Recognition sequence and DNA ends	Take a look at the following examples of DNA restriction enzyme sequences: Kas I (G^GCGCC) Nar I (GG^CGCC) Ehe I (GGC^GCC) Bbe I (GGCGC^C) You see that the same sequence is recognized by four isoschizomers that break the phosphodiester backbone differently. The ends generated by these four would consequently be different: Kas I NNNNNG GCGCCNNNNNN NNNNNCCGCG GNNNNNN Nar I NNNNNGG CGCCNNNNNN NNNNNCCGC GGNNNNNN Ehe I NNNNNGGC GCCNNNNNN NNNNNCCG CGGNNNNNN Bbe I NNNNNGGCGC CNNNNNN NNNNNC CGCGGNNNNNN
	The point here is that enzymes leave different types of DNA ends, and this is a matter that is independent of recognition sequence. In the example above, a digestion product using Kas I would not be compatible with a digestion product of Nar I, because they could not hydrogen bond. Kas I NNNNNG CGCCNNNNNN Nar I NNNNNCCGCG GGNNNNNN Bbe I has a GCGC-3' overhanging end, and similarly it cannot anneal to any of the other three examples. It does have ends that are compatible with ends generated by Hae II (RGCGC^Y) however. Here is an example: Hae II NNNNNAGCGC CNNNNNN Nar I NNNNNT CGCGGNNNNNN In this situation, the ends would match perfectly and the phosphodiester bonds could be sealed with the enzyme T4 DNA ligase.
Blunt ends: the great equalizer	Blunt ends are always compatible with each other, because there are no H-bonds being formed that would define compatibility or incompatibility. So, a DNA end generated by Ehe I is compatible with a DNA end generated by EcoRV (GAT^ATC): Ehe I NNNNNGGC ATCNNNNNN EcoRV NNNNNCCG TAGNNNNNN This is a mixed blessing, for while the ends will always fit together there is a lack of specificity in assembly. Having cohesive ends gives better control of the assembly process because you can force the DNA fragment to be inserted in a single orientation. For example: BamHI BamHI EcoRI EcoRI NNNNNNG GATCCNNNNNNNG AATTCNNNNNNNNN NNNNNNCCTAG GNNNNNNNCTTAA GNNNNNNNNN In this example, the green DNA fragment (center) can only be inserted with the BamHI site on the left and EcoRI site on the right. This is called forced cloning, and it is not possible when the ends are blunt.
	We can make a cohesive end into a blunt end using DNA polymerases such as Klenow (fragment of E. coli DNA polymerase I), T4 DNA polymerase, or Pfu polymerase. Let's review: In this case, a 5' overhanging end is being filled in with newly-synthesized DNA. Restriction enzymes typically leave small overhanging ends, and they are usually of the 5' overhanging type. Bam HI NNNNNNG GATCCNNNNN NNNNNNCCTAG GNNNNN Fill in one G nucleotide, and you have: NNNNNNGG GATCCNNNNN NNNNNNCCTAG GGNNNNN Fill in the next A nucleotide, and you have: NNNNNNGGA GATCCNNNNN NNNNNNCCTAG AGGNNNNN Then the next T nucleotide, and you have: NNNNNNGGAT GATCCNNNNN NNNNNNCCTAG TAGGNNNNN Finally the next C nucleotide, and you have a blunt end: NNNNNNGGATC GATCCNNNNN NNNNNNCCTAG CTAGGNNNNN Now the enzyme cannot add additional nucleotides to the 3' end because it requires a template: If there is a 3' overhanging end, then the 3' to 5' exonuclease removes it, leaving a blunt end also. Here are some examples of what the enzymes mentioned earlier (Klenow, T4 DNA polymerase, or Pfu) would do to ends left by the four restriction enzymes mentioned earlier: Kas I NNNNNG GCGCCNNNNNN NNNNNCCGCG GNNNNNN ...would be filled in to yield: NNNNNGGCGC GCGCCNNNNNN NNNNNCCGCG CGCGGNNNNNN Nar I NNNNNGG CGCCNNNNNN NNNNNCCGC GGNNNNNN ...would be filled in to yield: NNNNNGGCG CGCCNNNNNN NNNNNCCGC GCGGNNNNNN Ehe I NNNNNGGC GCCNNNNNN NNNNNCCG CGGNNNNNN ...would be unchanged Bbe I NNNNNGGCGC CNNNNNN NNNNNC CGCGGNNNNNN ...would be subject to the 3'-5' exo, leaving: NNNNNG CNNNNNN NNNNNC GNNNNNN Having modified the DNA ends left by these four enzymes, all are now mutually compatible, and would be compatible with other blunt ends. Note that where modifications have taken place, the enzyme site is generally destroyed upon religation. Sometimes that's exactly what you want.
Putting it together - the right way.	We use the enzyme T4 DNA ligase to make covalent connections in the phosphodiester backbone. It was indicated that 5' ends of DNA usually have a phosphate group, and we know that the phosphate group is required for ligase activity (as is ATP as a source of energy). We've also already discussed an enzyme (T4 polynucleotide kinase) that can be used to add a 5' phosphate where one is lacking, for example on a PCR oligonucleotide primer. When DNA is treated with the enzyme alkaline phosphatase, the 5' phosphate groups are removed. Activity of alkaline phosphatase - removal of 5' phosphates Here's a nice application: If a linearized vector is dephosphorylated in this way, it cannot reclose upon itself because the enzyme T4 DNA ligase requires that a 5' phosphate group be present. A DNA fragment that has 5' phosphates still present can form a bridge between the dephosphorylated ends, so insertions are favored! When you are trying to combine two molecules, this removal of 5' phosphates from the vector (alone) keeps it from reclosing on itself and spoiling the construction. Preventing reclosures by use of a dephosphorylated vector What you get in the end: There are two widely separated nicks in the final product, because two of the four ligation events were prevented by the lack of 5' phosphates. Still, two out of four is good enough! The bacteria will fix the remaining nicks after the DNA is transformed. Two sources of alkaline phosphatase are commonly used for this work: Calf intestinal alkaline phosphatase (CIAP, or CIP) Shrimp alkaline phosphatase (SAP) The shrimp alkaline phosphatase is heat sensitive (it is derived from an Arctic shrimp that loves the cold!), so the enzyme can easily be inactivated at a moderately high temperature. The calf intestinal alkaline phosphatase is relatively stable, so it must be inactivated at higher temperature, or via digestion with proteinase-K enzyme. Why is it so important to inactivate the alkaline phosphatase enzyme? Because if it contaminates your ligation reaction, it will strip the 5' phosphates off of the DNA insert as well. That will block all ligation events, including the ones you want!