Promise of Gene Therapy
Studies in the first half of the twentieth century demonstrated that bacteria were able to exchange genetic material, resulting in permanent and heritable changes in the properties of the recipient strain. The subsequent understanding of the molecular basis of inheritance in the second half of the twentieth century naturally led to the concept of using deoxyribonucleic acid (DNA) fragments comprising whole genes to overcome the effects of genetic defects in various human diseases. The experiences with the transfer of genes between bacteria (transformation) encouraged the development of a fairly simplistic view of gene therapy for human diseases: a single introduction of a normal copy of a defective gene into affected cells should result in long-term, stable production of the missing protein, leading to a complete cure! The promise for 100% cure seemed to be very different from anything else available in the armamentarium of twentieth-century medicine and caught the imagination of many scientists, as well as the attention of the news media and of the investors. Rather than treat disease by repeated administration of pharmacological agents, gene therapy offered the prospect of a complete cure after a single application and with no side effects! This concept has formed the premise for much of the gene therapy research that has been conducted over the last 30 years, and underlies much of the hype that has been associated with it.
Gene therapy was originally envisaged as a method for delivering normal copies of genes for the treatment of patients with various rare genetic diseases. This concept has gradually changed to include the use of gene fragments, including polymerase chain reaction (PCR) fragments as well as chimaeric and antisense oligonucleotides. Furthermore, the increasing understanding of the molecular mechanisms underlying cancer development and tissue repair with embryonic and adult stem cells is opening day by day many more applications for gene therapy. This is indicated by the fact that gene therapy trials for cancer now comprise about 70% of all gene therapy trials. These developments have further excited interest in the potential of gene therapy to transform the quality of life of most human beings in the twenty-first century.
On the other hand, it did not take very long for gene therapy research to come up against some serious obstacles. By the early 1980s, it was clear that most human genes were much more complex in organization than anything encountered in bacteria. While bacterial genes are composed of continuous coding sequences and adjacent regulatory elements, the coding sequences of human genes (exons) were found interspersed with noncoding intervening sequences (introns), ranging in size from a few base pairs to hundreds of thousands of base pairs. By what wondrous mechanisms could coding sequences placed apart over such distances be brought accurately together, again and again, to make continuous coding sequences in the mature messenger ribonucleic acids (mRNAs), so as to ensure the faithful production of the thousands of proteins needed by each cell? Studies on the regulation of the expression of the -globin locus have also revealed not only the presence of regulatory elements in the introns and the untranslated sequences immediately adjacent to each globin gene but also regulatory elements conferring tissue and developmental specificity as far as 50 kb (50000 base pairs) away from the target genes.
The early concept of gene therapy could thus not be applied directly to the treatment of human diseases, as there are still no methods for packaging and efficiently delivering DNA fragments in the 100300 kb range, which could encompass most human genes with their natural regulatory sequences. The discovery of the technique to reverse transcribe mRNA back into complementary DNA (cDNA) seemed to offer a way out of the difficulty. Minigenes produced in this way could be readily cloned and manipulated in bacteria before delivery into human cells. Furthermore, such minigenes were easily accommodated in various viral vectors, thus facilitating efficient delivery into different cell types. However, such synthetic minigenes carried few if any of their natural regulatory elements and, not surprisingly, failed in most cases to approach normal levels of expression under physiologically relevant conditions. With the exclusion from the viral vectors of most of the essential regulatory elements that underlie the developmental, tissue and locus specificity of gene expression, the goal of gene therapy proved much more elusive than was ever anticipated.
The task of identifying one by one individual regulatory elements followed by the inclusion and testing of combinations of these elements in the context of synthetic minigenes in viral vectors has proved very slow work. Thus, although the sequence of the intact -globin locus has been known for about 15 years and numerous elegant studies have examined the regulation of the expression of the different globin genes, we still do not have a detailed picture of globin gene expression under various physiologically relevant conditions, while it has proved very difficult to achieve a therapeutic level of globin expression with any viral constructs. Our understanding of the mechanisms underlying the developmental, tissue and locus specificity of expression of all the other human genes remains of course at an even lower level. Although the completion of the sequencing phase of the Human Genome Project has now revealed the sequences of the regulatory elements for most genes, it will probably take research during most of the twenty-first century before a good understanding of the regulatory mechanisms for most human genes is achieved. The development of effective viral gene therapy vectors is of course conditional upon such an understanding for each gene of interest. In contrast, the delivery of large DNA fragments carrying intact functional loci and/or the direct correction of mutations in genomic DNA are not conditional on such detailed knowledge and should enable the faster exploitation of knowledge of the human genome for the effective therapy of genetic and somatic gene diseases.
The development of several high-quality, large insert genomic bacterial artificial chromosome (PAC/BAC) libraries for the Human Genome Sequencing Project is for the first time opening the real possibility of using intact human genes for the treatment of gene diseases. These resources have already proved extremely popular for gene mapping, isolation and sequencing. Similarly, the use of fully sequenced genomic fragments from PAC/BAC libraries for functional studies is already shifting the attention of many researchers to packaging, delivery and long-term maintenance of intact human genes in human cells as independent minichromosomes or episomes, or by targeted integration into specific sites in the genome using homologous or site-specific recombination mechanisms.
At the same time, it is also likely that new methods will be developed for the targeted correction of mutations by enhancing the endogenous mismatch repair and/or homologous recombination mechanisms of the cells. The high fidelity of these mechanisms should reduce the risks associated with random integration of viral vectors in the human genome, while enabling the corrected gene to function under the control of the endogenous regulatory mechanisms.
Finally, it should be noted that no drugs have been developed so far to treat human diseases by modifying the expression of specific genes so as to complement other defective genes, or by modifying the way specific mutations interfere with gene expression. However, the completion of the Human Genome Sequencing Project in combination with DNA chip expression profiling is expected to lead to an unprecedented degree of understanding of genetic networks under different physiological states, thus facilitating the identification of genes playing key regulatory roles in health and disease. The coupling of this knowledge with technologies for high-throughput screening for agents acting on the mechanisms that underlie the developmental, tissue and locus specificity of gene expression should facilitate the development of drugs that modify the expression of specific genes in a tissue-specific manner, or that are able to suppress or reverse the effects of specific types of mutations. Although such approaches may not involve delivery of any DNA fragments and would not be strictly speaking in the realm of gene therapy, they could provide a basis for the effective and safe management of most genetic and somatic gene diseases.
Ex Vivo Gene Therapy with Retroviruses
Viruses have naturally evolved efficient mechanisms to deliver their nucleic acids into eukaryotic cells. Typically, a few particles per cell are sufficient to lead to productive infection for most viruses. Many different viruses are now known, ranging in the type of nucleic acid they use (RNA or DNA), the size of their genomes and their species and host cell specificity. Many of these viruses, including murine retroviruses (MMLV, Moloney murine leukaemia virus), lentiviruses, adenovirus, adeno-associated virus (AAV), Vaccinia virus and Herpes simplex type 1 virus (HSV1), have been intensively studied as potential gene delivery systems.
Murine retroviruses were the first to be studied as gene delivery vectors and still remain very popular because of their high transduction efficiency, their ability to infect rapidly dividing cells and their efficient integration into genomic DNA. The maximum size of the therapeutic insert with murine retroviruses is about 8 kb, thus precluding the use of most genes with their natural regulatory elements. This is also true for most of the other viral vectors. Lentiviral vectors, despite the obvious safety concerns, are also being used increasingly in gene therapy research, because of their ability to integrate therapeutic genes into nondividing or quiescent cells, a property that may be particularly useful for therapeutic gene delivery into haematopoietic stem cells and neuronal cells.
In retroviral vectors, the therapeutic gene is inserted between the long terminal repeats (LTRs), in place of the viral genes that are needed for replication and packaging. The packaging of the recombinant retroviral vector is carried out in a packaging cell line that provides all the viral proteins that are required for the formation of mature viral particles. A major concern in this approach has been to minimize the chances of generating replication-competent retroviruses by recombination between the therapeutic vector and the helper sequences in the packaging cell line. The latest packaging systems have been designed to make the production of wild-type virus through homologous recombination highly unlikely. However, cells have a remarkable ability to join together DNA molecules through nonhomologous recombination mechanisms, and thus the generation of modified replication-competent retroviruses remains a small but not negligible possibility.
Retroviral vectors are best suited for ex vivo gene delivery to cultured cells (e.g. cultured haematopoietic stem cells). After transduction, the recombinant retroviral construct is efficiently integrated into the chromosomes of the recipient cells. However, the random nature of the integration events could lead to a number of undesirable complications:
- The expression of the therapeutic gene may be affected by the overall organization of the chromatin in the region of integration, or by nearby regulatory sequences, leading to variability of expression and/or silencing.
- One or more integration events may take place near a gene involved directly or indirectly in the cellular processes leading to cell division. Modification of the expression of such genes by the LTR elements or other regulatory sequences in the recombinant vector may lead to tumorigenesis. Since there have not been systematic long-term studies to assess the magnitude of this risk, this needs to be kept constantly in mind as gene therapy reaches the stage of clinical effectiveness with these vectors.
- The ability of retroviral and lentiviral vectors to achieve efficient integration may be linked to equally efficient mechanisms for replicational excision and reintegration at secondary sites in the presence of complementing functions, thus enhancing considerably the risk of complications.
- An integration event may take place near one or more of the numerous retrovirus-like elements that are scattered in the human genome, with the potential to activate a dormant element into a new type of transmissible virus.
- Another cause for concern is the potential of interaction between an ongoing viral infection and a recombinant viral vector, to produce a virus with a new potential. This is of particular concern in gene therapy trials on human immunodeficiency virus (HIV)-infected patients, where the high rate of virus production and the high rate of spontaneous mutagenesis are already challenging all the tools of modern medicine.
Ex vivo gene therapy allows in principle the possibility of careful evaluation of the effectiveness and safety of the procedure before returning the cells to the patient, although other considerations may necessitate the return of the cells to the patient before a thorough evaluation can be carried out. It is hoped that advances in stem cell research will enable long-term maintenance of human stem cells in culture after gene delivery, to allow a thorough evaluation to be carried out. Similarly, the possibility to direct dedifferentiation and redifferentiation of various types of stem cells is opening new avenues for the combination of ex vivo cell and gene therapy approaches.
In vivo Gene Therapy
The potential for ex vivo gene therapy is currently limited to haematopoietic stem cells. This approach is not applicable to most other cell types, either because the tissue cannot be cultured or because the tissue forms too large a part of the body to be effectively repopulated after gene therapy of stem cells in culture. Besides these practical limitations to the application of ex vivo gene therapy for genetic diseases, in vivo gene therapy is a necessity for cancer and other somatic gene diseases.
Besides the difficulties encountered with ex vivo gene therapy as discussed above, in vivo gene therapy also faces the additional challenge of delivering enough therapeutic vector to the target tissue without toxicity to other tissues. While in some situations the correction of a defect in a proportion of the cells in a tissue may be sufficient to restore health, in the case of cancer most, if not all, cells have to receive the therapeutic gene. Thus, a high degree of specificity during in vivo gene delivery is essential. This may be achieved by one of the following methods.
- Altering the glycoprotein coat of the recombinant vector so that it can only get into cells carrying the appropriate receptor. Such modifications can be carried out either by insertion of a peptide on the glycoprotein coat, to alter its cell specificity, or by chemical modification (e.g. chemical attachment of lactose to the surface of MMLV vectors facilitates transduction into hepatocytes via the asialoglycoprotein receptor).
- Inclusion of a tissue-specific promoter to drive a therapeutic gene may ensure that the gene is expressed only in certain tissues, although the vector may be delivered to a wider range of tissues.
Nonviral Gene Delivery
Nonviral approaches to gene delivery date back to the middle of the last century, when Avery, MacLeod and McCarthy showed in 1944 that genes were transferred between bacteria by nucleic acids. Calcium phosphate-mediated transfection was developed in the early 1970s as the first nonviral technique for eukaryotic cells and is still the method of choice for the production of recombinant viral vectors. The size limitations and safety concerns of viral-based gene delivery approaches have spurred on efforts to develop safer, nonviral vectors, without limitations on insert size. A variety of nonviral approaches are now available, although the efficiency of the most effective nonviral techniques in stable gene expression is still less than that of transduction with retroviral vectors. The key differences between viral and nonviral approaches seem to lie not so much in getting DNA across the cellular membrane but in the efficiency of transport into the nucleus and the stable integration of the therapeutic genes. Viruses have evolved over a long period to go through these steps very efficiently, while nonviral delivery systems have paid little attention to these aspects of gene delivery.
This procedure is labour intensive, as DNA (or RNA) is injected into the nuclei of individual cells under a light microscope. The method is proving particularly useful in the generation of transgenic animal models with large genomic fragments from YAC, PAC and BAC clones. About 10% of the surviving embryos normally carry the microinjected transgene integrated randomly into the genome. Since such clones carry most genes as intact functional units, transgenic animals generated in this manner can be very useful in unravelling the mechanisms that underlie developmental, tissue and locus specificity in gene expression.
Electroporation usually involves the application of high voltage to a mixture of DNA and cells in suspension, although techniques for the electroporation of DNA into muscle and other tissues have also been described. The high voltage opens small holes in the cell membrane, through which the DNA enters into the cytoplasm. In bacteria, this process is very efficient, resulting in over 10 billion clones per microgram of DNA for small plasmids. Although the efficiency decreases considerably for large plasmids, the process is efficient enough to make the generation of high-quality genomic PAC and BAC libraries (clone size range 100300 kb) a rewarding endeavour.
Electroporation of mammalian cells is much less efficient because starting numbers of cells are much smaller and the DNA is delivered naked into the cytoplasm, where it can be subject to degradation. It is possible that complexing the DNA with agents that may protect it from degradation and facilitate its uptake into the nucleus will overcome some of these limitations.
Liposomes are formed by a variety of amphiphilic lipids. Chemical synthesis of a large variety of liposomes has allowed gene delivery in vitro and in vivo in various cell types and tissues, under conditions that overcome most of the safety concerns with viral vectors. Major advantages of the liposome technology over viral vectors include the defined chemical composition of the liposomes and the absence of any obvious size limitation in the DNA fragments that can be packaged. Thus, as the liposome technology improves, it should be possible to package and deliver intact genomic loci, if not whole chromosomes, into specific tissues with good efficiencies.
Cationic liposomes interact with DNA electrostatically to form condensed particles, which can then be taken up by cells. Once inside the cells, the DNA is slowly released and some of it makes it to the nucleus where it can be transiently expressed. In a very small proportion of cells, the DNA is eventually integrated into the chromosomes, leading to the establishment of stable cell lines. A large variety of cationic liposomes are commercially available and the effectiveness of these seems to vary greatly with different cell types and the conditions for lipofection. In general, lipofection uses 10
Cationic liposomes are generally unsuitable for in vivo delivery, since they tend to interact nonspecifically with many tissues. In contrast to cationic liposomes, neutral liposomes entrap DNA inside the liposome particles. Although such liposome formulations have generally been less efficient than cationic liposomes, they are potentially more useful in vivo, since they can stay in circulation for much longer and can thus be targeted more effectively to various tissues. The compaction of DNA with peptides before encapsulation to facilitate nuclear uptake may eventually enable the in vivo delivery of large genomic fragments to specific tissues.
Naked DNA injection
A variety of tissues show transgene expression after delivery of naked DNA, with expression in striated muscle persisting for long periods after a single injection. The naked plasmid DNA appears to persist primarily as free plasmid, with no significant integration into the host genome after intramuscular injection. The efficiency of DNA uptake and expression in striated muscle is affected by various parameters, including age of animals and species. Plasmids up to about 20 kb in size have been successfully expressed after naked DNA delivery to muscle. A major limitation of this approach is the low percentage of expressing myofibres. However, expression of the erythropoietin gene under such circumstances seems to be sufficient to bring about changes in the haematocrit levels of the treated animals. Similarly, injection of DNA into muscle can be used to induce the production of antibodies to the encoded protein.
Ballistic DNA injection
The gene-gun was originally developed for the introduction of DNA into plant cells, but it has since been modified to transfer genes into mammalian cells both in vitro and in vivo. The technique is restricted to skin, muscle or other organs that can easily be exposed surgically. The most exciting application of ballistic plasmid DNA injection is in DNA-based immunization. In contrast to immunization with foreign antigens, which generates only an antibody-mediated immunity, DNA-based immunization is more likely to result in a cell-mediated immune response and thus it may be more effective in immunization against a variety of viruses.
The ideal form of gene therapy involves the direct correction of the underlying genetic defect, enabling the mutant gene to recover its activity under its normal regulatory mechanisms. This type of repair may be achieved through homologous recombination, a process that involves the exchange of genetic information between two similar DNA sequences, with or without the involvement of mismatch repair. In mammalian cells, this process allows the introduction of specific changes into the genome by interaction of the endogenous sequences with homologous sequences in DNA constructs that are delivered into the cells by a variety of vectors.
A major obstacle to successful gene therapy through homologous recombination is the low efficiency of the targeting process. A large number of different proteins are involved in homologous recombination. The mechanisms are best understood in bacteria, where inducible homologous recombination systems have recently been developed. Similar studies in eukaryotic cells have yielded only marginal improvements. Further advances in our understanding of the mechanisms of homologous recombination will no doubt open novel possibilities for the therapy of genetic diseases.
Correction of mutations directly in the genome may also be achieved through the enhancement of the endogenous mismatch repair mechanisms. In order for this process to become operational, it is necessary to induce heteroduplex formation over the site to be corrected. Short PCR fragments, homologous sequences delivered with the AAV vector and specially designed chimaeric RNA/DNA oligonucleotides have all been used to induce mismatch repair correction, with variable results.
Site-specific recombination mechanisms are well known in bacteria and a small number of such systems have been adapted to eukaryotic cells. Each bacterial integrase has a large recognition site and thus there are only a limited number of potential endogenous recognition sites in the human genome. As more and more integrases become characterized, it may be possible to have a whole spectrum of different integrases for targeting therapeutic constructs into specific regions of the genome.
The Success and the Problems of Gene Therapy
Gene therapy research has gone through many ups and downs in recent years. The hype driven by the concept of gene therapy as an elixir for genetic diseases, and by the rush of investors to make a quick profit, has given way to a more careful assessment of potential risks and benefits. The death of a patient taking part in a gene therapy trial in September 1999 has renewed safety concerns over viral vectors and has rekindled the debate on the ethical aspects of gene therapy. At the same time, the successful therapy of a number of patients with severe combined immunodeficiency (SCID)-X1 disease, after ex vivo retroviral transduction of bone marrow cells, has given new momentum to the field. However, the subsequent development of a form of leukaemia in two of the patients in the SCID trial as a direct consequence of insertional mutagenesis has underscored the need for a more careful assessment of the risks involved in gene therapy. These recent developments underscore the potential of gene therapy in treating human diseases but also demonstrate that there are still some major obstacles to overcome before gene therapy becomes an essential tool of modern medicine.
The main challenges still facing gene therapy research are the following.
What sequences to deliver?
Use of cDNA sequences in therapeutic vectors can hardly ensure the regulated expression of the cDNA under a variety of physiologically relevant conditions. While some cDNA constructs may have useful applications, the availability of most genes as intact functional units in genomic DNA fragments from PAC/BAC libraries should encourage efforts at delivering intact functional loci. The delivery of an intact functional locus should overcome expression problems, as it should come under the endogenous regulatory mechanisms, even if it is integrated at a different site. Similarly, the direct correction of defects in genomic DNA by the delivery of short PCR fragments or chimaeric oligonucleotides will also restore the ability of the gene to function appropriately under a variety of physiologically relevant conditions.
How to deliver?
Viruses have evolved a number of complex mechanisms that endow them with extraordinary efficiency and specificity in delivering their genomes into the nuclei of different cell types. Similarly, some viruses have very efficient mechanisms for integration into the genome, while others can maintain their genomes extrachromosomally. The design of viral vectors for various gene therapy applications has concentrated primarily on taking advantage of some of these properties while deleting potentially harmful viral DNA sequences from the final constructs. Thus, most retroviral, lentiviral and AAV gene therapy vectors retain only the corresponding LTR sequences. However, in a number of viral vectors, some of the viral proteins and mRNAs from the packaging cell lines that may be included in the viral particles may precipitate a variety of short-term acute reactions in the recipient host cells. These short-term acute reactions and the continuing need for a careful assessment of the long-term risks to individual patients and to society further support the development of alternative gene delivery systems.
The development of nonviral gene delivery systems is aiming to take advantage of our increasing understanding of the mechanisms for the compaction of DNA in various physiological states, as well as for the packaging and delivery of therapeutic sequences across the cellular and nuclear membranes. The incorporation of additional mechanisms for the maintenance of the therapeutic DNA as independent minichromosomes or episomes, or for its targeted integration through homologous or site-specific recombination mechanisms, should enable nonviral gene delivery to become a safe and indispensable tool of medicine in this century.
Target cell specificity
Delivery of therapeutic genes to a specific tissue and/or cell type in vivo represents one of the major hurdles in gene therapy for somatic gene diseases. While the design of viral vectors has often taken advantage of the natural tissue specificity of different viruses, various strategies have also been developed to modify such targeting specificity, so as to modify the types of cells that are susceptible to transduction by each viral vector. Some of these approaches are readily adaptable for use in nonviral delivery systems. Since many cDNA constructs do not display a high degree of cell specificity in the expression of the cDNA, a high degree of cell specificity in the delivery is desirable to reduce unwanted effects on other cell types. In contrast, since expression of intact functional loci is highly cell specific, it is anticipated that there will be less need for cell targeting during delivery of intact functional loci.
Fate of DNA in cells
Retroviruses integrate very efficiently into the genome of dividing cells and this has been one of the main reasons for their development as gene therapy vectors. Lentiviruses appear better in this respect, since they are also capable of gene delivery and integration in nondividing cells. The only elements that appear to be needed for integration are the LTR sequences and the viral integrase. It has generally been assumed that integration is not reversible. However, the recent demonstration that lentiviral vectors can be mobilized to integrate at additional sites in the course of HIV infection is of particular concern, since it highlights the potential for interaction between viral vectors and other ongoing viral infections in patients.
It is clear that the random integration of viral vectors into patients chromosomes cannot provide a safe approach for the widespread use of gene therapy to alleviate human disease. The enhancement of homologous or site-specific recombination mechanisms to facilitate targeted integration of therapeutic genes into specific sites in the genome, or their maintenance in independent artificial minichromosomes or episomes, may provide a safe alternative approach.
Gene therapy is emerging as one of the most potent tools of medicine for the treatment of genetic and somatic gene diseases. Although the problems associated with the design, delivery and fate of therapeutic constructs into patient cells have been grossly underestimated and oversimplified, leading to unrealistic and overoptimistic expectations, the completion of the sequencing phase of the Human Genome Project is expected not only to catalyse the development of effective therapies for many diseases in the first half of this century but also to provide a solid basis for the biological emancipation of mankind by the end of the twenty-first century.
- Antisense oligonucleotide
- A short synthetic fragment of nucleic acid of a defined sequence that is complementary (antisense) to a specific sequence in the target messenger RNA (mRNA). Such binding may induce degradation of the mRNA, or modify its interaction with various factors involved in mRNA processing and translation.
- cDNA (complementary DNA) is a DNA copy of a messenger RNA, usually made with the enzyme reverse transcriptase.
- Chimaeric oligonucleotide
- A short synthetic fragment of nucleic acid of a defined sequence made from both RNA and DNA bases. The RNA sequence facilitates binding to a target sequence in genomic DNA to induce heteroduplex formation and correction of mutations.
- A DNA molecule (usually circular) that is not part of the cell's chromosome(s) and has its own mechanisms for replication and maintenance. Plasmids are an example.
- A double-stranded DNA molecule or DNARNA hybrid, where the two strands are of different origin and have very similar but not identical sequences. Such hybrids will typically have one or more mismatched base pairs.
- An enzyme that helps integrate DNA fragments into the genome. This usually takes place through site-specific recombination, although the degree of sequence specificity and the recognition sequence varies greatly between different integrases.
- Long terminal repeats (LTR)
- A particular sequence of nucleotides that appears on both ends of the DNA or RNA genome of some viruses and has a number of key functions for viral gene expression, replication and integration.
- Mismatch repair
- One of the endogenous mechanisms for correcting mispaired bases in DNA resulting from damage to one of the strands or through heteroduplex formation during recombination.
- PAC/BAC libraries
- Large collections of bacterial clones, usually arranged in 384-well microtitre plates, each of which contains a fragment of genomic DNA up to about 300000 base pairs long. In BACs, such fragments are cloned as bacterial artificial chromosomes in a vector based on the F plasmid origin of replication. PACs are very similar, except that the origin of replication is derived from the P1 bacteriophage.
- Transfer of recombinant nucleic acids from one cell to another by way of a viral vector.
- Introduction of nucleic acids into eukaryotic cells by nonviral means.
- Yeast artificial chromosome is an additional yeast chromosome created by the cloning of a fragment of genomic DNA, usually in the range of 100000 to 1000000 bases long in a YAC vector.