AlkD's Novel Approach to DNA Repair
By: Carol A. Rouzer, VICB Communications
Published: October 29, 2010
The Eichman lab’s structures of the DNA glycosylase AlkD reveals a totally new mechanism by which the enzyme recognizes and removes damaged DNA bases.
Cellular DNA is subject to damage from a multitude of chemical and physical insults. Among these are methylating agents, which are present as environmental toxins or produced endogenously during normal metabolic processes. In addition, some commonly used anti-cancer drugs are DNA methylating agents. Methylation of DNA results in lesions such as N3-methyladenine (3mA, Figure 1) and N7-methylguanine (7mG), both of which are toxic and may lead to mutagenic outcomes. In order to prevent such outcomes, cells contain DNA glycosylases, which hydrolyze the glycosidic bond between the altered base and the deoxyribose, leaving an abasic site to be repaired by other enzymes. A large number of DNA glycosylases exist, each designed to recognize and remove a specific lesion or group of lesions. The enzymes that repair 3mA are of particular interest to Vanderbilt Institute of Chemical Biology member Brandt Eichman and his laboratory. Realizing the importance of these enzymes to the maintenance of DNA in healthy cells, and to reversing the effects of cancer chemotherapy, the Eichman lab set out to determine exactly how these enzymes work. Their model enzyme was the 3mA-specific glycosylase from Bacillus cereus, AlkD [E. Rubinson et al. (2010) Nature, published online October 3, DOI: 10.1038/nature09428].
Figure 1. Structures of 3mA, 3d3mA, and THF.
Our current knowledge of DNA glycosylases indicates that most work by flipping the damaged base out of its position in the DNA double helix and into an active site pocket. The enzyme stabilizes the disrupted DNA structure by inserting the side chain of an amino acid into the void left by the extruded base. Often, an active site residue acts as a general base to promote glycosidic bond cleavage. DNA glycosylases specific for 3mA, however, do not have an active site residue that acts as a general base. These enzymes may rely on the fact that 3mA’s positive charge makes its glycosidic bond highly susceptible to hydrolysis even in the absence of a catalytic active site residue. To explore this possibility, the Eichman lab determined the crystal structure of AlkD complexed to a double stranded oligonucleotide containing 3-deaza-3-methyladenine (3d3mA, Figure 1) as a surrogate for 3mA. The substitution of carbon for nitrogen at position 3 of the purine ring in 3d3mA produced a stable analog with similar structural characteristics to 3mA, so its complex with AlkD should reveal how AlkD interacts with a substrate containing 3mA. Then, to learn how cleavage of the glycosidic bond alters the AlkD-DNA interaction, the Eichman lab obtained the crystal structure of a complex of AlkD with an oligonucleotide bearing tetrahydrofuran (THF, Figure 1) as a surrogate for the product abasic site.
Figure 2. Stylized represntation of the structure of AlkD. Each colored cylinder represents a stretch of alpha helix. the five heat repeats are indicated by the different colors (orange, yellow, green, blue, and violet). The helical DNA substrate is shown interacting with the concave surface formed by these heat repeats. Figure curtesy of Brandt Eichmann, copyright 2010.
Analysis of the crystal structure data revealed some exciting surprises. AlkD is made up of 5 pairs of alpha helical segments called HEAT repeats that form a positively charged concave surface (Figure 2). HEAT repeats are not usually involved in protein-DNA interactions or catalytic activity, but in both crystal structures, the concave surface generated by these helical segments was the point of contact with the oligonucleotide. Half of the DNA double helix was surrounded by the C-shaped protein over a 10 base-pair region. Unexpectedly, the DNA strand in contact with the protein was NOT the strand containing the lesion, so the lesion itself actually faced away from the enzyme, making it accessible to solvent. In the case of the 3d3mA structure, the 3d3mA lesion remained stacked in the DNA helix, but the complementary thymine (dT) was displaced into the minor groove of the DNA. Although oriented next to the enzyme, the displaced dT had no direct interactions with the protein. Rather, it was held in place by a distortion of the DNA helix, which was secured by a hydrogen bond network between the deoxyribose backbone of the DNA and Asp 113, Arg 148, and Arg 190 of AlkD along with Van der Waals interactions between the DNA backbone and AlkD’s Trp 109 and Trp 187 residues. In the case of the THF structure, the DNA helix was even more distorted, with the THF moiety rotated completely out of position into the major groove and the complementary dT shifted even further into the minor groove.
Figure 3. Structures of the DNA substrate complexed to AlkD. The enzyme induces a distortion of the DNA helix that causes the lesion (violet) to be displaced out of the helix, exposing it to solvent, and the opposing base (blue) to be displaced into the minor groove, facing the protein. Figure curtesy of Brandt Eichmann, copyright 2010.
These findings suggested that AlkD can not possibly work by the mechanisms used by most DNA glycosylases. The Eichman lab postulated that the protein recognizes and cleaves the glycosidic bond of 3mA without actually interacting directly with the lesion. Rather, AlkD identifies lesions on the basis of instability of the helix that results from disruption of Watson-Crick base pairing. This instability leaves the region around the lesion susceptible to distortion upon binding to the enzyme. Once AlkD places the DNA in the distorted conformation observed in the crystal structure, the alkylated base is exposed to solvent, rendering the unstable glycosidic bond susceptible to rapid hydrolysis (Figure 3). If this hypothesis is correct, then AlkD should be able to cleave a wide variety of lesions, even bulky ones, since they would not have to fit neatly into an active site pocket. However, the lesion would have to be charged so that the glycosidic bond would be easily cleaved upon solvent exposure. This prediction proved to be true as indicated by AlkD’s ability to hydrolyze bulky pyridyloxobutyl (POB) adducts, as long as they carried a charge (Figure 4). The hypothesis also predicts that AlkD should recognize any lesion that causes instability in the DNA double helix. The Eichman group showed that AlkD could capture oligonucleotides bearing a dG-dT mismatch, binding it in a structure nearly identical to that observed for the 3d3mA-containing oligonucleotide. Thus, the enzyme could recognize the unstable helical structure introduced by the mismatch, even though neither base was alkylated.
Figure 4. Structures of pyridyloxobutyl (POB) adducts of deoxyguanine. The N7 adduct is charged and is cleaved by AlkD. The O6 adduct is not charged, and is not cleaved by the enzyme.
The results indicate that AlkD recognizes and cleaves alkylated nucleobases by a totally novel mechanism. The mechanism capitalizes on intrinsic properties of the lesion, base mispairing and glycosidic bond instability, but does not depend on the specific geometry of the lesion. This allows AlkD to find and remove a wide variety of alkylated bases while guaranteeing that it does not cleave bases from native DNA. Knowledge of how AlkD works provides a new foundation for understanding how alkylation of DNA leads to mutations and cancer, and how better alkylating agents can be designed for cancer chemotherapy.