"Northern blots" or RNA blot‑hybridization
In the reverse of Southern blot‑hybridizations, one can separate RNAs by size on a denaturing agarose gel, and transfer them to nylon or other appropriate solid support. Labeled DNA can then be used to visualize the corresponding mRNA (Figure 3.33). Ed Southern initially used labeled rRNA to find the complementary regions in immobilized, digested DNA, so this "reverse" of Southern blot-hybridizations, i.e. using a labeled DNA probe to hybridize to immobilized RNA, is often referred to as "Northern" blot‑hybridizations.
One can hybridize a labeled DNA clone to a panel of RNA samples from a wide variety of tissues to determine in what tissues a particular cloned gene is expressed (top panel of Figure 3.33. More precisely, this technique reveals the tissues in which the genes is transcribed into stable RNA. The results allow one to determine the tissue specificity of expression, e.g. a gene may only be expressed in liver, or only in erythroid cells (e.g. the b-globin gene). This helps give some general idea of the possible function of the gene, since it should reflect the function of that tissue. Other genes are expressed in almost all cells or tissue types (such as GAPDH); these are referred to as housekeeping genes. They are involved in functions common to all cells, such as basic energy metabolism, cell structure, etc. The relative amounts of RNA in the different lanes can be directly compared to see, e.g., which tissues express the gene most abundantly.
One can hybridize a labeled DNA clone to a panel of RNA samples from a progressive stages of development to determine the developmental stagewhen during development a particular cloned gene is expressed as RNA (bottom panel of Figure 3.33). For instance, a gene product may be required for determination decisions early in development, and only be expressed in early embryos.
Once the DNA sequence of the gene of interest is known, and its intron-exon structure determined, highly sensitive RT-PCR assayscan be designed (Figure 3.34). The RNA from the cell or tissue of interest is copied into cDNA using reverse transcriptase and dNTPs, and then primers are annealed for PCR. Ideally, the primers are in different exons so that the product of amplifying the cDNA will be smaller than the product of amplifying the genomic DNA.
Figure 3.34.Reverse transcription-PCR (RT-PCR) assay for mRNA.
In situ hybridizations / immunochemistry
In complementary approaches, the labeled DNA can be hybridized in situto thin sections of a tissue or embryo or other specimen, and the resulting pattern of grains visualized along the specimen in the microscope (Figure 3.35). Also, antibody probes against the protein product can be used to localize it in the specimen. This gives a more detailed picture of the pattern of expression, with resolution to the particular cells that are expressing the gene. The RNA blot-hybridization techniques described in a. above look at the RNA in all the cells from a tissue, and do not provide the level of resolution to single cells.
As large numbers of sequenced mRNAs and genes become available, technology has been developed to look at expression of very large numbers of genes simulatneously. DNA sequences specific for each gene in a bacterium or yeast can be spotted in a high-density array with 400 r more spots. Some technologies use many more spots, with mutliple sequences per gene. Microarrays, or “gene chips” are available for many species, some with tens of thousands of different sequences or “probes.” RNA from different tissues can be converted to cDNA with a distinctive fluorescent label, and then hybridized to the gene chip. Differences in level of expression can be measured. Thus global changes in gene expression can now be measured.
Figure 3.36.Hybridization of RNA to high density microarrays of gene sequences, or “gene chips”.
An increasingly powerful approach is to determine candidates for the the function of your gene by searching the databaseswith the sequence, looking for matches to known proteins and genes. These matches provide clues as to protein function.
The power of this approach increases as the amount of sequences deposited in databases expand. Sequences of many genes are already known. The sequenced genes from more complex organisms, such as plants and animals, tend to be the ones more easily isolated using the techniques discussed in recombinant DNA technology. However, the sequences of genes expressed at a low level are starting to accumulate in the databases.
One remarkable advance in the past few years is the increasing number of organisms whose entire genome has been sequenced. About 10 bacterial genomes have been sequenced, and the number increases every few months. Genomics sequences for two eukaryotes are now available. That of the yeast Saccharomyces cerevisiaehas been known for a few years, and the genome of the nematode Caenorhabditis eleganswas completed in 1998. These sequences are being analyzed intensively, and a very high fraction of all the genes in each genome can be reliably detected using computational tools (one part of bioinformatics). It has become clear that many of the enzymes used in basic metabolism, regulation of the cell cycle, cellular signaling cascades, etc. are highly conserved across a broad phylogenetic spectrum. Thus it is common to find significant sequence matches in the genomes of model organisms when they are queried by the sequence of a previously unknown gene, e.g. from humans or mouse. The function already established for that gene in worms or yeast is a highly reliable guide to the function of the homologous gene in humans. The worm C. elegansis multicellular, and fate of each of its cells during development has been mapped. Thus it is possible that many functions involved in cellular interactions and cell-cell signaling will be conserved in this species, thus expanding the list of potential targets for a search in the databases.
This potential is being realized as working draft sequences of the human and mouse genomes are being analyzed. Within these data is a good approximation of sequences from virtually all human and mouse genes. Random clones have been partially sequenced from libraries of cDNAs from various human tissues, normalized to remove much of the products of abundant mRNAs and thus increasing the frequency of products of rare mRNAs. These sequences from the ends of the cDNA clones are called expressed sequence tags, or ESTs. The name is derived from the fact that since they are in cDNA libraries, they are obviously expressed at the level of mRNA, and some are used as tags in generating high-resolution maps of human chromosome. Hundreds of thousands of these have now been sequenced in collaborative efforts between pharmaceutical companies, other companies and universities. The database dbEST records all those in the public domain, and it is a strong complement to the databases recording all known sequences of genes. Many different parts of the same, or highly related, cDNAs, are recorded as separate entries in dbEST. Projects are underway to group all the sequences from the same (or highly related) gene into a a unified sequence. One example is the Unigene project at NCBI. The number of entries grows continually, but in the summer of 1998 there are about 50,000 entries, each representing about one gene. The number is higher now. Current estimates of the number of human genes are around 30,000, so it is possible that some UniGene clusters represent only parts of genes, and some genes match more than one cluster.
Very efficient search engines have been designed for handling queries to these databases, and several are freely available over the World Wide Web. One of the most popular and useful sites for this and related activities is maintained by the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/). Their Entrez browser provides integrated access to sequence, mapping and some functional information, PubMed provides access to abstracts of papers in journals in the National Library of Medicine, and the BLAST server allows rapid searches through various sequence databases. dbEST and the Unigene collection are maintained here, many genome maps are available, and three-dimensional structures of proteins and nucleic acids are available.
Make the protein product and analyze it
It is often possible to express the gene and make the encoded protein in large amounts. The protein can be purified and assayed for various enzymatic or other activities. Hypotheses for such activities may come from database searches.
The previously describe approaches give some idea about gene function, but they do not firmly establish those functions. Indeed, this is a modern problem of trying to assign a function to an isolated gene. Several “reverse genetic” approaches can now be taken to tackle this problem. The most powerful approach to determining the physiological role(s) of a gene product is to mutate the gene in an appropriate organism and search for an altered phenotype.
The easiest experiment to do, but sometimes most difficult to interpret, is a gain of function assay. In this case, one forces expression of the gene in a transgenic organism, which often already has a wild type copy of the gene. One can look for a phenotype resulting from over-expression in tissues where it is normally expressed, or ectopic expression in tissues where it is normally silent.
In some organisms, it is possible to engineer a loss of functionof the gene. The most effective method is to use homologous recombination to replace the wild type gene with one engineered to have no function. This knock-outmutation will prevent expression of the endogenous gene and one can see the effects on the whole organism. Unfortunately, the efficiency of homologous recombination is low in many organisms and cell lines, so this is not always feasible. Other methods for knocking out expression are being developed, although the mechanism for their effect (when successful) is still being studied. In some cases, one can block expression of the endogenous gene by forcing production of antisenseRNA. Another method that is effective in some, but currently not all organisms, is the use of double-stranded, interfering RNA (RNAi). Duplex RNAs less than 30 nucleotide pairs long from the gene of interest can prevent expression of genes in worms, flies, and plants. Some success in mammals was recently reported.
Another way to generate a loss-of-function phenotype is to express dominant negative alleles of the gene. These mutant alleles encode stable proteins that form an aberrant structure that prevents functioning of the endogenous protein. This usually requires some protein-protein interaction (e.g. homodimers or heterodimers).
Localization on a genetic map
Sometimes the gene you have isolated maps to a region on a chromosome with a known function. Of course, many genes are probably located in that region, so it is critical to show that a candidate gene really is the one that when mutated causes an altered phenotype. This can be done by showing that a wild type copy of the candidate gene will restore a normal phenotype to the mutant. If a marker is known to be very tightly linked to the candidate gene, one can test whether this marker is always in linkage disequilibrium with the determinant of the mutant phenotype, i.e. in a large number of crosses, the marker for the candidate gene and the mutant phenotype never separated by recombination.
The mapping is often done with gene‑specific probes for in situ hybridizations to mitotic chromosomes. One then aligns the hybridization pattern with the chromosome banding patterns to map the isolated gene. Another method is to hybridize to a panel of DNAs from hybrid cells that contain only part of the chromosomal complement of the genome of interest. This is particularly powerful with radiation hybrid panels.