The first evidence and the first demonstration that cDNA could be cloned in a bacterial plasmid came from work by F. Rougeon, P. Kourilsky and B. Mach in 1975, that was published in Nucleic Acids Research [1]. They showed that double stranded DNA made by reverse-transcribing rabbit globin messenger RNA could be recombined in vitro with a pBR22 plasmid DNA, and the recombinant thus obtained used to transfect E. coli cells, enabling one to further isolate clones of globin cDNAs. This was soon followed by similar work by the groups of Rabbitts [2] and Maniatis [3].

The recombinant DNA technology later resulted in a wealth of experiments giving rise to the cloning of many eukaryotic genes including some early examples obtained at the Pasteur Institute on chicken ovalbumin [4], mouse immunoglobulin chains [5–8], human growth hormone [9], mouse histocompatibility antigens [10], mouse rennin [11] and the acetylcholine receptor alpha subunit of Torpedo marmorata [12].

Major observations were made between 1976 and 1978 about the physiological modifications of eukaryotic RNA transcripts. For example, it was shown by G. Brawerman and U. Littauer that, in most cases, eukaryotic messengers are polyadenylated at their 3′ ends. Although the role of this poly-A tail is not fully understood, specific poly-A binding proteins have been described and it has been postulated that they could be responsible for the extra nuclear transport of primary transcripts and for controlling their metabolic stability. At the 5′ end, another type of modification taking place is the phenomenon of ‘capping’ consisting upon a terminal guanylation by a guanylyl-transferase, plus a methylation of the guanine residue by a 7-methyl transferase, as shown initially independently by Lewin and Furuichi [13]. This guanine-rich cap is believed to exert some transient protection of RNA transcripts and to prevent them to adopt a spurious secondary structure.

But what came out as the greatest surprise – particularly to molecular biologists trained in microbial and phage genetics – was the discovery of ‘split genes’ by P. Sharp and the observation by many others, including P. Chambon, that the product of these genes are long transcripts that have to be processed (spliced), according to a mechanism of exon shuffling, in order to give rise to mature messenger RNAs. The existence of self-splicing mechanisms operating in the excision of particular introns from Tetrahymena ribosomal RNA (T. Cech), or from plant viral RNAs, further led to the exciting concept of catalytic RNAs, so called ‘ribozymes’, a concept which has had considerable implications in the field of molecular evolution, and which is finding recent support in the manifestations of the so called ‘RNA world’ (interfering RNA, riboswitches, non sense-mediated mRNA decay, etc., see for example [14–18]).

But, between 1980 and 1985, molecular and developmental biologists turned most of their attention to the control of eukaryotic transcription itself. Among the names that have best illustrated this field in its early phase one should mention people like R.G. Roeder, P. Chambon, M. Yaniv, A. Klug, etc. The main message that came out from their work was that, contrary to the situation observed with prokaryotes, gene regulation in eukaryotes obeys, in most cases, a positive rather than negative control, which is achieved by the attachment of transactivating factors to cis-acting regulatory sequences lying upstream of the transcription start-site. As illustrated in Fig. 1, showing as an example the desmin gene control region, a three dimensional arrangement of these different factors has to be taken into consideration to understand how they operate to stimulate the activity of the RNA polymerase. The role of ‘enhancers’ presumably involved in this type of interaction at the chromatin level was shown to be a very crucial one.

It is clear that the technical analysis of cellular transcripts went through different phases some of which consisting of crude approaches to global cell sub-fractions, others being more sophisticated, including fractionation of mRNA by sucrose gradient ultracentrifugation or preparative electrophoresis [19]. For example, and as already mentioned, bacterial messenger RNA was initially defined as a rapidly labelled, heterogeneous, and metabolically unstable RNA fraction capable of reversible attachment to ribosomes. In other instances, its existence was also inferred from its capacity to stimulate protein synthesis in a reconstituted cell-free system. Later on, following work on RNA polyadenylation, the messenger subfraction from eukaryotic cells could be purified by chromatography on oligo-dT-cellulose columns. Then radioactive cDNA became available, the PCR amplification technique came to age, and more sophisticated methods could then be used. Instead of addressing messenger RNA as a global component of the cell, it then became possible to deal with gene-specific transcripts. Combining detection with labelled cDNA probes and electrophoretic separation, the ‘Northern blot’ analysis greatly facilitated the identification of specific gene transcripts in cell-free extracts.

More recently, with the onset of genomics, new technologies have been developed that give the possibility to monitor the expression of hundreds or thousands of genes at a time. These technologies are essentially based on the utilization of high-density microarrays (DNA chips). They usually involve hybridization of total cDNA samples derived from the whole collection of RNA transcripts present in a cell extract, to genes or to oligo-chips immobilized on solid supports such as nylon membranes or glass slides.

Today, using oligonucleotides, cDNAs or gene fragments present on these supports, it is possible to have access to the whole set of messenger RNAs that are expressed by the cell or the tissue under precise physiological or developmental conditions. This complex collection of transcripts is designated under the name of ‘transcriptome’. It is the ‘signature’, so to speak, of the total gene expression pattern of the cell at any given moment of its normal or pathological fate.

Recent advances in different technical fields such as material engineering, optics, electronics, robotics, chemistry, genetic engineering and informatics have permitted the development of these miniaturized platforms, the utilization of which is extremely versatile. They can be used to help biologists to scrutinize the cell physiology in general. For example, transcriptional regulatory circuits, cell division patterns, apoptosis, communication and signal transduction networks, etc., all of these parameters can influence gene expression in such a way as to affect the transcriptome.

The microarray techniques can also afford new tools for diagnosing diseases, for establishing cancer prognosis, for assisting the field of drug development (pharmacogenomics), or else to tailor therapeutics to specific pathologies, and, of course, to generate databases.

There exist several types of microarrays [20] (Fig. 2). Sometimes the oligonucleotides serving as probes for detecting cell transcripts or their cDNAs can be synthesized directly in situ, by photolithographic techniques (Affymetrix); the synthesis takes place on glass support and up to half a million probes can actually be synthesized on a very small surface (1.3 cm²). The same can be realized by an ink-jet system. On the other hand, pre-fabricated oligos or cDNAs can be printed on glass or Nylon supports by robots; this procedure was utilized to analyse gene expression in yeast, since close to 5–6000 genes from this organism could be deposited on a surface not exceeding 2 cm².

DNA arrays are also quite utilized by physicians or research scientists confronted with the nosology (categorization) and prognosis of various types of cancers.

For example, Lander and Golub [21] succeeded in establishing interesting comparisons at the transcriptome level between acute myeloid leukaemia (AML) and acute lymphoblastic leukaemia (ALL), two types of blood diseases that are not always easy to distinguish in their early stages using classical approaches. Gene profiling study of the RNA transcripts from 40 leukaemic patients (with micro-arrays containing 7000 human gene probes) could show that up to 50 genes were differentially expressed in AML and ALL.

The same type of approach was applied at Stanford by Brown and Botstein [22] in analysing the development of diffuse non-Hodgkin lymphoma. Using specific arrays containing up to 18 000 genes, they were able to observe distinctive traits between low-risk and relatively high-risk lymphomas. But blood cancers are not the only type of cancers that are amenable to comparative transcriptome analyses. Work done on melanomas [23–25] pointed out to the major role played by the Rho-C, Thymosin β-4 and fibronectin genes. High-level expression of these genes was found to be correlated with bad prognosis, namely conversion of local subcutaneous tumor cells into metastatic ones. Rho-C is a small GTPase, which functions as a mitogenic factor and is probably indirectly involved in cell adhesiveness. Thymosin β-4 exerts some direct effects on the acto-myosin containing filaments, making the active monomers more available to rapid polymerisation into lamellipodia. Fibronectin is a component of the extracellular matrix, the deposition of which favours the movement of the cell on a surface. Microarray studies have permitted to include these three components in the acquisition of the metastatic state by melanomas.

Much work has also been devoted, using similar approaches to the prognosis of cancers affecting breast (Fig. 3) [26], ovaries, lung and prostate with interesting results [27–29].