The results described here confirm that Tol2 integrated Gal4-VP16 transgenes not only self report via an UAS:eGFP
cassette, but can function as driver lines to activate UAS responder lines in trans
. The pilot screen has efficiently generated lines displaying restricted eGFP expression in a wide variety of tissues. The SAGVG vector is designed to take advantage of the amplification inherent in the Gal4-VP16/UAS
system (Koster and Fraser, 2001
) and is facilitated by recent advances in transgenic methods (Kawakami et al., 2000
). With Gal4-VP16 activation, detectable eGFP expression is expected even if transcription levels from endogenous regulatory elements are low. Ideally, this maximizes the number of driver lines that can be obtained in a screen.
Most Gal4-VP16 driver lines maintained robust expression of eGFP over 3 generations. In some cases, such as c218, we have observed a decrease in the frequency of embryos/larvae expressing eGFP, in each new generation. This occurrence could be due to transgene silencing, but as SAGVG trapping events are self reporting, green fluorescence can be used to easily identify the progeny that have escaped this phenomenon. Alternatively, as each line contains several copies of SAGVG distributed throughout the genome, random assortment could be reducing the copy number of UAS:eGFP insertions available for Gal4 activation, thereby affecting the amount of eGFP expressed. For some lines, special efforts may be needed to maintain adequate templates for transactivation.
To identify candidate genes associated with particular expression patterns, and to gain insight into the mechanisms underlying Gal4-VP16 expression, we determined the chromosomal localization of insertions responsible for eGFP expression patterns of interest. We examined 6 lines and found a total of 8 insertions co-segregating with eGFP expression (c223 had three linked insertions). Genuine gene trap events are potentially mutagenic (Friedrich and Soriano, 1991
). Although the SAGVG construct was designed to function as a gene-trap, we found that only three insertions (c218, c220, c223-C) were consistent with a gene trap mechanism (depicted in ), on the basis of the identity and orientation of flanking genomic sequences.
Potential mechanisms of eGFP expression
The c220 insertion is in the first intron of a known gene and does indeed create a mutant fusion transcript, as demonstrated by sequencing the RT-PCR product. When bred to homozygosity, c220 fish display abnormal morphology and late lethality (supplemental Fig.S2
mutations in humans lead to a range of defects that include ectodermal dysplasia and immunodeficiency (Smahi et al., 2000
). The Ikkγ
gene is X-linked in mammals, and male mice hemizygous for a null Ikkγ
mutation die during mid-gestation. Although heterozygous females are viable, likely due to random X inactivation of the one wild type copy of the gene they display ectodermal dysplasia (Schmidt-Supprian et al., 2000
). It is clear that SAGVG can function as a bona fide gene trap vector with the potential to be mutagenic, and work is underway to demonstrate whether the linked late phenotype seen in c220 fish is solely due to the disruption in zebrafish ikkγ
Two other lines (c229 and c237) appear to represent Gal4-VP16 dependent enhancer traps (mechanism depicted in ), because the SAGVG insertions are located outside of known transcribed regions. Activation of eGFP expression is most likely mediated by Gal4-VP16, which in these lines is capable of activating other UAS reporters in trans
(see below). Specific enhancer trap studies in the zebrafish have shown that active insertions are often in close proximity to genes with similar patterns of expression (Balciunas et al., 2004
; Ellingsen et al., 2005
). In the case of c229 and c237, the SAGVG construct has inserted within 20kb of genes with tissue-specific expression predicted to be similar to the observed eGFP labeling patterns. This suggests that nearby endogenous genes and Gal4-VP16 are under the control of a common regulatory enhancer(s). In these two cases, the site of transcription initiation is unknown: it may be that there are cryptic promoter sequences in upstream genomic DNA or within the vector itself.
Lines c228 and c236 are most likely examples of Gal4-VP16 independent eGFP expression (mechanism depicted in ). Both insertions generate highly restricted eGFP expression, in hindbrain rhombomeres (c228) or in the heart (c236), respectively. However, when fish bearing either insertion were bred to Tg(UAS:Kaede) fish, the resultant eGFP+ larvae failed to show any evidence of Kaede photoconversion (see below). This finding is consistent with an SAGVG insertion in which transcription from the minimal promoter of eGFP is under the direct influence of a local enhancer () rather than mediated via Gal4-VP16 activation.
Although a somewhat unexpected finding, the multiple mechanisms by which eGFP expression can be achieved from the SAGVG construct likely accounts for the very high frequency with which distinct expression patterns were observed during the screening phase of this study. Fifty-seven percent (28/49) paired crosses of F0 fish resulted in restricted patterns of eGFP expression. We also demonstrated in this report the utility of the SAGVG insertions to function as Gal4-VP16 driver lines by transactivating 3 different UAS-regulated genes to produce mCherry, Kaede and NTR-mCherry. Of the 15 lines assayed for transactivation, only 2 (c228 and c236) failed to show labeling of other fluorescent proteins within eGFP-expressing tissues. Significant resources will be required to maintain a large panel of transactivating fish lines with restricted patterns of eGFP expression generated from a more extensive screen for Gal4-Vp16 insertions. Hence, it is satisfying that Tol2 insertional events failing to lead to transactivation are uncommon.
Spatially restricted photoconversion of Kaede can be used to label individual cells, such as neurons (e.g., ) and to trace their axonal projections (Sato et al., 2006
). This system is also well suited to cell tracking (Hatta et al., 2006
). Using this technique we tracked an unknown population of photoconverted cells over time and demonstrated that they migrated rostrally and came to lie in close proximity to the liver. Cell position at 8 dpf strongly correlated with the tissue location of eGFP positive cells in the mesothelium of the adult liver. Thus, transactivation of Kaede by SAGVG transgenic lines and subsequent photoconversion is a powerful method to monitor the behavior of cells and their ultimate tissue specific fate.
In several experiments where Tg(UAS:Kaede)
fish were bred to Gal4-VP16 lines, we noted a mosaic pattern of transactivation. For instance larvae carrying the c237 allele possess interneurons that fluoresce green. In some interneurons photoconversion of Kaede was not observed (), even though adjacent cells with similar morphology converted to red fluorescence following exposure to UV light. This presumed variegation in transactivation may represent differences in chromatin state between otherwise similar cells. Such epigenetic effects are known to cause variegation in transcription levels and to be influenced by Gal4 activity (Ahmad and Henikoff, 2001
). Tissue variability in expression has also been previously documented in Drosophila
using the Gal4-VP16/UAS system (Fischer et al., 1988
) and may also influence transactivation in zebrafish. Such potentially complicating phenomena must be taken into account when interpreting experiments where Gal4-VP16 drivers are used to misexpress genes in given cell types and may necessitate fluorescently tagging activated proteins of interest.
Two Gal4-VP16 lines were bred to fish transgenic for UAS:nfsB-mCherry
. The production of the NTR-mCherry fusion protein led to prodrug dependent cell ablation. Self-reporting Gal4-VP16 drivers can be used for selective cell ablation in a tissue-specific manner and for monitoring cell loss over time. From our work and the work of others it is clear that NTR activity can be used to ablate a wide range of cell types throughout the embryo/larva (Pisharath, 2007
) and D. Stainier (personal communication). As more Gal4-VP16 driver lines become available, the UAS:nfsB-mCherry
transgenic line will constitute an important resource in studying cellular regeneration and cell-cell interactions.
Gal4-VP16 technology revolutionized experimental design in Drosophila
(Brand and Perrimon, 1993
; Duffy, 2002
) and we envision the same advantages can be readily applied to zebrafish. Future applications of in the zebrafish could include the rescue of mutations by expression of wild-type genes in specific tissues (Zars et al., 2000a
; Zars et al., 2000b
), the generation of cancer models by expression of oncogenes in tissues of interest (Folberg-Blum et al., 2002
), dissection of signal transduction pathways in specific cells by expression of dominant negative (Elefant and Palter, 1999
) or constitutively active pathway components, and probing the neuronal basis of behavior (Brand and Dormand, 1995
). The binary nature of the Gal4-VP16 activation system will permit the production of many genetic tools for zebrafish that cannot be generated as simple transgenic lines due to lethality.
In conclusion, the current results demonstrate that distribution of SAGVG throughout the zebrafish genome by Tol2 transposition can generate a collection of effective Gal4-VP16 driver lines. Self-reporting transgenic lines are useful for labeling cell types, and for transactivating other UAS regulated genes. Mapping has shown that there are apparently multiple mechanisms by which transcription of Gal-VP16 can be induced and may explain the efficient recovery of eGFP-expressing lines. We anticipate that the widespread application of this technology will provide an important new resource for genetic manipulation in zebrafish.