Since the pioneering work of Stuart et al. (9,15), several groups have reported the generation of stable transgenic lines of zebrafish. Although, the frequency with which transgenic fish were produced was low, some important features have emerged about generating stable transgenic lines.
The preferred method of introduction of the DNA sequences is injection (9,10,13,15-17). An alternative is by electroporation (18-21), but these studies have not demonstrated that the transgenic lines are genetically stable or that the transgenes are expressed. Therefore, microinjection of the DNA into the one-cell stage blastula is the most reliable method.
The amount of DNA injected is difficult to estimate and can vary with each needle used. In general, solutions ranging from 25-50 |g/mL give the highest number of survivors, which may also be transgenic in their germline (9,15,16). Phenol red may be added to the solution at 0.1-2.0% as a visual guide to the amount injected. Alternatively, the volume may be estimated by using a radioactive nucleotide in the injection mix and counting the embryo in a scintillation counter. The volume injected should be about 1-3 nL (14).
It appears to make little difference to transient transgene expression or the frequency of generating transgenic germlines whether injected DNA is linear or circular DNA. In all cases reported, all the founder fish (F0), which transmit the transgene to their progeny, were found to be mosaic. In most cases where the transgenic lines have been subjected to Southern analysis, the transgene is inserted at a single chromosomal location, but in multiple copies probably in a tandem array. However, the single enhancer trap insert reported by Bayer and Campos-Ortega (10) appears to be a single copy, and 12 out of the 15 integrations reported by Gibbs et al. (13) were of one to two copies. The reason for this lower copy number in the latter study is not clear. The DNA was injected as supercoils at 10 |g/mL in 0.01X TE containing 0.05% phenol red. They estimated injecting a total of 100 pg of DNA/embryo and obtained a survival of 10-20% to sexual maturity.
A problem with transgenic technology in zebrafish is the high level of mosaicism in the germlines of the founders. This means that only a small proportion of the F1 will inherit the transgene (frequencies reported range from 240%). This makes the identification and live isolation of the relatively rare transgenic F1s very time-consuming if they are screened by PCR. Founder fish with mosaic germ-lines can be identified by extracting DNA from a proportion of embryos from each individual spawning. If the frequency of transgenic F1 fish is low, there is chance that transgenic fish may be missed or that the siblings that are raised to sexual maturity do not carry the transgene. The alterna-
tive is to raise all F1 fish until they are near maturity and then take a fin biopsy. This means that many fish must be reared before they can be tested. These problems can be overcome if transgenic fish carry a dominant marker to identify individual transgenic embryos. One possibility is to include a lacZ reporter gene in the injected DNA and screen the F1 progeny with a substrate that can be used on living cells. The use of two such substrates has been reported. Westerfield et al. (12) used the Imagene Green™ substrate to identify transient expression of injected hox-lacZ constructs. Lin et al. (22) used a similar fluoresceinated substrate (FDG) on transgenic embryos expressing a lacZ fusion construct with the Xenopus elongation factor 1a (EF1a) transcriptional regulatory element. In the latter study, four out of five different transgenic lines expressed the lacZ gene in early embryos. However, the pattern of expression was distinct for each line, with two lines first beginning expression at the midblastula transition. One of these expressed solely in motor neurons, whereas the other showed patchy expression. Since the different lines contain the same transgene construct, the different expression patterns must be a consequence of the site of insertion in the genome. Positive embryos of the most highly expressing transgenic lines were identified by permeabilizing with dim-ethylsulfoxide and soaking the embryos for 2-3 min in the FDG substrate.
The lacZ gene as the reporter gene has the disadvantage that weakly expressing embryos may not be identified. Prolonged incubation in the substrate gives rise to false positives, probably through the activity of endogenous P-gal. This problem may be overcome with transgene constructs with luciferase, which is not normally found in zebrafish. Gibbs et al. (17) injected CMV-luciferase plasmids into embryos, and showed transient expression by incubating the embryos in luciferin solution and detecting the luminescence on X-ray film. However, the transgenic lines in this study failed to express the luciferase enzyme.
The most promising alternative to the P-gal substrates and to luciferase described above is the green fluorescent protein (GFP) from Aequorea victoria (23). GFP produces fluorescence without the need to add an exogenous substrate. Amsterdam et al. (24,25) have fused a truncated version of the 4.6 kb Xenopus elongation 1a enhancer/promoter to the GFP cDNA. Expression of such a construct starts at about 4 h after injection at the one-cell stage when incubated at 28°C. This corresponds to onset of transcription at the midblastula transition (26). Fluorescence continues to be detected for at least 3 wk after injection.
A further three different constructs were made, which all had the EF1a enhancer/promoter in different fusion configurations with GFP cDNA (25). One, with an intron from rabbit P-globin, yielded five transgenic lines, each of which expressed the fluorescence in the F1 fish and, where tested, in the F2. Most expressing lines showed a similar pattern of expression. In transgenic embryos, fluorescence was not seen until 20 h after fertilization, when it was seen uniformly expressed throughout the embryo. Fluorescence was greatest between 24 and 36 h, but persisted for up to 5 d being particularly strong in the eye. A single copy of the transgene was sufficient to detect fluorescence.
The absence of expression in the F1 and subsequent generations is a serious problem with zebrafish transgenic technology. The study of Gibbs et al. (17) demonstrated that the CMV-luciferase transgenes were highly methylated in the F1 and subsequent generations, which could account for the lack of activity. Partial activity could be recovered by incubating in 0.3 mM 5-azacytidine, which can inhibit methylation in vivo. However, survival was very much reduced, and many of the embryos were deformed. It is possible that incorporation of extra sequences into the vectors to "buffer" the promoter-gene fusions may allow subsequent expression of the transgenes after passage through the germline. The incorporation of the intron from rabbit P-globin in the GFP construct described by Amsterdam et al. (25) appears to be beneficial with respect to the ability of the transgenes to be transcribed. However, they report that when included with additional insulating sequences flanking the same construct, only one out of three transgenic F1 lines expressed the GFP. The number of transgenic lines are low, so these results should be treated with caution. However, they suggest that transcription of the transgenes may depend on some inherent property of the particular DNA sequences.
Caldovic and Hackett (27), have tested the ability of border elements incorporated into transgenic constructs to affect integration and expression of a carp P-actin gene enhancer/promoter-CAT reporter gene fusion. Border elements are sequences of DNA that are part of the scaffold attachment regions involved in forming the nuclear scaffold and can impart position-independent expression of a gene in transgenic mice (28). Caldovic and Hackett (27) made transgenic constructs with border elements from the Drosophila heat-shock 87A7 locus and the attachment-element sequence from the chicken lysozyme locus. These constructs were problematical to make, since plasmids containing two copies flanking the transgene construct failed to amplify in Escherichia coli. To overcome this problem they generated linear concatamers by ligation of separate purified fragments in vitro. Transgenic fish carrying the border elements showed uniform expression at about the same level in all tissues. Without the border elements, the expression was restricted and variable between different transgenic lines.
The recent work of Higashijima et al. (29) has shown that transgenic zebrafish, which express GFP reliably, can be generated at high frequencies. Unlike the previous studies, which used promoter constructs of heterologous origins, they have made constructs in which the expression of GFP is driven by a zebrafish muscle-specific actin promoter (a-actin) promoter. Transgenic zebrafish carrying the a-actin-GFP promoter constructs reliably expressed the protein muscle. Further constructs with a cytoskeletal P-actin promoter produced transgenic zebrafish that expressed GFP throughout the body.
The use of GFP as the reporter gene allows embryos, injected at the one-cell stage, to be examined for transient expression of the protein a day later. Only those embryos expressing the GFP need to be raised to adulthood. These founder fish are crossed with wildtype fish and the F1 progeny examined for expression of GFP. In the case of a-actin-GFP constructs, 41/194 (21%) of the founder fish transmitted the GFP expression to their progeny. Of the 41 lines generated, 40 expressed the GFP in the same spatially restricted pattern as the normal a-actin gene. It also appears that the a-actin-GFP construct can be incorporated into a plasmid that also carries other transgenes driven by other promoters. Thus the a-actin-GFP transgene can be used as a transformation marker and will be invaluable in generating transgenic zebrafish carrying different gene constructs. Important points arising from this work are that the DNA to be injected should be linearized and the constructs made such that any plasmid sequence is 3' to the a-actin-GFP transgene and that those embryos expressing GFP in only a few cells, a day after injection, should be rejected (usually about 25%). In a similar study by Lin's group (30) tissue-specific expression of GFP has been achieved using a modified promoter of the zebrafish GATA-1 gene.
Was this article helpful?