DNA Libraries 

Recombinant DNA technologies developed in the early 1970s (1-6) form the core of all DNA-based molecular combinatorial libraries. Genetic sequences of interest are recombined with a replication-competent DNA vector, such as a plasmid or bacteriophage. Each individual genetic sequence fragment is enzymatically ligated into a receptive vector, to form a unique recombinant clone. The population of individual recombinants is then reintroduced into an appropriate host cell in order to constitute a functional library. Modern cloning technologies allow the preparation of up to 109 recombinants in a single library, with greater numbers requiring a substantial scale-up that is beyond the resources of most researchers.

 A different type of recombinant design is seen in some oligonucleotide-based libraries that require no vector for their survival (7-9). The replication elements in these oligonucleotides are built into their sequences in the form of primer binding sites. These libraries can be amplified in vitro by the polymerase chain reaction (PCR) using additional oligonucleotides capable of annealing to the termini of the library molecules (10). Other amplification technologies can be used to replicate oligonucleotide libraries, including the ligase chain reaction (11), self-sustained sequence replication (12) , and strand displacement amplification (13). Because oligonucleotide-based libraries are prepared chemically and require no cloning steps, they can achieve very high sequence complexities on the order of 1 × 10¹⁵ different species or more.

 Recent technological advances have introduced a novel type of oligonucleotide-based DNA library in which defined sets of sequences are synthesized chemically and displayed in a two-dimensional array. A highly publicized approach developed by Affymetrix employs photolithographic masking procedures to construct very high-density oligonucleotide arrays on silicon chips. One such product developed by Affymetrix is the p53 chip, which displays a large number of common disease-related p53 gene mutations represented as oligonucleotide sequences. Any clinical specimen can be rapidly assayed over this chip by hybridization of nucleic acids in order to determine the genotype of p53 genes expressed by the specimen tissue. The development of further disease-related gene chips has enormous potential to carry clinical diagnosis to the molecular level where appropriate highly individualized therapies may be ascertained (eg, pharmacogenomics (14)).

The intended use of recombinant libraries determines numerous aspects of library design and construction. The vector provides essential functions, such as an origin of replication, that enable replication and survival of the recombinant clone within the host cell. A selectable marker such as an antibiotic-resistance gene provided by the vector provides a means to eliminate any cells that have not taken up a recombinant clone. Unique restriction enzyme sites in the vector allow efficient insertion of foreign DNA fragments into defined contexts. Various expression control elements engineered into the vector permit selective expression of RNA or protein encoded by the cloned DNA fragment. Vectors that contain certain viral signal sequences can be packaged into viral particles in infected cells, thereby enabling highly efficient viral-mediated gene transfer into noninfected cells. In addition, cloning vectors often provide genetic markers that confer desirable properties upon the expression products. For example, DNA encoding an open reading frame may be fused in frame to the C-terminus of glutathione-S--transferase (GST) (15). The resulting GST-fusion protein can then be easily purified by affinity chromatography over glutathione-agarose resin. Finally, “shuttle” vectors contain multiple origins of replication, supporting growth of the recombinant clone in bacteria as well as in relevant experimental systems, such as yeast and mammalian cells.

Plasmid vectors are by far the most commonly used cloning vectors for relatively short DNA fragments ranging from several hundred to several thousand nucleotides. This size range is sufficient to carry the coding sequence portions of most genes. However, because genes from many organisms contain varying amounts of nonexpressed intervening sequences (introns), cloning of complete genes or gene clusters often requires the use of vectors that can accommodate DNA inserts exceeding 10 kilobase pairs (kbp) or more. The lambda phage particle can package approximately 50 kbp of DNA, but the lambda phage itself requires at least 30 kbp of essential phage genes for viability, limiting its insert carrying capacity to no more than 20 kbp. Some improvement is seen with “cosmid” vectors, which are composite vectors containing both plasmid replication elements and the lambda phage cos packaging signal. One type of cosmid vector can package up to 40 kbp of foreign DNA into lambda phage capsids (16). Gene mapping studies and investigations into chromosome structure require cloning vectors with even larger capacity, called artificial chromosomes (ACs). Thus, YACs are yeast artificial chromosomes that contain yeast centromere, telomere, and autonomous replication sequence (ARS) elements, as well as appropriate yeast selection markers (17, 18). YACs allow the cloning and maintenance of fragments of genomic DNA up to several hundred kilobase pairs in length. Similarly, ACs have been constructed that allow introduction of large foreign sequences into bacteria (called BACs) and mammalian cells (called MACs) (19).

DNA libraries may be categorized according to the source and the context of foreign DNA within the library. Genomic libraries often contain relatively large DNA inserts in phage or AC vectors. These inserts are typically not expressed, although there are exceptions. cDNA library inserts are derived from DNA copies of messenger RNA that are generated by the enzyme reverse transcriptase. mRNA is a useful source for protein-coding DNA sequences, as the nonexpressed intervening sequences have been removed through the process of pre-mRNA processing. Thus, protein-coding sequences that are interspersed throughout a gene are arranged contiguously within a cDNA, whereupon it becomes trivial to infer the encoded protein sequence using the genetic code. cDNA libraries usually provide some means to express open reading frames encoded within the DNA insert, both as RNA transcripts and protein translation products. Expression libraries may, however, also be based on genomic DNA and oligonucleotides, either cloned into a vector or as linear molecules that can be amplified in vitro.

 “Virtual libraries” are a relatively new concept in DNA libraries. Sequence information emerging from large-scale sequencing projects is incorporated into ever-expanding computer databases, where cross-references can be established between any sequence and its associated functional information. Thus, a database can serve as a virtual library that can greatly expedite the identification and recovery of molecules with desired properties. Examples of public sequence databases include

GenBank and the Cancer Genome Anatomy Project (CGAP). Such virtual libraries enable tasks that once required working with actual molecules in vitro to be accomplished much more rapidly in silico using only the information contained within those molecules. For example, a scientist that identifies a sequence fragment of a protein associated with a particular disease can, through the genome databases, identify a full-length DNA sequence associated with that protein. The DNA sequence in turn provides a great deal of useful information, including the likely sequence and structure of the protein, as well as a set of primer sequences that can be used to clone the full-length gene representing that protein. Many modern database resources also maintain banks of cosmid or AC clones that are (or will be) correlated with every sequence in the database. Such resources make it possible to move from an unknown protein band obtained through gel electrophoresis to a cloned and expressed gene for that protein in a matter of days—a task that often required several years to accomplish only a decade ago.

References

1. R. M. Lawn, E. F. Fritsch, R. C. Parker, G. Blake, and T. Maniatis (1978) Cell 15, 1157–1174. 

2. S. N. Cohen, A. C. Chang, H. W. Boyer, and R. B. Helling (1973) Proc. Natl. Acad. Sci. USA 7o, 3240-3244.

3. N. E. Murray and K. Murray (1974) Nature 251, 476–481. 

4. M. Thomas, J. R. Cameron, and R. W. Davis (1974) Proc. Natl. Acad. Sci. USA 71, 4579–4583.

5. T. Maniatis, R. C. Hardison, E. Lacy, J. Lauer, C. O''Connell, D. Quon, G. K. Sim, and A. Efstratiadis (1978) Cell 15, 687–701. 

6. G. K. Sim, F. C. Kafatos, C. W. Jones, M. D. Koehler, A. Efstratiadis, and T. Maniatis (1979( Cell 18, 1303–1316. 

7. C. Tuerk and L. Gold (1990) Science 249, 505–510. 

8. A. D. Ellington and J. W. Szostak (1990) Nature 346, 818–822. 

9. D. E. Tsai, D. S. Harper, and J. D. Keene (1991) Nucleic Acids Res. 19, 4931–4936. 

10. K. Mullis, F. Faloona, S. Scharf, R. Saiki, G. Horn, and H. Erlich (1986) Cold Spring Harb. Symp. Quant. Biol. 51, 263–273. 

11. F. Barany (1991) Proc. Natl. Acad. Sci USA 88, 189–193. 

12. R. R. Breaker and G. F. Joyce (1994) Proc. Natl. Acad. Sci. USA 91, 6093–6097. 

13. G. T. Walker, M. S. Fraiser, J. L. Schram, M. C. Little, J. G. Nadeau, and D. P. Malinowski (1992) Nucleic Acids Res. 20, 1691–1696. 

14. A. Persidis (1998) Nat. Biotechnol. 16, 209–10. 

15. D. B. Smith and K. S. Johnson (1988) Gene 67, 31–40. 

16. B. Hohn and J. Collins (1980) Gene 11, 291–298. 

17. P. Guzman and J. R. Ecker (1988) Nucleic Acids Res. 16, 11091–11105. 

18. A. Coulson, R. Waterston, J. Kiff, J. Sulston, and Y. Kohara (1988) Nature 335, 184–186. 

19. M. Ikeno, B. Grimes, T. Okazaki, M. Nakano, K. Saitoh, H. Hoshino, N. I. McGill, H. Cooke, and H. Masumoto (1998) Nat. Biotechnol. 16, 431–439.