Bacteria have the ability to restrict bacteriophage (virus) infection by cutting up the 'phage DNA once it's injected into the cell. The enzyme that cuts DNA is called a restriction enzyme, or more properly, a restriction endonuclease.
Endonucleases are enzymes that cleave DNA internally by binding to the middle of a DNA strand and breaking one of the linkages that join the nucleotides. In the case of restriction endonucleases, both strands of the double helix are cut thus breaking apart the 'phage DNA before it can make new proteins and new virus particles.
The enzymes don't cut randomly. They bind to specific sequences and only cut at those sites. An example is one of the restriction endonucleases from Escherichia coli called EcoR1. It binds to the sequence GAATTC.
The top figure shows EcoR1 bound to DNA. Notice that the restriction site is palindromic—it reads the same way in opposite directions on opposite strands (see below). In order to appreciate this you need to understand that the two strands of DNA run in opposite directions and the direction of reading is important. We always read a DNA strand in the 5ʹ to 3ʹ direction. Thus, on the top strand the sequence is GAATTC while on the bottom strand the sequence is also GAATTC but read in the opposite direction.
The two identical subunits of EcoR1 (blue and purple) bind to opposite strands of the double helix and cut at exactly the same spot; in this case between the G and the A. The DNA is chopped in two.
The sequence GAATTC will occur, on average, once every 4096 base pairs (46). This means there's a good chance of cutting any bacteriophage DNA that enters the cell since most bacteriophage genomes are much larger than 4096 base pairs.
This is an effective way of restricting bacteriophage infection except for one minor problem. How does the bacterium prevent it's own DNA from being cut by the restriction endonucleases?
The secret lies in blocking the restriction site in host DNA so that the restriction enzyme doesn't recognize it. One of the nucleotides is modified by an methylase enzyme (modification enzyme) that attaches a methyl group to one of the bases. The restriction enzyme doesn't bind to sites where one of the bases is modified by methylation; so all you have to do is make a methylase enzyme that recognizes the same site as the restriction enzyme.
EcorR1 methylase binds to the sequence GAATTC and methylates the first A to form N6-methyladenine (see Monday's Molecule #12). Like the endonuclease, the methylase has two subunits that bind symmetrically to double-stranded DNA. Both of the A's on opposite strands are methylated so neither strand can be recognized by EcoR1 and neither strand will be cut.
So far, so good, but we still have a problem. Why isn't the 'phage DNA methylated as well?
This is the cool part. Look at the figure on the left. Imagine that both strands are methylated. Following DNA replication, the sequence GAATTC will be copied but the newly synthesized strand isn't methylated. The hemimethylated DNA won't be recognized by the restriction enzyme so it's in no danger of being cut. In a very short time the methylase will bind and methylate the new strand of DNA.
The methylase binds specifically to hemimethylated DNA and not to unmethylated DNA. Thus, it will keep the host DNA fully methylated but it won't methylate the incoming 'phage DNA since the 'phage DNA is completely unmethylated.
As bacteria grow and divide they continue to inherit DNA that's methylated at the restriction site even though this inheritance isn't your typical genetic inheritance. It's called epigenetic inheritance. If something happens to the methylase, the cell will commit suicide by chopping up its own DNA. If both the methylase and restriction enzyme genes are mutated the cell survives quite nicely except that it's more susceptible to bacteriophage infection.
Epigenetic inheritance is common in mammals, including us. In this case it's not related to restriction/modification. It's a separate phenomenon where gene expression is controlled by the presence or absence of 5-methyl cytosine. The cytosine methylase works just like the modification enzymes—it binds preferentially to hemimethylated DNA. In this way, methylated regions of DNA are inherited from one generation to another.