Targeted gene ^knock-down’ in zebrafish.

The classical method o f genetic research involves the identification o f a mutant phenotype, which is studied to reveal the fimctional significance o f the mutated gene. Mapping and cloning can then identify the gene itself. The direction o f progress is from phenotype to the gene identification. Reverse genetics takes advantage of sequencing technology, which provides us with primary genetic information. This sequence data can then be used to target genes within the genome for inactivation.

allowing us to create a specific mutant lacking activity o f a gene o f interest. In reverse genetics the progress is from gene sequence to mutant phenotype.

The ability to create an organism in which a specific gene product is absent or non­ functional provides an invaluable tool to investigate the role o f the targeted gene. Such techniques have been available in a number of model species for some time. Homologous recombination within cultured mouse embryonic stem cells (ES cells) has allowed the creation of transgenic mice in which activity o f a specific gene is removed. A DNA construct is designed with flanking regions homologous to the targeted gene, with the central region containing an antibiotic resistance gene and sometimes also a reporter gene. ES cells that have integrated the ‘knock-out’ construct correctly in their DNA are selected for (by antibiotic resistance) and are then injected into the blastula at the 32 cell stage or co-cultured with treated 8 cell stage embryos (reviewed in Galli-Taliadoros et al., 1995). The embryos then require re-implantation in a pseudo-pregnant female for development to proceed. The resulting embryos are chimeras. Progeny that contain the knock-out construct within cells in the germ line are selected for and are crossed together to make stable mutant lines.

A similar system generating zebrafish chimeras using a contribution o f cells from culture has been developed (Ma et al., 2001). However, this system is not yet useful as a research tool because homologous recombination in the cell culture is not possible and the success rate of chimera development is very low.

The nematode C. elegans was the first multicellular organism to have its genome sequenced in its entirety (C. elegans sequencing consortium, 1998) and so lends itself to reverse genetics. RNA interference (RNAi) involves the injection of antisense, sense or double stranded RNA, which have all been successfully used to ‘silence’ translation of specific genes in C. elegans (Fire et al., 1998) by binding the specific mRNA transcripts of the targeted gene.

It is thought that the use o f RNAi has been such a success in C. elegans as this organism uses production o f antisense RNA transcripts to bind mRNA as an

endogenous method of transcriptional gene silencing. This phenomenon is called co­ suppression, and has also been observed in various plant species, fungi, and in Drosophila, all of which show a response to RNAi (Ketting & Plasterk, 2000).

A number o f approaches for targeted blocking of translation have been attempted in zebrafish. One such technique involves injecting DNA encoding a ribozyme (a biologically active RNA molecule, similar to a protein enzyme) designed to specifically cleave a sequence found on the mRNA to be targeted. The no tail mRNA was successfully destroyed in vivo by this method (Xie et al., 1997).

The no tail gene has also been targeted using double stranded RNA injection (Li et al., 2000). However, another study suggests that RNAi is not an ideal method to block gene function in zebrafish as it generates many non-specific phenotypes. When the T-box gene tbx 16/spade tail was targeted, one o f the non-specific effects observed was a reduction in the presence o f a variety o f mRNA transcripts including no tail mRNA (Oates et al., 2000). A new method to block the translation o f specific mRNAs has become available to zebrafish researchers recently.

Morpholino phosphorodiamidate oligonucleotides (Morpholinos, MOs) are synthetic DNA analogues. MOs have a morpholine ring (instead o f a ribose sugar) and a neutrally charged backbone (instead of the negatively charged backbone o f DNA). They were originally developed as therapeutic agents, and block specific translation by a non-classical RNAse H-independent mechanism (reviewed by Summerton, 1999; Heasman, 2002).

Morpholinos are designed to bind the 5’ region o f the mRNA transcript, to either the leader sequence or the translation start site closely down stream of this. This prevents scanning o f the mRNA by the 40S ribosomal subunit so that the translation complex is unable to form. This blocking of the leader sequence and surrounding bases is a method employed endogenously to regulate translation. The 5’ UTR from the RNA prior to post-transcriptional processing is able to form a secondary structure that works identically to the MO (Ekker & Larson, 2001).

The use of morpholinos in zebrafish has been prolific over the past year. All known major pathways in the developing zebrafish have now been targeted successfully using MOs. The Wnt pathway has been disrupted using MOs targeted to frizzled-2, (Sumanas et al., 2001), silberbick/wntlla andpipetail/wnt5a mRNA transcripts (Lele et al., 2001). The Hedgehog pathway has been investigated with a morpholino approach by a number of groups. Hedgehog pathway mutants are generally weaker than those observed in other species, probably as a result o f functional redundancy between at least two hedgehog genes, sonic hedgehog {shh) and tiggie winkle hedgehog {twhh) (Nasevicius & Ekker, 2000). A third hedgehog gene is also present in zebrafish. MOs allow a fast and effective method o f generating embryos that lack function in more than one gene, as they can be co-injected in a single embryo. Both shh and twhh were ‘knocked-down’ to achieve a zebrafish model o f the human condition holoprosencephaly (Nasevicius & Ekker, 2000), which is due to mutations in the shh gene (Roessler et al., 1996). The ‘double-morphanf revealed a novel phenotype in the forebrain (partial cyclopia, as observed in the human condition), which is not seen in either single mutant or morphant line.

An identical approach was used to investigate the role of the hedgehog pathway in brachiomotor neurons (Bingham et al., 2001) and floor plate development (Etheridge et al., 2001).

The nodal pathway has been studied using morpholinos to phenocopy the squint, one­ eyed-pin-head (oep) (Feldman & Stemple, 2001) and cyclops (Karlen & Rebagliati, 2001) mutant lines. The MO targeted to oep highlights a loss of activity o f the morpholino by the second day of development. The late differentiating floor plate, requiring Oep, develops normally in morphants while is absent in mutants. This is a limitation o f the morpholino approach, although they appear adequate to block specific protein translation for at least the first 36 hours o f development - a considerable length of time in the context of zebrafish development.

Mutants from the BMP pathway including sw irllbm plb, snailhouselbmp? and somitabun/smad5 have also been successfully phenocopied using morpholinos (Imai & Talbot, 2001 ; Lele et al., 2001).

O f particular interest to this work is the ‘knock-down’ o f ptx genes, although both such studies have been carried out inXenopus (Chang et ah, 2001; Schweickert et ah, 2001). Neither study focused on the anterior pituitary, but instead on the cement gland, an ectoderm derived structure at the anterior extreme o f the developing embryo. Ectopic cement glands were first generated by the injection o f otx mRNA, and then development o f these ectopic glands were shown to be reduced by the co­ injection o f either X p tx l MO or Xptx2c MO, and completely prevented by the co­ injection of both morpholinos (Schwiekert et al., 2001).