9.3 Restriction Enzymes Catalyze Highly Specific DNA-Cleavage Reactions

We next consider a hydrolytic reaction that results in the cleavage of DNA. Bacteria and archaea have evolved mechanisms to protect themselves from viral infections. Many viruses inject their DNA genomes into cells; once inside, the viral DNA hijacks the cell’s machinery to drive the production of viral proteins and, eventually, of progeny virus. Often, a viral infection results in the death of the host cell. A major protective strategy for the host is to use restriction endonucleases (restriction enzymes) to degrade the viral DNA on its introduction into a cell. These enzymes recognize particular base sequences, called recognition sequences or recognition sites, in their target DNA and cleave that DNA at defined positions. We have already considered the utility of these important enzymes for dissecting genes and genomes (Section 5.1). The most well-studied class of restriction enzymes comprises the type II restriction enzymes, which cleave DNA within their recognition sequences. Other types of restriction enzymes cleave DNA at positions somewhat distant from their recognition sites.

Restriction endonucleases must show tremendous specificity at two levels. First, they must not degrade host DNA containing the recognition sequences. Second, they must cleave only DNA molecules that contain recognition sites (hereafter referred to as cognate DNA) without cleaving DNA molecules that lack these sites. How do these enzymes manage to degrade viral DNA while sparing their own? In E. coli, the restriction endonuclease EcoRV cleaves double-stranded viral DNA molecules that contain the sequence 5′-GATATC-3′ but leaves intact host DNA containing hundreds of such sequences. We shall return to the strategy by which host cells protect their own DNA at the end of this section.

Restriction enzymes must cleave DNA only at recognition sites, without cleaving at other sites. Suppose that a recognition sequence is six base pairs long. Because there are 46, or 4096, sequences having six base pairs, the concentration of sites that must not be cleaved will be approximately 4000-fold higher than the concentration of sites that should be cleaved. Thus, to keep from damaging host-cell DNA, restriction enzymes must cleave cognate DNA molecules much more than 4000 times as efficiently as they cleave nonspecific sites. We shall return to the mechanism used to achieve the necessary high specificity after considering the chemistry of the cleavage process.

Cleavage is by in-line displacement of 3′-oxygen from phosphorus by magnesium-activated water

A restriction endonuclease catalyzes the hydrolysis of the phosphodiester backbone of DNA. Specifically, the bond between the 3′-oxygen atom and the phosphorus atom is broken. The products of this reaction are DNA strands with a free 3′-hydroxyl group and a 5′-phosphoryl group at the cleavage site (Figure 9.31). This reaction proceeds by nucleophilic attack at the phosphorus atom. We will consider two alternative mechanisms, suggested by analogy with the proteases. The restriction endonuclease might cleave DNA by mechanism 1 through a covalent intermediate, employing a potent nucleophile (Nu), or by mechanism 2 through direct hydrolysis:

270

Mechanism 1 (covalent intermediate)

Mechanism 2 (direct hydrolysis)

Each mechanism postulates a different nucleophile to attack the phosphorus atom. In either case, each reaction takes place by in-line displacement:

The incoming nucleophile attacks the phosphorus atom, and a pentacoordinate transition state is formed. This species has a trigonal bipyramidal geometry centered at the phosphorus atom, with the incoming nucleophile at one apex of the two pyramids and the group that is displaced (the leaving group, L) at the other apex. Note that the displacement inverts the stereochemical conformation at the tetrahedral phosphorous atom, analogous to the interconversion of the R and S configurations around a tetrahedral carbon center (Section 2.1).

Figure 9.31: Hydrolysis of a phosphodiester bond. All restriction enzymes catalyze the hydrolysis of DNA phosphodiester bonds, leaving a phosphoryl group attached to the 5’ end. The bond that is cleaved is shown in red.

The two mechanisms differ in the number of times that the displacement takes place in the course of the reaction. In the first type of mechanism, a nucleophile in the enzyme (analogous to serine 195 in chymotrypsin) attacks the phosphate group to form a covalent intermediate. In a second step, this intermediate is hydrolyzed to produce the final products. In this case, two displacement reactions take place at the phosphorus atom. Consequently, the stereochemical configuration at the phosphorus atom would be inverted and then inverted again, and the overall configuration would be retained. In the second type of mechanism, analogous to that used by the aspartyl- and metalloproteases, an activated water molecule attacks the phosphorus atom directly. In this mechanism, a single displacement reaction takes place at the phosphorus atom. Hence, the stereochemical configuration at the phosphorus atom is inverted after cleavage. To determine which mechanism is correct, we examine the stereochemistry at the phosphorus atom after cleavage.

A difficulty is that the stereochemistry is not easily observed, because two of the groups bound to the phosphorus atom are simple oxygen atoms, identical with each other. This difficulty can be circumvented by replacing one oxygen atom with sulfur (producing a species called a phosphorothioate). Let us consider EcoRV endonuclease. This enzyme cleaves the phosphodiester bond between the T and the A at the center of the recognition sequence 5′-GATATC-3′. The first step is to synthesize an appropriate substrate for EcoRV containing phosphorothioates at the sites of cleavage (Figure 9.32). The reaction is then performed in water that has been greatly enriched in18O to allow the incoming oxygen atom to be marked. The location of the18O label with respect to the sulfur atom indicates whether the reaction proceeds with inversion or retention of stereochemistry. This experiment revealed that the stereochemical configuration at the phosphorus atom was inverted only once with cleavage. This result is consistent with a direct attack by water at the phosphorus atom and rules out the formation of any covalently bound intermediate (Figure 9.33).

Figure 9.32: Labeling with phosphorothioates. Phosphorothioate groups, in which one of the nonbridging oxygen atoms is replaced by a sulfur atom, can be used to label specific sites in the DNA backbone to determine the overall stereochemical course of a displacement reaction. Here, a phosphorothioate is placed at sites that can be cleaved by EcoRV endonuclease.
Figure 9.33: Stereochemistry of cleaved DNA. Cleavage of DNA by EcoRV endonuclease results in overall inversion of the stereochemical configuration at the phosphorus atom, as indicated by the stereochemistry of the phosphorus atom bound to one bridging oxygen atom, one16O, one18O, and one sulfur atom. Two possible products are shown, only one of which is observed, indicating direct attack of water at the phosphorous atom.

271

Restriction enzymes require magnesium for catalytic activity

Many enzymes that act on phosphate-containing substrates require Mg2+ or some other similar divalent cation for activity. One or more Mg2+ (or similar) cations are essential to the function of restriction endonucleases. What are the functions of these metal ions?

Direct visualization of the complex between EcoRV endonuclease and cognate DNA molecules in the presence of Mg2+ by crystallization has not been possible, because the enzyme cleaves the substrate under these circumstances. Nonetheless, metal ion complexes can be visualized through several approaches. In one approach, crystals of EcoRV endonuclease are prepared bound to oligonucleotides that contain the enzyme’s recognition sequence. These crystals are grown in the absence of magnesium to prevent cleavage; after their preparation, the crystals are soaked in solutions containing the metal. Alternatively, crystals have been grown with the use of a mutated form of the enzyme that is less active. Finally, Mg2+ can be replaced by metal ions such as Ca2+ that bind but do not result in much catalytic activity. In all cases, no cleavage takes place, and so the locations of the metal ion-binding sites are readily determined.

272

As many as three metal ions have been found to be present per active site. The roles of these multiple metal ions are still under investigation. One ion-binding site is occupied in essentially all structures. This metal ion is coordinated to the protein through two aspartate residues and to one of the phosphate-group oxygen atoms near the site of cleavage. This metal ion binds the water molecule that attacks the phosphorus atom, helping to position and activate it in a manner similar to that for the Zn2+ ion of carbonic anhydrase (Figure 9.34).

Figure 9.34: A magnesium ion-binding site in EcoRV endonuclease. The magnesium ion helps to activate a water molecule and positions it so that it can attack the phosphorus atom.

The complete catalytic apparatus is assembled only within complexes of cognate DNA molecules, ensuring specificity

We now return to the question of specificity, the defining feature of restriction enzymes. The recognition sequences for most restriction endonucleases are inverted repeats. This arrangement gives the three-dimensional structure of the recognition site a twofold rotational symmetry (Figure 9.35).

Figure 9.35: Structure of the recognition site of EcoRV endonuclease. (A) The sequence of the recognition site, which is symmetric around the axis of rotation designated in green. (B) The inverted repeat within the recognition sequence of EcoRV (and most other restriction endonucleases) endows the DNA site with twofold rotational symmetry.

The restriction enzymes display a corresponding symmetry: they are dimers whose two subunits are related by twofold rotational symmetry. The matching symmetry of the recognition sequence and the enzyme facilitates the recognition of cognate DNA by the enzyme. This similarity in structure has been confirmed by the determination of the structure of the complex between EcoRV endonuclease and DNA fragments containing its recognition sequence (Figure 9.36). The enzyme surrounds the DNA in a tight embrace.

Figure 9.36: EcoRV embracing a cognate DNA molecule. (A) This view of the structure of EcoRV endonuclease bound to a cognate DNA fragment is down the helical axis of the DNA. The two protein subunits are in yellow and blue, and the DNA backbone is in red. Notice that the twofold axes of the enzyme dimer and the DNA are aligned. One of the DNA-binding loops (in green) of EcoRV endonuclease is shown interacting with the base pairs of its cognate DNA-binding site. Key amino acid residues are shown hydrogen bonding with (B) a CG base pair and (C) an AT base pair.
[Drawn from 1RVB.pdb.]

273

Figure 9.37: Distortion of the recognition site. The DNA is represented as a ball-and-stick model. The path of the DNA helical axis, shown in red, is substantially distorted on binding to the enzyme. For the B form of DNA, the axis is straight (not shown).

An enzyme’s binding affinity for substrates often determines specificity. Surprisingly, however, binding studies performed in the absence of magnesium have demonstrated that the EcoRV endonuclease binds to all sequences, both cognate and noncognate, with approximately equal affinity. Why, then, does the enzyme cleave only cognate sequences? The answer lies in a unique set of interactions between the enzyme and a cognate DNA sequence.

Within the 5′-GATATC-3′ sequence, the G and A bases at the 5′ end of each strand and their Watson–Crick partners directly contact the enzyme by hydrogen bonding with residues that are located in two loops, one projecting from the surface of each enzyme subunit (Figure 9.36). The most striking feature of this complex is the distortion of the DNA, which is substantially kinked in the center (Figure 9.37). The central two TA base pairs in the recognition sequence play a key role in producing the kink. They do not make contact with the enzyme but appear to be required because of their ease of distortion. The 5′-TA-3′ sequence is known to be among the most easily deformed base pairs.

Figure 9.38: Nonspecific and cognate DNA within EcoRV endonuclease. A comparison of the positions of the nonspecific (orange) and the cognate DNA (red) within EcoRV. Notice that, in the nonspecific complex, the DNA backbone is too far from the enzyme to complete the magnesium ion-binding sites.
[Drawn from 1RVB.pdb.]

The structures of complexes formed with noncognate DNA fragments are strikingly different from those formed with cognate DNA: the noncognate DNA conformation is not substantially distorted (Figure 9.38). This lack of distortion has important consequences with regard to catalysis. No phosphate is positioned sufficiently close to the active-site aspartate residues to complete a magnesium ion-binding site (Figure 9.34). Hence, the nonspecific complexes do not bind the magnesium ions and the complete catalytic apparatus is never assembled. The distortion of the substrate and the subsequent binding of the magnesium ion account for the catalytic specificity of more than a millionfold that is observed for EcoRV endonuclease. Thus, enzyme specificity may be determined by the specificity of enzyme action rather than the specificity of substrate binding.

274

We can now see the role of binding energy in this strategy for attaining catalytic specificity. The distorted DNA makes additional contacts with the enzyme, increasing the binding energy. However, the increase in binding energy is canceled by the energetic cost of distorting the DNA from its relaxed conformation (Figure 9.39). Thus, for EcoRV endonuclease, there is little difference in binding affinity for cognate and nonspecific DNA fragments. However, the distortion in the cognate complex dramatically affects catalysis by completing the magnesium ion-binding site. This example illustrates how enzymes can utilize available binding energy to deform substrates and poise them for chemical transformation. Interactions that take place within the distorted substrate complex stabilize the transition state leading to DNA hydrolysis.

Figure 9.39: Greater binding energy of EcoRV endonuclease bound to cognate versus noncognate DNA. The additional interactions between EcoRV endonuclease and cognate DNA increase the binding energy, which can be used to drive DNA distortions necessary for forming a catalytically competent complex.

Host-cell DNA is protected by the addition of methyl groups to specific bases

How does a host cell harboring a restriction enzyme protect its own DNA? The host DNA is methylated on specific adenine bases within host recognition sequences by other enzymes called methylases (Figure 9.40). An endonuclease will not cleave DNA if its recognition sequence is methylated. For each restriction endonuclease, the host cell produces a corresponding methylase that marks the host DNA at the appropriate methylation site. These pairs of enzymes are referred to as restriction-modification systems.

Figure 9.40: Protection by methylation. The recognition sequence for EcoRV endonuclease (left) and the sites of methylation (right) in DNA protected from the catalytic action of the enzyme.

The distortion in the DNA explains how methylation blocks catalysis and protects host-cell DNA. The host E. coli adds a methyl group to the amino group of the adenine nucleotide at the 5′ end of the recognition sequence. The presence of the methyl group blocks the formation of a hydrogen bond between the amino group and the side-chain carbonyl group of asparagine 185 (Figure 9.41). This asparagine residue is closely linked to the other amino acids that form specific contacts with the DNA. The absence of the hydrogen bond disrupts other interactions between the enzyme and the DNA substrate, and the distortion necessary for cleavage will not take place.

Figure 9.41: Methylation of adenine. The methylation of adenine blocks the formation of hydrogen bonds between EcoRV endonuclease and cognate DNA molecules and prevents their hydrolysis.

275

Type II restriction enzymes have a catalytic core in common and are probably related by horizontal gene transfer

Type II restriction enzymes are prevalent in Archaea and Bacteria. What can we tell of the evolutionary history of these enzymes? Comparison of the amino acid sequences of a variety of type II restriction endonucleases did not reveal significant sequence similarity between most pairs of enzymes. However, a careful examination of three-dimensional structures, taking into account the location of the active sites, revealed the presence of a core structure conserved in the different enzymes.

These observations indicate that many type II restriction enzymes are indeed evolutionarily related. Analyses of the sequences in greater detail suggest that bacteria may have obtained genes encoding these enzymes from other species by horizontal gene transfer, the passing of pieces of DNA (such as plasmids) between species that provide a selective advantage in a particular environment. For example, EcoRI (from E. coli) and RsrI (from Rhodobacter sphaeroides) are 50% identical in sequence over 266 amino acids, clearly indicative of a close evolutionary relationship. However, these species of bacteria are not closely related. Thus, these species appear to have obtained the gene for these restriction endonucleases from a common source more recently than the time of their evolutionary divergence. Moreover, the codons used by the gene encoding EcoRI endonuclease to specify given amino acids are strikingly different from the codons used by most E. coli genes, which suggests that the gene did not originate in E. coli.

Horizontal gene transfer may be a common event. For example, genes that inactivate antibiotics are often transferred, leading to the transmission of antibiotic resistance from one species to another. For restriction-modification systems, protection against viral infections may have favored horizontal gene transfer.