Vanderbilt Institute of Chemical Biology



Discovery at the VICB







Fresh Insight into the Mechanisms of DNA Repair


By: Carol A. Rouzer, VICB Communications
Published:  November 1, 2015


Structural studies of the DNA glycosylase AlkD reveal a novel mechanism of action and a new role for base excision repair in the removal of bulky DNA lesions.


The genome is under constant attack by both chemical and physical insults that can damage DNA, leading to potentially deleterious or even lethal mutations. To survive, cells have evolved multiple DNA repair pathways, tailored to the widely varying forms of damage that can occur. One of the most important of these pathways is base excision repair (BER), which is recognized as primarily responsible for repair of oxidized, alkylated, or deaminated bases. Because of BER’s central role in maintaining genome integrity, the pathway enzymes, particularly the DNA glycosylases that recognize the damage and catalyze the first step, have been the focus of extensive study. This has led to a general consensus regarding how these enzymes work. Now, however, Vanderbilt Institute of Chemical Biology member Brandt Eichman and postdoctoral fellow Elwood Mullins present an entirely new paradigm for the mechanism of action of DNA glycosylases and a different perspective on the role of BER in genome maintenance (E. A. Mullins, et al. Nature, published online October 28, 2015, DOI 10.1038/nature15728).


The first step of BER is removal of the damaged nucleobase through the action of a DNA glycosylase, which cleaves the glycosidic bond between the nucleobase and the sugar (Figure 1). The result is an apurinic/apyrimidinic (AP) site, which is subsequently removed by an AP endonuclease, forming a single-strand break in the DNA duplex. The break is repaired by the consecutive action of a DNA polymerase and a DNA ligase, which add the correct complementary base and reunite the sugar-phosphate backbone, respectively. Prior research had indicated that most DNA glycosylases recognize damaged nucleobases by flipping the base into a specialized pocket in the enzyme’s active site. As this occurs, the enzyme stabilizes the DNA duplex by inserting an active site amino acid into the space vacated by the flipped base (Figure 2, top). In most DNA glycosylases, a residue in the active site serves as a general acid and protonates the damaged base to destabilize the glycosidic bond. The active site further facilitates cleavage of the bond through the presence of a conserved aspartate or glutamate residue that helps to stabilize the positively charged oxocarbenium ion as it forms on the deoxyribose during the bond-breaking process. The same acidic residue assists in deprotonating the water molecule that will subsequently react with the oxocarbenium ion in the last step of the reaction. This mechanism enables repair of a wide range of DNA damage as long as the altered base is small enough to be flipped into the enzyme’s active site pocket. This has led to the generally accepted conclusion that BER is reserved for damage that does not radically alter the size of the nucleobase. In contrast, the nucleotide excision repair (NER) pathway handles damage involving the presence of large adducts.



Figure 1. The base excision repair pathway. (A) A lesion in DNA is removed by a DNA glycosylase, which cleaves the glycosidic bond between the nucleobase and the deoxyribose forming an AP site. (B) An AP endonuclease removes the AP site by cleaving bonds in the sugar-phosphate backbone. The result is a single-strand break. (C) A DNA polymerase adds the appropriate base to fill the void left by removal of the AP site. (E) DNA ligase reunites the sugar-phosphate backbone. Figure reproduced with permission from S. S. David and S. D. Williams (1998) Chem. Rev., 98, 1221. Copyright 1998 American Chemical Society.


Figure 2. (Top) Diagrammatic representation of the crystal structure of AAG (a base-flipping DNA glycosylase) in complex with a substrate. The damaged nucleobase (magenta) is no longer paired to its complement (green). Instead, it has flipped out of its position in the DNA duplex and occupies the AAG active site. An intercalating residue from the enzyme (gray) fills the space left by the damaged base. (Bottom) Diagrammatic representation of the crystal structure of AlkD in complex with a substrate. Note that the damaged nucleobase (magenta) and its complement (green) remain paired in the DNA duplex. Figure kindly provided by Brandt Eichman (Copyright 2015, Brandt Eichman).


The Eichman lab began to question the dogma about DNA glycosylases through their study of AlkD, a glycosylase from Bacillus cereus that lacks both a nucleobase binding pocket and an intercalating residue. Their initial work with this enzyme indicated that, in addition to these unusual traits, AlkD was capable of removing bulky DNA lesions, as long as they were positively charged. These findings suggested that AlkD must be acting by a mechanism quite distinct from that of most DNA glycosylases, leading the Eichman investigators to question just how the enzyme works. To answer this question, they first sought to obtain structural data to explore how the enzyme interacts with its substrate. However, this posed a problem. Positively charged nucleobases, like those cleaved by AlkD, commonly arise from alkylation at certain sites such as the N3 and N7 positions of adenine. The positive charge renders the glycosidic bond of these adducts unstable, and they will spontaneously depurinate over time even in the absence of a glycosylase. Thus, most AlkD substrates were not good candidates for obtaining crystal structure data. To solve this problem, the researchers crystallized AlkD in complex with a DNA duplex containing 3-deaza-3-methyladenine (3d3mA), a stable analog of the AlkD substrate N3-methyladenine (3mA) (Figure 3). They hypothesized that, because 3d3mA lacks a positive charge, it would not spontaneously depurinate, nor would it be cleaved by AlkD.



Figure 3. Structures of altered nucleobases used for crystallographic studies.



The structural data revealed that the binding of the 3d3mA-containing duplex in the concave surface of the protein induced a 30o bend in the DNA helix, accompanied by a 4 Å widening of the minor groove. Within this distorted helix, 3d3mA adopted two distinct conformations that existed in equilibrium (Figure 4). In the first conformation, 3d3mA was paired with the complementary thymidine through conventional Watson-Crick hydrogen bonding interactions. In the second conformation, 3d3mA was sheared away from the thymidine and displaced by 5 Å into the minor groove, although it remained partially stacked within the double helix. The sheared conformation of 3d3mA enabled the substrate to establish interactions with three residues in the AlkD active site. Tryptophan-109 and tryptophan-187 cradled the backbone and the nucleotide on each side of 3d3mA, while aspartate-113 aligned with the 3d3mA glycosidic bond, positioning it perfectly to stabilize the oxocarbenium ion that will form on bond cleavage. All three of these residues are highly conserved and required for AlkD’s activity.



Figure 4. Figure 4. Structure of a 3d3mA-containing DNA duplex in complex with AlkD. The 3d3mA (purple) can be seen in two conformations, one (shown with thin bonds) in which it is base-paired to the complementary thymidine (yellow), and a second (shown with thick bonds), in which it is sheared away from the thymidine and partially occupies the minor groove. In the sheared conformation interactions with aspartate-113, tryptophan 187, and tryptophan 109 are established. Figure reproduced by permission from Macmillan, Ltd.



As noted above, the Eichman lab investigators designed 3d3mA with the expectation that it would not be cleaved by AlkD. Yet, to their surprise, the crystal structure data suggested that slow cleavage was occurring. They explored this phenomenon more closely by flash freezing crystals of the AlkD-3d3mA-DNA complex at various time points over a prolonged incubation in an attempt to capture the cleavage reaction in various stages. They found that crystals incubated for a long period of time (>96 h) contained no intact 3d3mA bound to the DNA. Rather, an AP site was present where the 3d3mA had been located, and the nucleobase, now free from its attachment to the DNA duplex, had turned 180o from its original position (Figure 5). Structures of crystals incubated for 24 to 96 h revealed the presence of both intact 3d3mA and the AP site/free nucleobase. As would be expected, the relative amount of AP site/free nucleobase increased with increasing incubation time. These results indicated that AlkD catalyzed cleavage of the 3d3mA glycosidic bond with a rate constant of 4.6 x 10-6/s, about 1/800th the rate of removal of positively charged 3mA lesions. The investigators explained this finding on the basis that the pH of the crystallization buffer was 5.7, which allowed protonation of approximately 1% of 3d3mA nucleobases (pKa = 3.8) at equilibrium. The resulting positive charge would then enable removal of the protonated 3d3mA by AlkD. Consistent with this hypothesis, no AlkD-mediated cleavage of 3d3mA occurred in crystallization buffers at pH 7.



Figure 5. Structures of the 3d3mA nucleobase within the AlkD-3d3mA-DNA complex over prolonged incubation. At early time points, 3d3mA (purple) can be seen in both its base-paired conformation and its sheared conformation. Over time, the base-paired conformation disappears, and the free nucleobase and a corresponding AP site can be seen (pink). With prolonged inclubation, only the AP site/free nucleobase is present. Figure reproduced by permission Macmillan, Ltd. (E. A. Mullins, et al. Nature, published online October 28, 2015, DOI 10.1038/nature15728). Copyright 2015.



The ability of AlkD to cleave 3d3mA confirmed that the binding conformation observed in the crystal structures represented a true enzyme-substrate complex. The time course studies revealed both the starting points (reactant) and endpoints (products) of the reaction in association with the enzyme active site. To explore the structure of the intermediates along the pathway, the investigators synthesized a DNA duplex containing 1′-aza-2′,4′-dideoxyribose (1aR) to mimic the oxocarbenium intermediate and a free 3mA nucleobase. A second duplex contained tetrahydrofuran (THF) to mimic the product deoxyribose in addition to free 3mA (Figure 3). The crystal structure of a complex of AlkD with the 1aR + 3mA-DNA duplex revealed that the 3mA retained its position in the duplex even though it was not covalently bound. The 1aR moiety was shifted more toward the AlkD surface than the deoxyribose ring of 3d3mA, and this shift enabled stronger interactions with aspartate-113 and the incoming water molecule (Figure 6). The crystal structure of AlkD in complex with the THF + 3mA DNA duplex exhibited a very similar overall conformation of enzyme and substrate, except that the THF was rotated away from aspartate-113 relative to the position of 1aR.



Figure 6. Structure of a 1aR + 3mA-containing DNA duplex in complex with AlkD. The 3mA (purple) remains in the duplex, even though it is not covalently bound. The 1aR interacts with aspartate-113 and an incoming water molecule. Figure reproduced by permission from Macmillan, Ltd. (E. A. Mullins, et al. Nature, published online October 28, 2015, DOI 10.1038/nature15728). Copyright 2015.



The crystal structure data clearly indicated that aspartate-113 acts to stabilize the oxocarbenium ion and facilitate water addition in the AlkD active site. The exact role of tryptophan-109 and tryptophan-187, however, was not at all obvious. The only clues lay in evidence that the two tryptophans form CH-π interactions with C2′, C4′, and C5′ of the deoxyribose moiety of 3d3mA. The Eichman lab used computational methods to explore the effects of those interactions on glycosidic bond breakage in the case of a substrate containing 3mA. They discovered that both the nucleobase and the deoxyribose of 3mA exhibited substantial cationic character, and that higher cationic character correlated with increased binding energy to AlkD. As expected, the positive charge became concentrated on the deoxyribose as the glycosidic bond elongated. During bond cleavage, the CH-π interactions became more polar, and this polarization contributed substantially to the stability of the transition state, thereby increasing the rate of glycosidic bond cleavage by a factor of 103 to 104. The investigators note that such a role for CH-π interactions in transition state stabilization has not previously been reported.


Having solved the mystery of how AlkD binds to and catalyzes the removal of a positively charged abnormal nucleobase, the Eichman group quickly realized the important implications of their discovery. The results confirmed that AlkD is different from the majority of DNA glycosylases in that it does not rely on base flipping into a restricted active site pocket for activity. Rather, the substrate interacts with a much less confined concave depression on AlkD’s surface. This explains the prior observations that AlkD can accommodate much larger DNA lesions than can most DNA glycosylases. To further explore AlkD’s substrate repertoire, the Eichman group turned their attention to YtkR2, an enzyme produced by Streptomyces sp TP-A0356. These microorganisms biosynthesize yatakemycin, a natural product with antifungal and antitumor activities. Yatakemycin’s bioactivity is a direct result of its capacity to alkylate DNA, exclusively at adenine nucleobases in the minor groove. The result is the 815-Da positively charged N3-yatakemycyladenine (YTMA) adduct (Figure 7). The microorganisms that synthesize yatakemycin express YtkR2, a DNA glycosylase that is highly proficient at repairing YTMA adducts, as a self-defense mechanism against the toxicity of the natural product. Noting a strong sequence similarity between YtkR2 and AlkD, the Eichman lab hypothesized that AlkD should be able to remove YTMA adducts with reasonable efficiency, despite their large size. Indeed, molecular modeling predicted that the YTMA adduct could be accommodated in the AlkD active site. Biochemical assays confirmed that AlkD could remove YTMA adducts with an efficiency similar to that of YtkR2. Furthermore, genetic deletion of the gene for AlkD in Bacillus anthracis increased the sensitivity of the bacteria for yatakemycin, but not methyl methanesulfonate, a methylating agent. These results suggest that B. anthracis required AlkD for protection against an alkylating agent that forms bulky adducts, but not against an agent that forms small adducts, likely because other DNA glycosylases were present in the cell to serve that function.




Figure 7. Structure of YTMA. Figure reproduced by permission from Macmillan, Ltd. (E. A. Mullins, et al. Nature, published online October 28, 2015, DOI 10.1038/nature15728). Copyright 2015.



From these new findings combined with their previous studies, the Eichman group concluded that AlkD is the prototype of a group of DNA glycosylases that do not rely on the well-characterized base flipping mechanism. Instead, interactions with key residues along the surface of the protein enable the enzyme to recognize and remove damaged nucleobases. AlkD’s mechanism suggests that the primary function of these enzymes is to remove bulky DNA lesions, a role heretofore ascribed to the NER pathway. In exchange, AlkD sacrifices the ability to remove lesions that do not carry a positive charge. These conclusions mark a major shift in our understanding of the role of BER in DNA repair and the biochemical mechanism of the DNA glycosylase family of enzymes.



View Nature article: The DNA glycosylase AlkD uses a non-base-flipping mechanism to excise bulky lesions






The Vanderbilt Institute of Chemical Biology, 896 Preston Building, Nashville, TN 37232-6304, phone 866.303 VICB (8422), fax 615 936 3884
Vanderbilt University is committed to principles of equal opportunity and affirmative action. Copyright © 2014 by Vanderbilt University Medical Center

  • Areas of Interest