Restriction-modification is a term for bacterial enzyme systems that cleave DNA sequences. Each system consists of two distinct enzyme activities: a DNA methylase and an endonuclease that catalyzes the double-strand DNA break. Type II restriction endonucleases have been of great value to research because most of them cut within the recognition sequence, making cleavage absolutely sequence specific. A divalent cation is required for cleavage, but ATP is not required. Each type II nuclease has a counterpart methylase, which binds to the same recognition sequence and methylates one nucleotide within that sequence. A hemimethylated DNA (i.e., DNA with a methyl group on one strand only) is a preferred substrate for the methylase but not for the nuclease, which generally cleaves only when the recognition site is unmethylated on both strands. Cleavage almost always creates 3' hydroxyl and 5' phosphate termini. Cleavage sites on the two strands may be offset by as much as four nucleotides (as in EcoRI), giving cuts with short, self-complementary, single-strand termini. Some enzymes cleave to give a 5'-terminated single-strand end ("overhang"), whereas others generate a 3' overhang. Other type II nucleases, including SmaI and HindII, generate blunt-ended fragments, in which the cutting sites are not offset. Most recognition sites are four, five, or six nucleotides in length, although a few type II enzymes recognize an eight-nucleotide sequence. Most show 2-fold rotational sequence symmetry, suggesting that the two enzyme subunits are also arranged symmetrically. Table 25.2 shows the recognition sites for several widely used type II nucleases. Several hundred enzymes of this type have now been isolated. Some type II nucleases, such as HindII, recognize multiple specific sequences, and some enzymes (such as HgaI) cleave at a site outside the recognition sequence.

Figure 25.7 shows one polypeptide subunit of the dimeric EcoRI restriction enzyme in contact with its DNA recognition sequence. The DNA is bound in a cleft of the enzyme, and the protein has an N-terminal "arm" that wraps about the DNA. Sequence specificity is maintained by 12 hydrogen bonds, which link the purine residues in the site to a glutamate and two arginine residues (not shown in the figure). The other subunit (not shown in the figure) contacts the substrate identically, accounting for the ability of the enzyme to catalyze symmetrical cleavages within the cutting site.