Over the past few thousand years, the physical characteristics of sheep and their underlying genetic make-up have been slowly altered by selective breeding from animals with desirable characteristics. The result is that sheep are now mobile protein factories that can survive in harsh environments and still provide abundant quantities of milk, meat and wool. But although our ancestors succeeded in significantly altering this species, selective breeding is limited by the pool of desirable variants in the sheep population. Transgenic technology, developed over the past 15 years, has begun to address this problem. But, although this approach can be used to add new genetic information, it is not possible to specifically modify the genes of the species itself by this route. McCreath and colleagues, writing on page 1066 of this issue1, have finally overcome this problem by making use of the now familiar technique of cloning.
Transgenic technology has been widely applied to livestock species because
it is technically quite simple, requiring only the injection of 'naked'
DNA — unencumbered by its protein partners — into the nucleus
of a fertilized egg2. Genes encoding useful human proteins such
as 
1-antitrypsin (an enzyme that is mutated in the disorder familial
emphysema) have been inserted into sheep eggs in this way, resulting in transgenic
sheep that express the human protein in their milk3. But despite
the ease with which transgenic animals can be generated, this technology has
limitations, because the injected DNA integrates randomly into the sheep genome.
Often the injected DNA does not land in a site in the genome that allows the
foreign gene (transgene) to be expressed in the desired tissue or at the
appropriate level. Moreover, the sheep's endogenous genes cannot be specifically
altered using this technique.
By contrast, researchers have been able to modify endogenous genes in laboratory mice for over ten years4. This type of genetic manipulation became possible only because cells known as embryonic stem (ES) cells could be isolated from mouse embryos5. Specific mutations can be generated, or specific gene sequences altered, in cultured ES cells; rare variants with the desired genetic change can then be selected. These modified cells retain their developmental potential and, when inserted into a developing embryo, can contribute to all of its tissues, including the germ cells (egg and sperm). So the changes selected in cultured cells can be transferred into the genetic repository of the animal.
In the mouse, ES cells have provided the essential link between selection schemes in cell culture and the whole organism. The usefulness of this system for producing genetic changes has been widely exploited6. For more than a decade, researchers have searched for ES cells from a multitude of other species, including livestock. But, although some cell lines have been reported, none has been able to pass the crucial test — that of contributing to germ cells.
So the hope that livestock could be genetically manipulated like mice languished until a few years ago, when it was revived by the cloning of livestock species7. This technique, later extended to the use of nuclei obtained from cultured adult cells8, has circumvented the need to isolate the elusive ES cell from species other than mice, and is a powerful way of transferring transgenes to livestock via the culture dish9, 10. However, all the genetic changes established so far by cloning could have been achieved using conventional transgenic technology. Until now, no one had shown that it would be possible to specifically modify endogenous genes by cloning.
The success reported by McCreath et al.1 hinges in part on strategies designed to overcome some of the perceived problems of performing gene targeting in fibroblasts — the cell type of choice for cloning — rather than ES cells. For instance, they selected the COLIA1 genetic locus as the endogenous sequence to be altered, or 'targeted', because this gene is highly expressed in fibroblasts. The authors were concerned that targeting efficiencies in fibroblasts might be much lower than the 5–20% typically observed in mouse ES cells (these percentages are ratios of desired to random integration events). The high level of expression of COLIA1 allowed McCreath et al. to use a 'promoter-trap' strategy (similar to that used in early attempts to target genes in ES cells11) to increase their chances of detecting targeting events at the COLIA1 locus.
In one set of experiments, the authors inserted a neo cassette —
a stretch of DNA encoding a protein that allows cultured cells to grow in
the drug G418 — downstream of the COLIAI gene in sheep fibroblasts.
In a second set of experiments, the targeting sequence also included a transgene
downstream of the neo cassette (Fig. 1). This
transgene encoded human 1-antitrypsin, as well as a sequence to ensure
that the transgene would be switched on only in the mammary gland tissue of
lactating ewes. The targeting efficiency with both vectors was similar to
that obtained using ES cells.
Figure 1: Modifying specific genes in sheep.
McCreath et al.1 have created sheep with a human gene
inserted into a particular genetic locus. a, The targeting vector —
which carried the neomycin-resistance gene (neo) and a foreign gene
(transgene) encoding human 
1-antitrypsin — was inserted
into sheep fibroblast cells. The internal ribosome-entry site (IRES) allowed
the messenger RNA produced from the transgene to be translated into protein.
The transgene and selection cassette were incorporated between the translational
stop site and the polyadenylation (poly(A)) site of the sheep COLIAI
locus. Transfected cells expressed the neo gene and so could be grown
in the drug G418 (a neomycin analogue). b, The DNA of such cells was analysed
by the polymerase chain reaction to verify that the COLIAI locus was
targeted. c, These cells were used for nuclear transfer. Nuclei were removed
from sheep eggs, which were then fused to single transfected cells. The resulting
embryos were transferred to foster ewes. d, Some of the implanted eggs developed
into healthy lambs expressing 1-antitrypsin in their milk.
Fibroblasts cannot proliferate for long in vitro, and acquire chromosomal changes quite quickly. This meant that the authors had to achieve targeting and identify targeted clones rapidly. So they transferred the targeting vectors to the fibroblasts early in their growth history, and identified targeted cells by their ability to grow in G418 and, later, by molecular techniques (the perennially useful polymerase chain reaction).
The next step was to fuse the transfected fibroblasts with an egg from
which the nucleus had been removed, and to allow the egg to develop into
an embryo (Fig. 1) — the protocol developed for
cloning. Using this strategy, the authors established two types of targeted
loci in sheep. When the targeting sequence encoding 1-antitrypsin was
used, high levels of this protein were found in the sheep's milk. So the
COL1A1 locus might be a good site in the genome to place mammary-gland-specific
transgenes.
This technique should provide a general way to produce specific genetic changes in several mammalian species. But the opportunities extend far beyond the optimal positioning of foreign genes. Mouse models of several human diseases — such as cystic fibrosis12 — are not always precise because of differences between the two species. Perhaps the application of McCreath et al.'s technique to more closely related mammals might be of use here. In agriculture, undesirable genes could be removed, such as PrP, in this case producing flocks of sheep that are resistant to scrapie13. We are clearly at the dawn of a new era in mammalian genetic technology.