Identifying the Elusive Protein-DNA Cross-Link
By: Carol A. Rouzer, VICB Communications
Published: October 22, 2013
Structural elucidation of dibromoethane-mediated cross-links between DNA and a DNA repair enzyme pave the way for understanding mutagenesis.
Many small molecule electrophiles can react with the nucleophilic sites on DNA, leading to alkylation damage. If not repaired, these damaged sites may be replicated incorrectly, resulting in mutations and ultimately cancer or cell death. Recent accumulating evidence indicates that bifunctional electrophiles can react with nucleophiles on both DNA and protein to form a protein-DNA cross-link. Such cross-link formation carries an even greater potential for toxicity, but precise characterization of the chemical structure of this form of DNA damage has been elusive. Now, VICB member Fred Guengerich, his collaborator Anthony Pegg (University of Pennsylvania), and their laboratories have solved this puzzle for cross-links formed between the bifunctional electrophile 1,2-dibromoethane (DBE) and the enzyme O6-alkylguanine-DNA alkytranferase (AGT) [G. Chowdhury, et al. (2013) Angew. Chem. Int. Ed., published online Oct. 15, doi:10.1002/anie.201307580].
In the past, DBE was used as a gasoline additive, a pesticide, and a soil fumigant. Following the discovery that DBE is carcinogenic in animals and mutagenic in multiple organisms, including humans, most industrial and agricultural uses of DBE have been curtailed. Prior studies have shown that products formed during microsomal oxidation or glutathione transferase-mediated detoxication likely contribute to DBE’s toxicity. However, the unexpected observation that increased expression of AGT augments the toxicity of DBE led to the hypothesis that DBE-mediated AGT-DNA cross-link formation might play a role. A mechanism by which cross-links could form is shown in Figure 1. The sulfhydryl group of cysteine-145, located in AGT’s active site, first attacks DBE, displacing one bromide and forming a half-mustard. Then, attack of the sulfur on the second carbon of DBE produces a reactive episulfonium ion. AGT is a DNA repair enzyme that removes alkyl groups from the O6 position of guanine and O4 position of thymine bases by transferring the groups to cysteine-145. Thus, the natural function of the enzyme places cysteine-145 in close proximity to the bases of DNA. As a result, an AGT bearing a cysteine-145 episulfonium ion can easily react with a nucleophilic site of DNA, forming an ethylene cross-link between the two macromolecules.
Figure 1. Mechanism for AGT-DNA cross-link formation in the presence of DBE. Cysteine-145 in the AGT active site, in thiolate anion form, attacks one carbon of DBE, displacing bromide. Then, the sulfur attacks the other carbon of DBE to form a cyclic episulfonium ion. This reactive electrophile can react with water, leaving just an alkylated enzyme. If, however, it reacts with a nucleophilic site on DNA, a cross-link is formed.
Past attempts to characterize the structure of protein-DNA cross-links have met with limited success. Standard enzymatic protocols used to digest protein or DNA to obtain individual amino acids or nucleosides, respectively, for analysis frequently fail due to steric hindrance resulting from the close proximity of the two large molecules. Most chemical methods for breakdown of protein or DNA are too harsh and would result in degradation of the cross-link structure. Mass spectrometry methods that are optimal for protein analysis are not ideal for analysis of DNA and vice versa. The only exception to this conundrum is the case of unstable cross-links, such as those involving alkylation at the N7 position of guanine or the N3 position of adenine. Adduction at these positions leads to rapid depurination by thermal decomposition. This frees the adducted base from the DNA while maintaining the bond to the protein, allowing analysis by standard methods for protein adduct detection.
Studies using purified AGT treated with DBE in vitro revealed that the AGT episulfonium ion can react with all four bases of DNA. The order of reactivity under these conditions was G>T>C>A. Incubation of the AGT episulfonium ion with DNA clearly produced cross-links, but the only previously characterized structure was a labile N7-guanyl adduct (Figure 2). Due to its high susceptibility to depurination, a cross-link at the N7-position of guanine is most likely to give rise to a G:C to T:A transversion mutation. This results from the propensity of polymerases to place adenine across from an abasic site during replication. However, DBE treatment of cells overexpressing AGT results in G:C to A:T transitions most frequently, suggesting that the most mutagenic cross-link is not the N7-guanyl adduct. This led the Guengerich and Pegg labs to hypothesize that other, nonlabile cross-links are also formed.
Figure 2. Structures of the cross-links formed between AGT and DNA in the presence of DBE. The cross-links are shown from left-to-right in decreasing order of prevalence.
To investigate the structures of stable DBE-mediated AGT-DNA cross-links, the investigators incubated AGT with DBE in the presence of a 15 base pair GC-rich double-stranded synthetic oligonucleotide. Heating the resulting cross-linked sample at 90oC for 30 minutes destroyed any labile adducts, leaving only stable cross-links for further analysis. Although treatment with trypsin did not fully digest the cross-linked AGT, it did reduce the protein to a 12 amino acid peptide, suitable for mass spectrometry (Figure 3). As expected, the results confirmed the presence of molecules containing both the peptide and the oligonucleotide joined by an ethylene bridge, but the data provided no information on the exact site of cross-link formation on the oligonucleotide. To address this problem, the investigators repeated the experiment using a double stranded oligonucleotide with the sequence T5G2T4. In this case, mass spectrometric analysis using collision-induced dissociation yielded an ion with an m/z of 1450, consistent with the structure of the peptide linked to guanine through an ethylene bridge (Figure 3). Similarly, use of an oligonucleotide with the sequence (AT)6 yielded, upon mass spectrometric analysis, an ion with an m/z of 1474.6, indicating a structure comprising the peptide, adenine, and an ethylene bridge (Figure 3). These results indicated that DBE could mediate stable cross-link formation between AGT and both the adenine and guanine bases of DNA.
Figure 3. Detection of stable AGT-DNA adducts. AGT was incubated with a synthetic oligonucleotide in the presence of DBE. Following cross-link formation, the sample was heated to decompose labile adducts and then digested with trypsin. Mass spectral analysis using collision-induced dissociation of the resulting peptide-oligonucleotide adducts led to the detection of cross-links to both guanine and adenine..
To further characterize the structure of DBE-mediated cross-links between AGT and DNA, the investigators incubated calf thymus DNA with AGT in the presence of DBE. They treated the reaction products with trypsin or proteinase K to reduce the AGT to a simple peptide and then heated the sample to remove labile adducts. Based on prior work, the researchers hypothesized that reductive desulfurization using Raney nickel would successfully cleave the bond between the cysteine-145 sulfur atom of AGT and the ethylene bridge of the cross-link. This hypothesis proved to be correct, and the result was DNA bearing only an ethyl group at the adduct site. With the peptide no longer attached, routine treatment with nucleases and phosphatases readily reduced the DNA to individual nucleosides, which could easily be analyzed by mass spectrometry (Figure 4). The results indicated the presence of N6-ethyl-dA, N2-ethyl-dG, and O6-ethyl-dG (Figure 2).
Figure 4. Analysis of AGT-DNA cross-links formed in the presence of DBE. Calf thymus DNA, AGT, and DBE (radiolabeled or unlabeled) were incubated together to form cross-links. The AGT was digested with proteinase K or trypsin, and the resulting DNA-peptide cross-links were treated with heat to decompose the labile cross-links. Both stable and labile cross-links were then treated with Raney nickel to remove the peptide. The DNA in the stable cross-links was digested to individual nucleotides, and then these species, for both the stable and labile cross-links, were subjected to analysis by mass spectrometry in the case of unlabeled DBE or by HPLC with radiolabel detection when radiolabeled DBE was used.
The success of the Raney nickel approach led the investigators to repeat their experiment using calf thymus DNA, AGT, and DBE radiolabeled in both carbon atoms. Following proteinase K digestion, thermal decomposition, and Raney nickel desulfurization of the sample, the investigators analyzed both labile and stable adducts by high performance liquid chromatography using scintilation counting to detect the radiolabeled ethylene bridge that was derived from DBE (Figure 4). The results confirmed formation of N6-ethyl-dA, N2-ethyl-dG, and O6-ethyl-dG as detected in the prior experiment, in addition to an N1-ethyl-dG adduct formed in very low amounts. The experiment also confirmed the presence of the labile N7-ethyl-dG adduct and demonstrated that no other labile adducts had been formed. Quantification of the radiolabel revealed that cross-linking occurred primarily at the N7-position of guanine, which comprised approximately 80% of all detected species. The stable cross-links were formed in the relative order N6-ethyl-dA ≈ N2-ethyl-dG > O6-ethyl-dG ≈ N1-ethyl-dG.
The combination of enzymatic and mild chemical treatments solved the structure of the cross-links formed between AGT and DNA through the action of DBE. This approach is potentially applicable to the elucidation of other cross-links formed at protein sulfhydryl groups. The investigators note that the identification of stable cross-links confirms their hypothesis that linkages other than those at the N7-position of guanine could be giving rise to the G:C to A:T transitions observed in DBE mutagenesis experiments. If true, however, then one must conclude that the less prevalent stable cross-links are responsible for the majority of mutations. The mechanism by which these mutations arise is currently unknown. However, the fact that some translesion polymerases can bypass DNA-peptide cross-links suggests that these enzymes may also bypass the more complex DNA-protein cross-links. If so, then these error-prone polymerases may give rise to mutations. Alternatively, the process of homologous recombination, which is the primary route of repair of DNA-protein cross-links in eukaryotic cells, may lead to mutation-inducing errors. Further work will be required to explore these possibilities, but that work can now be based on solid knowledge of the structures of the relevant cross-links.