Differential display seems to be the technically simplest method. Its name stems from the end-point that is a comparison of a side-by-side display of the test and control preparations by electrophoresis. In its basic form, total RNA of the test and control samples is separately subjected to reverse transcription into cDNA that, in turn, is PCR-amplified using arbitrarily chosen primers. The products are applied to a gel electrophoresis and the band(s) that are specific for one of the preparations are cut from the gel, further amplified by PCR (using the same primers) and eventually sequenced 19.

In a more advanced version, mRNAs of the test and control cells are separately reverse transcribed to cDNA (Fig. 1). Each transcription is carried out in the presence of a oligo(dT) primers, directed to the poly(A) tail at the 3' terminus of the mRNA and constructed as 5'(NMT11)3' where N can be guanine (G), adenine (A), thymine (T), or cytosine (C), and M is G, A, or C 202122. The primers with G residues are superior to those having one C residue. Those ending in A or T are the least efficient. With use of an arbitrary decamer as the second primer, a PCR is carried out to amplify the transcript in order to obtain a sufficient working material. This is usually done in the presence of a radioactive nucleotide. Other methods are commonly used, such as silver staining. Amplified DNA fragments are separated on a denaturing polyacrylamide gel, the test preparation side by side with the control. Each band differing from those seen in the control electrophoresis is then used for sequencing, subcloning, or as a probe for cDNA library screening. Large amount of results can be obtained depending on the variation in N and M nucleotides. In spite of the basic simplicity of the procedure, the time and workload can be considerable, depending on the number of NM combinations tried.

The test mRNA is reverse transcribed into cDNA 23. This is hybridized with the mRNA of the control sample (Fig. 2). A portion of the test cDNA (corresponding to the activated gene) does not find any complementary part in the mRNA of the control sample and remains non-hybridized as a single-stranded cDNA (ss-cDNA). This can be isolated by chromatography on a hydroxylapatite column. The hybridization of the isolated ss-cDNA with control mRNA followed by another chromatography can be repeated to increase the purity of the isolated product 23. A cDNA library is produced and the subtracted sequence eventually identified. Alternatively, a second hybridization of the isolated ss-cDNA is carried out with the original test mRNA giving rise to a cDNA-mRNA hybrid which, after conversion to double stranded cDNA, is inserted into a vector, a cDNA library is constructed and several specific cDNA clones are isolated, leading to the identification of several genes 24.

In another variant 25, the test and control mRNAs are both reverse transcribed into cDNA. cDNA of the test sample is hybridized with cDNA of the control sample. The non-hybridized part of the test cDNA is a single-stranded DNA that is separated by hydroxylapatite. The single-stranded DNA is cloned into a vector to produce a subtracted library. Clones with a strong hybridization signal to the subtracted probe are selected and sequenced.

Isolation of a single-stranded test cDNA is not needed in this method. mRNAs of the test and control samples are prepared and each is reverse transcribed into cDNA. Each transcript is digested with the enzyme RsaI to obtain shorter, blunt-ended fragments. The test cDNA is divided into two portions (see Fig. 3). One of them is ligated with Adapter A, the second with adapter B. Each portion is hybridized with an excess of control cDNA. A mixture of hybridization products is formed (Fig. 3). A tiny fraction of cDNA remains unhybridized, single-stranded. This is a fragment that may be called specific, or differentially expressed, or subtracted. It originates from the gene that had been activated. It is absent in the control sample. This specific fragment is bound either to Adapter A or B in the two portions. In the second hybridization, the portions are mixed. After annealing, a small amount of the specific fragment is obtained double-stranded. It contains Adapter A on the one end and Adapter B on the other. After adding primers specific for the Adapters, the ends are filled and the specific fragment is amplified by PCR to make sure that sufficient amounts are available for a further processing. Cloning, sequencing and comparing with a gene database establish the identity of the gene(s) 2627 [http://www.clontech.com – "PCR-Select Subtraction kit"]. In contrast to the above methods, the primers for PCR amplification are clearly defined, avoiding thus problems with random primers. This method was used in a number of studies, such as the identification of genes upregulated in rats by E2 and progesterone treatment 28. A predecessor of this technique is the "representational difference analysis" 2930.

To describe the EST method, the following example is given. cDNA libraries were prepared by reverse transcription from mRNAs of the tissues to be examined 31. The libraries were converted to plasmids, transfected into Escherichia coli and plated. Hundreds of clones were picked at random. These were subjected to sequencing, followed by computer matching to known genes listed in the GenBank database. The average length of a sequence was 397 bases; ESTs longer than 150 bases were found to be most useful for similarity searches and mapping.

Subtractive hybridization (see above) was used to isolate the ESTs specific for one of the libraries. For example, a fibroblast cell line cDNA library was hybridized with a hippocampus library; the common sequences were removed and the specific hippocampus sequences remained. Using the EST method, more than 2000 human brain genes were identified 32.

The SAGE allows serial analysis of gene expression, an analysis of thousands of transcripts. It is based on the assumption that a short nucleotide sequence 10 base pairs (bp) – a tag – contains sufficient information to uniquely identify a transcript. In this respect SAGE differs from the EST approach.

The principle of SAGE is as follows: mRNA is reverse transcribed into cDNA with use of a biotinylated primer, the cDNA is cleaved with a restriction endonuclease and the 3' portions are then isolated by binding to streptavidin beads 33. In another version(http://www.invitrogen.com) (Fig. 4), mRNAs are captured prior to reverse transcription on oligo(dT) magnetic beads. Double stranded cDNAs are synthesized and digested with the restriction endonuclease NlaIII that cleaves most transcripts at least once. The part attached to the magnetic bead is further processed. The reaction mixture is divided into two portions. The portions are ligated via a restriction site R to an adapter A and B, respectively, each consisting of 40 bp. Taking advantage of the restriction sites R, both portions are cleaved with the restriction enzyme BsmFI in the distance of 14 bp. In this way "tags" are formed. Out of these 14 bp, 4 bp are a non-specific segment GTAC. These tags are blunt-ended with the Klenow fragment of DNA polymerase I. The two separate pools of tags are ligated together via a blunt-end ligation to produce "ditags". The ditags, flanked by the adapters A and B, are amplified by PCR with use of primers for A and B. The adapters are removed by the enzyme NlaIII and the ditags are concatenated. The resulting concatemers (a series of linked ditags) are cloned into a plasmid vector to create a SAGE library. Individual clones are then sequenced. SAGE is carried out for each sample to be compared.

Thanks to the concatenation, many tags can be detected in a single clone 33. As each tag is supposed to uniquely identify a transcript, SAGE can generate a comprehensive profile of gene expression. Indeed, many unique transcripts were identified with use of SAGE tags 34. The method is particularly useful for detecting genes of low level of expression or in rare tissues (e.g., early embryo) 3536. In addition, the amount of individual tags provides quantitative estimates of gene expression 37.

Still, the specificity of detection of genes with use of the short tags is not absolute. There are two main problems 38. The first one is that many SAGE tags have no match to known sequences in databases. These tags may represent so far unidentified genes, but their shortness makes it difficult to characterize the genes. The second problem is that the SAGE tags may find multiple matches in the databases 3940. Therefore, attempts have been made to increase the specificity by prolongation of the tags by various methods.

One such method is called GLGI (Generation of Longer cDNA fragments from SAGE tags for Gene Identification) 343840. The main feature of this method is the use of a SAGE tag as the sense primer for the PCR of a segment of cDNA. An anchored oligo(dT) serves as an antisense primer. In this manner a cDNA "tag" of up to several hundred bases is created. However, this method does not seem to improve the specificity of SAGE because even "non-specific" tags are co-amplified.

Better of seems to be another variant of SAGE, the LongSAGE 41. This is based on the use of tags 21 bp (out of which 4 represent a restriction site), tags longer than those in SAGE. The prolongation of tags is achieved by the use of the restriction endonuclease MmeI. The longer tags increase the power of identification of genes, while not diminishing the sensitivity of SAGE given by the use of PCR and concatenation. Theoretical calculations showed that >99.8% of the 21 bp tags were expected to occur only once in a genome.

SAGE was used for the investigation of differences in gene expression in various health conditions. In the studies of breast tumors 37, global gene expression profiles in breast carcinoma cells were compared with those in normal mammary epithelial cells. The patterns of gene clusters in normal tissue were distinctly different from those of tumors of different stage and histological grade. The most dramatic change occurred at the normal-to-in situ carcinoma transition. This change can be an important marker for an early diagnosis. In another study, several genes regulated by estrogen or tamoxifen were identified in an estrogen-dependent breast cancer cell line. One of them was studied closer. It appeared to play a significant role in estrogen-promoted cell growth 42.