The present invention provides a fast, simple and direct covalent bond formation between two strands of nucleotide sequences. Non-modified first strand nucleotide sequences are hybridized with second strand nucleotide sequences, of which certain specific base structure(s) is modified by chemical reagents in order to generate covalent bonding with the first strand. While the hybridization of these two strand nucleotide sequences generates double-stranded hybrid duplexes between their homologues, covalent bond formation occurs in the region of modified base-pairs. Since neither a polymerase chain restriction nor a restriction enzyme digestion can be performed with the covalently bonded hybrid duplexes, the present invention can be used to subtract common sequences during subtractive hybridization, to inhibit nonspecific contamination during subcloning and to increase binding stability of antisense probes during in situ hybridization as well as gene therapy.