Computational procedures to explain the different biological activity of DNA/DNA, DNA/PNA and PNA/PNA hybrid molecules mimicking NF-κB binding sites

Peptide nucleic acids (PNA) have recently been proposed as alternative reagents in experiments aimed to the control of gene expression. In PNAs, the pseudopeptide backbone is composed of N-(2-aminoethyl)glycine units and therefore is stable in human serum and cellular extracts. PNAs hybridize with high affinity to complementary sequences of single-stranded RNA and DNA, forming Watson-Crick double helices and giving rise to highly stable (PNA)2-RNA triplexes with RNA targets. Therefore, antisense and antigéne PNAs have been synthetized and characterized. The major issue of the present paper is to describe some computational procedures useful to compare the behaviour of PNA double stranded molecules and PNA/DNA hybrids with the behaviour of regular DNA duplexes in generating complexes with DNA-binding proteins. The performed computational analyses clearly allow to predict that the lack of charged phosphate groups and the different shape of helix play a critical role in the binding efficiency of NF-κB transcription factors. These computational analyses are in agreement with competitive gel shift and UV-cross linking experiments. These experiments demonstrate that NF-κB PNA/PNA hybrids do not interact efficiently with proteins recognizing the NF-κB binding sites in genomic sequences. In addition, the data obtained indicate that the same NF-κB binding proteins recognize both the NF-κB DNA/PNA and DNA/DNA hybrids, but the molecular complexes generated with NF-κB DNA/PNA hybrids are less stable than those generated with NF-κB target DNA/DNA molecules.