How do we know that a particular DNA fragment binds a particular transcription factor? One of the simplest ways is to perform a gel mobility shift assay. The basis for this assay is gel electrophoresis. Fragments of DNA can be placed in a depression at one end of a gel and an electric current run through the gel. The fragments will move toward the positive pole, and the distance each fragment travels in a given time will depend upon its mass and conformation. Larger fragments will run more slowly than smaller fragments. If the fragments of interest are incubated in a solution of Pax6 protein before being placed in the gel, one of two things will happen. If the Pax6 does not recognize any sequence in a DNA fragment, the DNA will not be bound and the fragment will migrate through the gel as it normally would. Alternatively, if the Pax6 protein does recognize a sequence in the DNA, it will bind to it, increasing the mass of the fragment and cause it to run more slowly through the gel. Figure 1A shows the use of such a gel to locate the binding site of Pax6 protein.
The results of this procedure are often confirmed by a DNase protection assay. If a DNA-binding protein such as Pax6 finds its target sequence in a DNA fragment, it will bind to it. If the fragment is then placed in a solution of DNase I (an enzyme that cleaves DNA randomly), the bound transcription factor will protect that specific region of DNA from being cleaved. Figure 1B shows the results of one such assay for Pax6 binding.
Reporter genes such as the lacZ gene mentioned above are also used to determine the functions of various DNA fragments. If a sequence of DNA is thought to contain an enhancer, it can be fused to a reporter gene and injected into an egg by various means. If the tested sequence is able to direct the expression of the reporter gene in the appropriate tissues, it is assumed to contain an enhancer.
One of the interesting concepts emerging from this work is that the transcription factors involved in the mechanisms of cell differentiation in a given region are often used to specify that region as a certain type of tissue. Thus, the Pax6 transcription factor is used in the specification of the eye-forming region as well as in the differentiation of the retina and lens, and the Pdx transcription factor is used both in the specification of the pancreatic rudiment and in the subsequent activation of the genes for insulin and somatostatin derived from that tissue (see Hui and Perfetti 2002).
One critically important experimental use of enhancers has been the conditional elimination of gene expression in certain cell types. For example, the transcription factor Hnf4αis expressed in liver cells. It looks like a hormone-binding protein, and it might be important in the transcription of certain liver proteins. But if this gene is deleted from mouse embryos, the embryos die before they even form a liver. This is because Hnf4α is critical in forming the visceral endoderm of the yolk sac, and if this tissue fails to form properly, the animal dies very early in development. So one needs to create a mutation that will be conditional—that is, a mutation that will appear only in the liver and nowhere else. How can this be done?
The Cre-Lox technique uses homologous recombination to place two Cre-recombinase recognition sites (loxP sequences) within the gene of interest, usually flanking important exons (see Kwan 2002). Such a gene is said to be “floxed” (“loxP-flanked”). For example, using cultured mouse ES cells, Parvis and colleagues (2002) placed two loxP sequences around the second exon of the mouse Hnf4α gene (Figure 2). These ES cells were then u¬¬¬sed to generate mice that had this floxed allele. A second strain of mice was generated that had a gene encoding bacteriophage Cre-recombinase (the enzyme that recognizes the loxP sequence) attached to the promoter of an albumin gene that is expressed very early in liver development. Thus, during mouse development, Cre-recombinase would be made only in the liver cells. When the two strains of mice were crossed, some o¬¬f their offspring carried both additions. In these double-marked mice, Cre-recombinase (made only in the liver cells) bound to its recognition sites—the loxP sequences—flanking the second exon of the Hnf4α genes. It then acted as a recombinase and deleted this second exon. The resulting DNA would encode a nonfunctional protein, since the second exon has a critical function in Hnf4α. Thus, the Hnf4α gene was “knocked out” only in liver cells.
The ability of an enhancer from one gene to activate other genes has been used by scientists to find new enhancers and the genes regulated by them. To do this, one makes an enhancer trap, consisting of a reporter gene (such as the E. coli lacZ gene or jellyfish GFP gene) fused to a relatively weak promoter. The weak promoter will not initiate the transcription of the reporter gene without the help of an enhancer. This recombinant enhancer trap is then introduced into an egg or oocyte, where it integrates randomly into the genome. If the reporter gene is ex¬pressed, it means that the reporter has come within the domain of an active enhancer (Figure 3). By isolating this activated region of the genome in wild-type flies or mice, the normal gene activated by the enhancer can be discovered (O’Kane and Gehring 1987).
One of the most powerful uses of this genetic technology has been to activate regulatory genes such as Pax6 in new places. Using Drosophila embryos, Halder and his colleagues (1995) placed a gene encoding the yeast GAL4 transcriptional activator protein downstream from an enhancer that was known to function in the labial imaginal discs (those parts of the Drosophila larva that become the adult mouth parts). In other words, the gene for the GAL4 transcription factor was placed next to an enhancer for genes normally expressed in the developing jaw. Therefore, GAL4 should be expressed in jaw tissue. Halder and his colleagues then constructed a second transgenic fly, placing the cDNA for the Drosophila Pax6 regulatory gene downstream from a sequence composed of five GAL4-binding sites. The GAL4 protein should be made only in a particular group of cells destined to become the jaw, and when that protein was made, it should cause the transcription of Pax6 in those particular cells (Figure 4A). In flies in which the Pax6 gene was expressed in the incipient jaw cells, part of the jaw gave rise to eyes (Figure 4B). Pax6 in Drosophila and frogs (but not in mice) is able to turn several developing tissue types into eyes (Chou et al. 1999). It appears that in Drosophila, Pax6 not only activates those genes that are necessary for the construction of eyes, but also represses those genes that are used to construct other organs.