The Referential Convergence of Gene Concepts Based on Classical and Molecular Analyses

Abstract

Kenneth Waters and Marcel Weber argue that the joint use of distinct gene concepts and the transfer of knowledge between classical and molecular analyses in contemporary scientific practice is possible because classical and molecular concepts of the gene refer to overlapping chromosomal segments and the DNA sequences associated with these segments. However, while pointing in the direction of coreference, both authors also agree that there is a considerable divergence between the actual sequences that count as genes in classical genetics and molecular biology. The thesis advanced in this paper is that the referents of classical and molecular gene concepts are coextensive to a higher degree than admitted by Waters and Weber, and therefore coreference can provide a satisfactory account of the high level of integration between classical genetics and molecular biology. In particular, I argue that the functional units/cistrons identified by classical techniques overlap with functional elements entering the composition of molecular transcription units, and that the precision of this overlap can be improved by conducting further experimentation.

Acknowledgements

I would like to thank Lindley Darden, Erika Milam, Eric Saidel, Pamela Henson, Joan Straumanis, Elizabeth Schechter, Brendan Ritchie, the DC History and Philosophy of Biology group, as well as two anonymous referees of this journal for helpful discussion and comments on earlier drafts. This work was supported by the Fonds de la recherche sur la société et la culture, Québec, Canada, grant number 127231.

Notes

[1] Waters initially argued that, in order to count as a gene, a segment of DNA must be ‘biochemically active’ (Waters 1990, 130). According to this conception, sequences required for recruiting the transcriptional machinery are part of the gene, suggesting that the DNA sequences to which Waters refers include open reading frames, as well as transcription units. In a subsequent development, Waters argues that molecular genes are sequences of DNA coding for product sequences generated at some point during gene expression (Waters 1994, 178). One peculiarity of this homology‐based conception is that only coding DNA counts as genes, while regulatory regions (promoters, enhancers) are left out the gene; this entails that the concept refers to open reading frames, but not to transcription units.

[2] Differences in DNA sequence can be investigated by molecular techniques, hence supporting Waters’s (2008) subsequent view that molecular biology contributed to an experimental/methodological ‘retooling’ of classical genetics.

[3] The position of the gene was narrowed down by classical analysis, via cytological band mapping on polytene salivary gland chromosomes in Drosophila (giant, easily observable chromosomes formed by multiple DNA replications without mitosis). Then, transcription units and open reading frames (i.e., molecular genes) were identified by sequencing and annotating DNA fragments binding specifically the same position on the chromosome. By inserting the mutant or the wild‐type version of the molecular gene, researchers were able to produce the same differences in phenotype associated with the mutant and wild‐type allele of the classical gene, thus proving that the classical and molecular genes referred to the same DNA sequences, located at the same chromosomal locus.

[4] In classical terms, such mutations are distinguished solely by different recombination frequencies (e.g., for the most part, Benzer worked only with the rapid mutant, phenotype caused by several distinct mutations in the rII region).

[5] The chromosomal fragments/DNA sequences that count as units of mutation and recombination vary with the resolution of the experimental techniques used. In this sense, Weber’s (2005, 224) claim that the evolution of gene concepts was driven by the development of new techniques is correct. However, I also want to point out that some concepts were irrevocably abandoned because of their incompatibility with new discoveries.

[6] I think that Weber’s concern with natural kinds is blurring the issue. Appealing to the kinds of entities classical and molecular geneticists thought genes must be in order to fulfil certain explanatory roles makes it extremely difficult to provide an account of how it is possible for geneticists to use side‐by‐side concepts issued from classical genetics and molecular biology. For example, as Weber (2005, 209) himself argues, early geneticists thought genes have two distinctive properties: they are units, such that they can segregate and assort; and they are causes of phenotypic traits. Later, Muller (1951) conjectured that, in order to perform the role classical genetics ascribes them, genes must possess three properties: the abilities to replicate, to contribute to phenotypes and to mutate; he conjectured that genes are most likely proteins. After the elucidation of the structure of DNA, it became clear that DNA is a linear polymer unable to perform enzymatic activities; genes were then said to ‘store information’. Finally, contemporary genetics defines genes in terms of structural motifs required for the recruitment of the gene expression machinery of the cell. If all these concepts refer, it follows that there are genes as causal entities, protein genes, blueprint genes and structural‐motifs genes. Set aside the obvious problem that not all these entities really exist, it is not at all clear how such radically different kinds of entities could overlap extensionally. I think that the key to understanding the high level of integration between classical genetics and molecular biology lies in the fact that, leaving aside for a moment the various causal roles attributed to genes at various points in time, the classical and molecular concepts retained until today consistently include reference to a chromosomal locus and its associated DNA sequence. Thus, the real question is not about kinds, but about the extent to which the DNA sequences picked by classical and molecular gene concepts overlap.

[7] There are only six chromosomal break‐points defining five functional regions of about 25,000 base‐pairs each in the achaete–scute study (Weber 2005, fig. 7.2). This is a very small number compared to the thousands of recombination points analyzed by Benzer and is immediately reflected in the resolution of the map (25,000 vs. single nucleotides).

[8] Weber (2005, 227) acknowledges that ‘if the regulatory regions are considered to be part of a molecular gene, then some molecular genes identified in Drosophila corresponded roughly to classical genes’, but doesn’t seriously consider this avenue of investigation. Yet regulatory sequences are part of the transcription unit concept (which can be seen as an open reading frame and its associated regulatory sequences). Not only is the latter an officially recognized and widely used concept (Wain et al. 2002 above; the NCBI GenBank guidelines), it also provides an indispensable theoretical framework for most genome annotation practices in use today: putative genes are identified on the basis of conserved promoter‐like regulatory sequences preceding coding/transcribed/open reading frame sequences (Altschul et al. 1990; Mandoiu and Zelikovsky 2008).