Rapid, large-scale genome as well as diagnostic DNA sequencing projects are, at present, dependent on the use of sensitive automated sequencers that rely on the detection of fluorescent signals. This emission is not an intrinsic property of the biomolecules but is a property of an optical label that must be incorporated before, during or after the sequence specific reaction. The choice of strategy, that is to be reviewed here, is a function of both the chemistry and the enzymology of the system as well as the nature of the fractionation and the physical parameters of the detection. One may differentiate between the labelling of the primer, incorporation of the label during elongation or base specific termination with labelled terminators. In reviewing the development of each of these strategies, the conclusion is reached that, whereas labelled primers have universal applicability, the current generation of fluorescent terminators have in, conjunction with appropriate enzymes, attained a refinement that strongly favours their use. However, since they are not available for all commercial sequencing systems, the alternative procedures are not merely of historic interest but are an essential component in DNA sequencing protocols.

Less than ten years after the milestone in molecular biology set by Maxam and Gilbert (1977) and Sanger et al (1977) with their development of DNA sequencing methods, means of automating at least some of the steps of these protocols were beginning to emerge (Ansorge et al., 1986; Smith et al., 1986, Prober et al., 1987). Fundamental to laying the foundation to large-scale sequencing, culminating in the Human Genome Project, was a replacement of conventional radioactive detection methods by optical systems much more amenable to automation (as well as reflecting the increasing environmental consciousness). Whereas hardware for stimulating the emission of fluorescence by biological molecules as well as highly sensitive systems for its detection was readily available, the lack of significant intrinsic fluorescent properties of any of the natural nucleobases was one key aspect which had to be considered at the outset. In contrast to radioactive labels, the introduction of an optical label - of necessity a chemical and physical perturbation of the system - leads to the questions of fundamental significance to both the biochemical and the physical part of the process. At what stage of the sequence reaction can one introduce a fluorescent tag? Which fluorophore is compatible with the excitation/detection system? The continued upgrading to new generations of the original instruments (Quesada and Zhang, 1996) and efforts to improve the properties of the dyes (Metzker et al., 1996; Hung et al., 1996a,b; Lee et al., 1997) confirm that these considerations are still of relevance.

The chain termination procedure introduced by Sanger et al. (1977) is dependent on the use of a short oligonucleotide primer, complementary to a target sequence on the DNA, which is enzymatically elongated and subsequently base-specifically terminated (Fig. 1).

Figure 1.Schematic representation of the DNA sequencing reaction, according to Sanger.

A. General structure of a chain terminating dideoxynucleotide (ddNTP). B. A DNA strand acts as a template for binding a short complementary synthetic primer. The primer is extended by the action of DNA polymerase in the presence of all four deoxynucleotides and is terminated base-specifically by partial incorporation of ddNTP. The terminated DNA fragments resulting from a partial inhibition of the DNA polymerase by the presence of a ddNTP is exemplified using ddTTP. Reactions with each of the other three ddNTPs generate a set of corresponding terminated strands. The reaction products are size fractionated by electrophoresis and the relative position of the fragments is directly correlated with the nature of the terminating ddNTP and hence of the DNA sequence.

With the advent of chemical methods for the efficient synthesis of oligonucleotides (Caruthers et al., 1987) and the emergence automated DNA synthesisers in the early '80s, the Sanger method has become the method of choice for all DNA sequencing projects. For the detection of the products of this reaction, requiring the incorporation of a label, a number of options exist irrespective of whether radioactive isotopes, fluorescent dyes or other tags are used. The available choices are summarised schematically in Fig. 2.

A. Labelling the primer (Fig. 2A)

The 5' terminus of the primer has been a popular target for the attachment of a variety of labels. It is, intuitively, a likely candidate since it is well separated from the 3' end where enzymatic processes during elongation, might be expected to be negatively affected by a chemical modification. On the other hand, as far as a direct enzymatic addition of a label is concerned, the 5' OH is particularly inert. The initial development of automated DNA sequencers depended on the ability of adding a reactive 5' terminal group to the oligonucleotide during its synthesis. This modification permitted a subsequent reaction of this residue with chemically reactive dye derivatives. The phosphoramidite monomers used for adding the 5' terminal moiety were indispensable until building blocks for the direct incorporation of labels became available and are still essential for tagging with novel reagents. Unfortunately, while being ideal for labelling standard primers, which are used repeatedly, it is totally uneconomical to use commercial dye-phosphoramidite preparations for the synthesis of specific or "walking" primers unless one can co-ordinate a demand for large numbers of such oligonucleotides (Hawkins et al., 1992). A further argument against primers modified during chemical synthesis is that by blocking the 5' end one is automatically excluding other potential applications of this primer.

For the purposes of introducing a single specific label at the 3' terminus enzymatically, the natural polymerisation activity of the enzyme terminal deoxynucleotidyl transferase (TdT) must be restricted. The analogs fluorescein- (or biotin)-modified ribotriphosphates were shown to be suitable substrates that could be efficiently used by TdT but that reduced the elongation of a sequencing primer to a single nucleotide (Igloi and Schiefermayr, 1993). The procedure remains to date the only direct enzymatic, template independent method for labelling pre-existing primers.

Automated sequencing hardware based on the detection of fluorescence suffers the disadvantage of a high initial outlay of funds. An alternative, avoiding radioactive methods, which has been shown to be capable of producing high volume data (Pohl and Maier, 1995) is sequencing by direct blotting. During this procedure the products of a gel-electrophoretic fractionation are blotted onto a membrane and visualised by an enzyme linked assay.

B. Labelling the growing DNA chain (Fig. 2B).

Conventional radioactive sequencing protocols almost without exception utilise the incorporation of a labelled dNTP as the basis of the detection. The direct transfer of this concept to non-isotopic systems is not trivial. One would anticipate that a bulky modification to the triphosphate might influence the system at several stages. a) The dNTP may no longer be an effective substrate for the DNA polymerase b) If incorporated, sequences requiring multiple adjacent labelled nucleotides may lead to chain termination c) The electrophoretic migration of the fragments would be perturbed in an unpredictable fashion, depending on the number of labels incorporated. Surprisingly and despite these misgivings, the use of fluorescein-dUTP, believed to bring about a multiple labelling, was reported to give unambiguous sequences (Voss et al., 1991) using T7 DNA polymerase. A screening of the polymerases, which had in the meantime become a target for genetic manipulation (Tabor and Richardson, 1995) showed that by no means all enzymes routinely used in sequencing would accept fluorescent dNTPs as substrates (Voss et al., 1997). In particular, only two of the thermostable DNA polymerases, used in the much favoured cycle sequencing protocol (Carothers et al., 1989), could be induced to incorporate these analogues. However, some other dye-labelled dNTPs, of relevance for the latest generation of automated sequencers using red laser excitation could be added to the repertoire. In contrast to the substrate selectivity of terminal transferase (Igloi 1996a), the remarkable tolerance of some DNA polymerases towards modified nucleotides (Voss et al., 1997) provides a useful pathway to template dependent incorporation of non-fluorescein dNTPs.

C. Labelling the termination (Fig. 2C).

The conceptually most satisfying approach to the introduction of a label during the Sanger sequencing protocol is via a labelled terminating dideoxyNTP. Using such fluorescent analogues bypasses the difficulties of labelling the primer or of designing the primer for a successful enzymatic internal labelling. The interpretation of the sequence pattern is simplified by the visualisation of only those products arising out of a defined enzymatic termination, rather than from a random fragmentation, reduces the background and facilitates the automated evaluation. Despite the elegance of this approach, in practice several non-trivial problems involving the enzymology of the system were only solved by establishing the molecular basis for the recognition of ddNTPs by DNA polymerase (Tabor and Richardson, 1995). It now became possible to manipulate the enzyme genetically to such an extent that not only ddNTPs but also dye-labelled ddNTPs became rather good substrates. Further tuning of the dye component of the terminators now provided dichlororhodamine and, in particular, coupled fluorescein-rhodamine energy transfer dye-terminators (Rosenblum et al., 1997). With this development the use of dye terminators has finally come of age and there is now no reason why it should not become the method of choice in future electrophoresis-based fluorescent automated sequencing.

References

Ansorge, W., Sproat, B.S., Stegemann, J. and Schwager, C. (1986). A non-radioactive automated method for DNA sequence determination. Journal of Biochemical and Biophysical Methods 13:315-323.

Carothers, A.M., Urlaub, G., Mucha, J., Grunberger, D. and Chasin, L.A. (1989). Point mutation analysis in mammalian gene: rapid preparation of total RNA, PCR amplification of cDNA and Taq sequencing by a novel method. BioTechniques 7:494-499.

Caruthers, M.H., Barone, A.D., Beaucage, S.L., Dodds, D.R., Isher, E.F.F., McBride, L.J., Matteucci, M., Stabinsky, Z. and Tang, J.-Y. (1987). Chemical synthesis of deoxyoligonucleotides by the phosphoramidite method. Methods in Enzymology 154:287-313.

Hawkins, T.L., Du, Z., Halloran, N.D. and Wilson R.K. (1992). Fluorescent chemistries for automated primer-directed DNA sequencing. Electrophoresis 13:552-559.

Hung, S.-C., Ju, J., Mathies, R.A. and Glazer, A.N. (1996a). Cyanine dyes with high absorption cross section as donor chromophores in energy transfer primers. Analytical Biochemistry 243:15-27.

Hung, S.-C., Ju, J., Mathies, R.A. and Glazer, A.N. (1996b). Energy transfer primers with 5- or 6-carboxyrhodamine-6G as acceptor chromophores. Analytical Biochemistry 238:165-170.

Igloi, G.L. (1996a). Substrate properties of fluorescent ribonucleotides in the terminal-transferase-catalyzed labelling of DNA sequencing primers. BioTechniques 21:1084-1092.

Igloi, G.L. and Schiefermayr, E. (1993). Enzymatic addition of fluorescein- or biotin-riboUTP to oligonucleotides results in primers suitable for DNA sequencing and PCR. BioTechniques 15:486-497.

Lee, L.G., Spurgeon, S.L., Heiner, C.R., Benson, S.C., Rosenblum, B.B., Menchen, S.M., Graham, R.J., Constantinescu, A., Upadhya, K.G. and Cassel, J.M. (1997). New energy transfer dyes for DNA sequencing. Nucleic Acids Research 25:2816-2822.

Maxam, A.M. and Gilbert, W. (1977). A new method for sequencing DNA. Proceedings of the National Academy of Sciences USA 74:560-564.

Metzker, M.L., Lu, J. and Gibbs, R.A. (1996). Electrophoretically uniform fluorescent dyes for automated DNA sequencing. Science 271:1420-1422.

Pohl, T.M. and Maier, E.(1995). Sequencing 500kb of yeast DNA using a GATC 1500 direct blotting electrophoresis system. BioTechniques 19:482-486.

Prober, J.M., Trainor, G.L., Dam, R.J., Hobbs, F.W., Robertson, C.W., Zagursky, R.J., Cocuzza, A.J., Jensen, M.A. and Baumeister, K. (1987). A system for rapid DNA sequencing with fluorescent chain-terminating dideoxynucleotides. Science 238:336-341.

Quesada, M.A. and Zhang, S. (1996). Multiple capillary DNA sequencer that uses fiber-optic illumination and detection. Electrophoresis 17:1841-1851.

Rosenblum, B.B., Lee, L.G., Spurgeon, S.L., Khan, S.H., Menchen, S.M., Heiner, C.R. and Chen, S.M. (1997). New dye-labelled terminators for improved DNA sequencing patterns. Nucleic Acids Research 25:4500-4504.

Sanger, F., Nicklen, S. and Coulson, A.R. (1977). DNA sequencing with chain terminator inhibitors. Proceedings of the National Academy of Sciences USA 74:5463-5467.

Smith, L.M., Sanders, J.Z., Kaiser, R.J., Hughes, P., Dodd, C.,Connell, C.R., Heiner, C., Kent, S.B.H. and Hood, L.E. (1986). Fluorescence detection in automated DNA sequence analysis. Nature 321:674-679.

Tabor, S. and Richardson, C.C. (1995). A single residue in DNA polymerases of the Escherichia Coli DNA polymerase I family is critical for distinguishing between deoxy- and dideoxyribonucleotides. Proceedings of the National Academy of Sciences USA 92:6339-6343.

Voss, H., Nentwich, U., Duthie, S., Wiemann, S., Benes, V., Zimmermann, J. and Ansorge, W. (1997). Automated cycle sequencing with Taquenase: protocols for internal labelling, dye primer and "doublex" simultaneous sequencing. BioTechniques 23:312-318.

Voss, H., Schwager, C., Wirkner, U., Zimmermann, J., Erfle, H., Hewitt, N.A., Rupp, T., Stegemann, J. and Ansorge, W. (1991). A new procedure for automated DNA sequencing with multiple internal labelling by fluorescent dUTP. Methods in Molecular and Cellular Biology 3:30-34.