred according to standard methods. Embryos were raised at 28.5°C in E3 embryo medium and staged by time and morphology (Kimmel et al., 1995). For in situ hybridization, embryos were fixed in 4% paraformaldehyde for 3 hours at room temperature or overnight at 4°C, washed briefly, dehydrated, and stored in 100% MeOH at −20°C until use. Cyclopamine treatments from 50% epiboly to 24 hpf were performed as described previously (Gribble et al., 2007).
Whole-mount in situ labeling for was performed as previously described (Oxtoby and Jowett, 1993) using probes for egfp (Dorsky et al., 2002), and dbx1a (Gribble et al., 2007). For combined in situ and antibody staining, embryos were first stained with the egfp probe then sectioned and processed for immunohistochemistry. Fixed embryos were frozen and sectioned on a cryostat at a thickness of 12 μm. Sections were stained with rabbit-anti-GFP primary antibody (Molecular Probes, A11122, 1:500), and goat anti-rabbit Cy3 (Jackson, 111–165–003, 1:1000) secondary antibody.
PCR primers used to clone dbx1a genomic fragments:
dbx1a-1.1kb forward primer: 5' CGGTTGGTTTGGAAAGAGAG 3'
dbx1a-3.5kb forward primer: 5' CAAGTGACCCCTTCTGAGTGA 3'
common reverse primer: 5' TCAGTTCAGAGCTTGCCAGA 3'
dbx1a-8kb forward primer: 5' GGAAAGAAAGAACAAAAGTTACAGTCAAGCACAC 3'
dbx1a-8kb reverse primer: 5' GCCAAAAGCATTTCACACATCTCAAACAG 3'
For cloning into pCS2+, PCR with Advantage Taq polymerase (Clontech) was performed using standard conditions from total genomic DNA, and PCR fragments were gel purified prior to cloning. PCR products were cloned into the pCRII-TOPO vector (Invitrogen), then subcloned into pCS2+ containing a gfp cDNA, digested to remove the CMV promoter.
For Gateway cloning (Invitrogen), PCR was performed by incorporating attB4 sequence into the dbx1a-3.5kb forward primer, and attB1 sequence into the common reverse primer. A BP reaction was then performed into a 5' donor vector (pDONRP4-P1R). LR reactions were then performed with this 5' clone, pME-EGFP and pME-mCherry middle clones, and a p3E-poly 3' clone, in a Tol2 plasmid backbone (pDestTol2pA) (Kwan et al., 2007).
The dbx1a-8kb fragment was made by PCR amplification using the specific forward and reverse primers listed above, followed by cloning into pCRII-TOPO. This fragment was then attached to the 5' end of the dbx1a-3.5kb using fusion PCR. The entire combined fragment was then BP and LR cloned using the protocol described above.
Injection of DNA constructs and raising of stable transgenic lines was performed essentially as described (Kwan et al., 2007). For pCS2+ plasmid injections, 50pg of DNA in 1nl was injected at the 1-cell stage. For Tol2 constructs, 20pg of DNA was co-injected with 20pg of Tol2 transposase mRNA in a total volume of 1nl at the 1-cell stage. Embryos were screened for GFP or mCherry expression at 24hpf, and embryos showing transient expression were raised to adulthood. Four stable lines were established for this project: Tg(−3.5dbx1a:gfp)zd2, made with a pCS2+ vector construct; Tg(−3.5dbx1a:gfp)zd3, made with a Tol2 construct and expressing GFP in the endogenous dbx1a pattern; Tg(−3.5dbx1a:gfp)zd4 and Tg(−3.5dbx1a:mcherry)zd5, made with Tol2 constructs and expressing reporters in non-endogenous patterns. While these lines were not verified as single-copy insertions, all their progeny express the transgene in 50% of offspring from an outcross, suggesting that they only express from a single locus or multiple closely-linked loci. All other founders analyzed for this work were not maintained as stable lines.
Whole-mount images were taken on a dissecting microscope using brightfield and fluorescence microscopy with embryos mounted in 70% glycerol. Sections were photographed on a compound fluorescent microscope. All images were obtained using Optronics PictureFrame software, and adjusted for brightness and contrast using Adobe Photoshop.
Genomic DNA was extracted from transgenic embryos by incubating in 500μg/ml proteinase K at 55°C for 2 hours, followed by phenol/chloroform extraction and ethanol precipitation. DNA was digested with HaeIII and then self ligated. Nested PCR was performed using primers complimentary to the Tol2 vector terminal inverted repeats, and products were sequenced.
To identify local regulatory elements that could potentially drive dbx1a expression, we searched for conserved genomic regions using the UCSC genome browser (Kent et al., 2002). Using the “conservation” track in the browser, we found 3 upstream elements within 8kb that were highly conserved between zebrafish and mammals (Fig. 1A). The most proximal element was located within 1.1kb upstream of the translation start site and likely included the gene promoter. The next element was approximately 3.5kb upstream, and the most distal element was approximately 8kb upstream. In addition, we noticed high conservation within the first intron. We decided to first test the 1.1kb and 3.5kb elements in transient and stable transgenic reporter assays to determine whether they contained enhancer activity.
Both upstream elements (dbx1a-1.1kb and dbx1a-3.5kb) were PCR amplified from genomic DNA and subcloned into the pCS2+ vector containing a gfp cDNA (Fig. 1B). The DNA constructs were injected into zebrafish embryos at the 1-cell stage, which were analyzed for GFP expression and raised to adulthood. Both constructs produced scattered GFP-positive cells in the brain and spinal cord of injected embryos (not shown), indicating that they contained functional enhancer and promoter elements. However, embryos injected with dbx1a-1.1kb had noticeably dimmer expression and fewer GFP positive cells. When these embryos were raised and outcrossed to screen for transgenic founders, 0/120 fish injected with dbx1a-1.1kb showed GFP expression in their offspring. We identified 1/156 fish injected with dbx1a-3.5kb as a transgenic founder, and embryos from this line [Tg(−3.5dbx1a:gfp)zd2] showed strong GFP expression in the posterior spinal cord, disappearing more rostrally and completely absent by 30hpf (not shown). Interestingly, in these embryos GFP was also expressed throughout the dorsal spinal cord (Fig 2A,B), unlike endogenous dbx1a, which is limited to an intermediate region (Gribble et al., 2007). These data suggested either that the 3.5kb enhancer was insufficient to drive endogenous spinal cord expression, or that expression in this line was modified by position effects from surrounding genomic DNA.
To test whether the dbx1a-3.5kb enhancer was sufficient to drive transgene expression in an endogenous pattern, we took advantage of the Tol2 transposon system, which allows a much higher transgenesis rate than plasmid injections (Kawakami et al., 2004). We constructed transgenes containing the dbx1a-3.5kb enhancer followed by either gfp or mcherry cDNAs in a Tol2 vector backbone and injected these constructs into 1-cell embryos with tol2 mRNA. After raising embryos with specific mosaic transgene expression, 75% (21/28) were identified as transgenic founders. Interestingly, in embryos from 20 out of 21 founders identified, we observed a similar expression pattern to our original plasmid-derived line. When we analyzed embryos from two stable lines [Tg(−3.5dbx1a:gfp)zd4 and Tg(−3.5dbx1a:mcherry)zd5], we found that the transgene was expressed only transiently in spinal progenitors (Fig. 2 C–I), and reporter expression was present throughout the dorsal spinal cord (Fig. 2J). Furthermore, when we tested whether reporter expression was sensitive to Hh signaling by incubating embryos in cyclopamine, we found that GFP expression in the spinal cord was not ventrally expanded in the same way as the endogenous gene (Gribble et al., 2007) (Fig 2 K,L).
Based on these results, we hypothesized that the dbx1a-3.5kb enhancer might not be sufficient to drive transgene expression in an endogenous pattern. We therefore tested a construct that included the entire 3.5kb fragment plus a third conserved element, located 8kb upstream (Fig. 1). Because we were unable to PCR amplify the entire 8kb fragment, we cloned the 8kb element separately and combined it with the dbx1a-3.5kb fragment by fusion PCR. The intervening region contains highly repetitive GC-rich DNA and did not contain any identifiable regulatory elements. The resulting enhancer construct (dbx1a-8kb) was subcloned upstream of gfp cDNA in a Tol2 backbone. When we raised fish injected with this construct, we did not observe any increase in transgene specificity; embryos from 16 founders exhibited transient expression throughout the dorsal spinal cord (Fig. 2M,N). Interestingly however in all these insertions we also observed GFP in the optic tectum, which is not an endogenous site of dbx1a expression at or before 24hpf (Fjose et al., 1994)(Fig. 2O). We therefore concluded that the endogenous spinal cord expression pattern of dbx1a was not encoded by the distal 8kb element.
Surprisingly, a single founder injected with the dbx1a-3.5kb enhancer produced embryos with transgene expression that closely resembled the endogenous pattern of dbx1a in the spinal cord, persisting in progenitors throughout the rostral/caudal axis (Fig. 3A–F). This line [Tg(−3.5dbx1a:gfp)zd3] showed specific expression in intermediate progenitors, with clear gaps both dorsally and ventrally (Fig. 3G–I), in the same region as endogenous dbx1a (Fig. 3J). Furthermore, expression was present in CoPA neurons, similar to the endogenous gene (Gribble et al., 2007) (Fig. 3K,L). The transgene was also modulated by Hh signaling in the same way as endogenous dbx1a expression, as we observed a clear ventral expansion in cyclopamine-treated embryos (Fig. 3M,N).
Together, these observations suggested that the dbx1a-3.5kb fragment is sufficient to drive transgene expression in the endogenous pattern, but is highly dependent on the local chromatin environment. To identify the insertion site in our single line with correct expression, we performed inverse PCR using the Tol2 vector backbone sequence. We found that the transgene insertion was located at position 25621762 on chromosome 2 (Ensembl genome assembly Zv8). This region is relatively gene-poor, with the closest identified gene approximately 80kb away. Intriguingly this gene, cdh7, is expressed in the spinal cord in a similar pattern to dbx1a (Liu et al., 2007). However we do not believe that this line simply represents an enhancer trap, because reporter expression in the spinal cord arises much earlier than endogenous cdh7 (Fig. 3A,D,G). Further analysis will be necessary to determine whether the local insertion region contains a unique chromatin structure, a discrete enhancer element, or is otherwise similar to the endogenous dbx1a locus.
In conclusion, we have identified a 3.5kb regulatory sequence capable of driving reporter expression in the spinal cord in the same pattern as the endogenous dbx1a gene, but only in a very specific chromatin environment. Although the stable lines that we analyzed expressed the transgene in a Mendelian ratio (not shown), we cannot rule out the possibility of multiple expressed or silent insertions in many of our transgenic founders. In fact it has been noted previously that Tol2-mediated transgenesis often results in multiple insertion events (Urasaki et al., 2006). The fact that 20 founders, carrying undoubtedly more than 20 insertions, all exhibited the same expression pattern strongly suggests that in most loci the 3.5kb enhancer drives ectopic dorsal spinal cord expression and is prematurely silenced in intermediate progenitors.
Repression of dbx1a in dorsal progenitors has been shown to be mediated by Msx1/2 factors (Timmer et al., 2002), and the failure of this repression in most of our insertions may reflect a loss of an Msx-responsive element in our enhancers, which was possibly rescued by a local element in the Tg(−3.5dbx1a:gfp)zd3 insertion. The transient expression in spinal progenitors also suggests that either distinct regulatory elements, or prevention of silencing of elements within the 3.5kb enhancer, are necessary for maintenance of dbx1a. All of these observations reinforce the idea that while individual regulatory elements may be evolutionarily conserved, their position in surrounding chromatin may be equally important. Thus the entire dbx1a locus is worth investigating in greater detail. Future characterization of sequences upstream, between the 8kb and 3.5kb enhancers, and in conserved intronic regions, as well as enhancer elements within the 3.5kb fragment, will lead to a better understanding of how dbx1a expression is maintained in intermediate neural progenitors throughout spinal cord development. We speculate that genes such as dbx1a, which function in multipotent neural progenitors, may require a specific chromatin environment to allow proper expression, similar to genes in pluripotent stem cells (Bernstein et al., 2006).