DNA Adduct Structure–Function Relationships: Comparing Solution with Polymerase Structures

Abstract

It has now been nearly two decades since the first solution structures of DNA duplexes covalently damaged by metabolically activated polycyclic aromatic hydrocarbons and amines were determined by NMR. Dozens of such high-resolution structures are now available, and some broad structural themes have been uncovered. It has been hypothesized that the solution structures are relevant to the biochemical processing of the adducts. The structural features of the adducts are considered to determine their mutational properties in DNA polymerases and their repair susceptibilities. In recent years, a number of crystal structures of DNA adducts of interest to our work have been determined in DNA polymerases. Accordingly, it is now timely to consider how NMR solution structures relate to structures within DNA polymerases. The NMR solution structural themes for polycyclic aromatic adducts are often observed in polymerase crystal structures. While the polymerase interactions can on occasion override the solution preferences, intrinsic adduct conformations favored in solution are often manifested within polymerases and likely play a significant role in lesion processing.

1. Introduction

Polycyclic aromatic hydrocarbons (PAHs)¹ and amines are well-known environmental chemicals present in tobacco smoke, automobile and diesel exhaust, and broiled fish and meats (1, 2). Metabolic activation of these substances can produce highly reactive products, which may subsequently react with DNA, yielding carcinogen–DNA adducts (1, 3, 4). These adducts, if not accurately repaired by the nucleotide excision repair machinery (5), may produce mutations during DNA replication. Such mutations, when present in DNA sequences that function in regulating the cell cycle, may lead to the initiation of cancer (1, 6).

It has now been nearly two decades since the first structures of DNA duplexes covalently damaged by metabolically activated PAHs and amines were determined by NMR solution methods. Dozens of such high-resolution structures are now available, and some broad structural themes have been uncovered. It has been hypothesized that the solution structures are relevant to the biochemical functioning of the adducts. The structural features of the adducts are considered to determine their mutational properties in DNA polymerases and their repair susceptibilities. Extensive reviews of NMR solution structures of damaged DNA have been presented by Geacintov et al. (7) and Patel et al. (8); recently, Cho has reviewed evidence for conformational heterogeneity in adduct structures (9), and Lukin and de Los Santos have presented a comprehensive survey of NMR structures of damaged DNA (10). In recent years, a number of crystal structures of DNA adducts of interest to our work have been determined in DNA polymerases (11–15). Accordingly, it now seems timely to consider how NMR solution structures relate to structures within DNA polymerases.

Of particular interest are adducts with available NMR solution structures as well as X-ray crystal structures within polymerases. Among the PAHs, we focus on benzo[a]pyrene (B[a]P) and benzo[c]phenanthrene (B[c]Ph); among the aromatic amines, we focus on 2-aminofluorene (AF) and N-2-acetyl-aminofluorene (AAF), with some selected consideration of other aromatic amine adducts for which NMR data are at hand. We first review conformational themes defined in NMR studies.

2. NMR Structural Themes

B[a]P and B[c]Ph represent two model PAHs with uniquely different topological features (Figure 1A, upper panel). The planar B[a]P has a characteristic bay region, while B[c]Ph, one of the most tumorigenic PAHs (16), has a sterically hindered fjord region, which causes nonplanarity in the aromatic ring system (17). In addition, the B[a]P ring system is extended while that of B[c]Ph is curved. Metabolic activation of B[a]P and B[c]Ph produces, among others, a pair of mirror image diol epoxides in each case (Figure 1A, lower panel); these can react with the amino groups of adenine and guanine by trans- and cis-epoxide opening to produce adducts differing in base-linkage site, stereochemistry, and ring topology (18).

Figure 1.

Structures of representative PAHs (A) and polycyclic aromatic amines (B), and their DNA adducts. PhIP is a heterocyclic aromatic amine. PAHs are adducted to DNA via the N² of guanine or the N⁶ of adenine. Polycyclic aromatic amines are adducted to DNA via the C8 of G. Chiral centers in the A ring of the PAH adducts and the linkage site of the polycyclic aromatic amine adducts are shown in blue. X is H or COCH₃.

NMR conformational themes that have been defined for these adducts (7, 10) are shown in Figure 2. For guanine (G) adducts derived from B[a]P, trans adducts in duplexes prefer to reside in the B-DNA minor groove with Watson–Crick pairing at the lesion site retained. The 10S adduct is oriented in the 5′-direction along the modified strand, while the 10R adduct is 3′-oriented (Figure 2A). The cis adducts favor a base-displaced intercalated conformation with Watson–Crick pairing disrupted at the modification site and the pyrenyl ring system inserted in place of the modified G and its partner cytosine (C), both of which are displaced into the major groove (Figure 2B) in the (−)-cis case. However, the displaced G is in the minor groove in the case of the (+)-cis adduct. Here, opposite orientations in 10S and 10R stereoisomeric adducts are manifested by protrusion of the puckered benzylic ring into the minor groove with the pyrenyl rings directed towards the major groove in the 10S adduct case and visa versa for the 10R adduct. There is evidence that equilibria between these conformational families are governed by base-sequence context in both cis and trans adducts (7, 10, 19, 20). The B[c]Ph G adducts adopt a third conformational theme, “classical intercalation”, with perturbed Watson–Crick base pairing retained at the lesion site and the aromatic ring system intercalated on the 5′-side of the modified base pair in the 1S case and on the 3′-side in the 1R case (Figure 2C). The helix is stretched and unwound to permit insertion of the aromatic rings. In the case of the adenine adducts, classical intercalation themes are observed in lesions derived from both B[a]P (Figure 2D) and B[c]Ph (Figure 2E), with 3′-side intercalation for 10S-B[a]P and 1S-B[c]Ph adducts and 5′-side intercalation for the respective R stereoisomeric adducts. While the overall conformational family is the same in the B[a]P and B[c]Ph adducts, there are important differences in detail stemming from the topological differences of the bay and fjord region adducts, which profoundly affect their repair susceptibilies, as has been discussed previously (21–23).

Figure 2.

Conformational themes for B[a]P and B[c]Ph adducts in duplex DNA in solution. The 5′- to 3′-directionality of all modified strands is given in panel A.

A second group of chemicals of interest in our structural studies are the polycyclic aromatic amines (Figure 1B, upper panel). Their metabolites form predominant adducts to the C8 of G (Figure 1B, lower panel), and additional minor products have also been observed (8, 10). The AF adducts have been investigated most extensively, and their representative structural themes have been defined (8, 9). Three conformational families have been observed as follows: (i) aromatic rings in the B-DNA major groove with Watson–Crick pairing intact and modified G anti (Figure 3A); (ii) base-displaced intercalated structures with Watson–Crick pairs disrupted at the lesion site, with aromatic rings inserted into the helix and the modified G in the syn conformation (Figure 3B); and (iii) adducts positioned in the B-DNA minor groove and with the modified G syn, characterized for the case of a G:A mismatch at the lesion site (Figure 3C). The equilibrium between major groove/anti and base-displaced intercalated/syn conformers is determined by base-sequence context (8), with enhanced carcinogen/base stacking likely accounting for preferred stabilization of the base-displaced structures in certain sequences. The conformational balance is dependent on ring topology as well. Thus, the 1-aminopyrene adduct with its multiple rings shows only a syn/base-displaced intercalated conformation (24). In the case of the 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) adduct, this conformation is about 90% of the population with the adduct in the remaining population solvent-exposed in a groove (25). For the 2-amino-3-methylimidazo[4,5-f]quinoline adduct, base-sequence context also plays a key role in determining the conformational balance between major groove and base-displaced intercalated conformations (26, 27). The flexible 4-aminobiphenyl adduct is predominantly in the major groove (28). The AAF adduct adopts a syn/base-displaced intercalated conformation in about 70% of the population; the conformations of the remaining adducts are uncharacterized (29). The syn conformation is particularly preferred by the AAF adduct due to crowding of the bulky acetyl group with adjacent sugar when the modified G is anti, as first noted by Grunberger et al. (30) and by Fuchs and Daune (31) in their base-displacement and insertion-denaturation models. However, molecular dynamics simulations show that modest sugar repuckering can alleviate the crowding and permit the anti conformation, thereby allowing Watson–Crick pairing (32). Observed preferential incorporation of dCTP opposite dG-AAF in DNA polymerases (33–37) is thus structurally explained (32).

Figure 3.

Conformational themes for AF adducts in duplex DNA in solution.

The variety of conformational themes adopted by these bulky aromatic adducts is governed by a subtle interplay of competing forces. To accommodate the adducted base, an energetic price is exacted in distorting the DNA through disturbed stacking, Watson–Crick pairing, and other perturbations such as stretching and unwinding. However, shielding of the adduct aromatic rings from solvent and/or stacking them with bases within the DNA helix is favorable. Van der Waals and electrostatic interactions between the adduct and the DNA backbone can also provide stabilizing forces. Thus, conformational preferences are a combined outcome of structural factors including adduct topology, nature of modified base (adenine or guanine), linkage site (e.g., C8 or N² of G), and base-sequence context that together determine the minimum free energy conformation.

3. DNA Polymerases

Crystallographic analyses of a number of replicative DNA polymerases have yielded excellent structures that defined common features of these enzymes. They are shaped like a right hand with palm, finger, and thumb domains, with the active site located in the palm domain (38). Replicative DNA polymerases employ an induced-fit mechanism in which the polymerase closes largely through rotation of the finger domain when the Watson–Crick partner to the template enters the active site. An α-helix of the finger domain closes around the nascent base pair with a tight steric fit (39). The resulting active site is catalytically competent and ready for the nucleotidyl transfer reaction. Two metal ions, usually Mg²⁺, are critical to the catalytic mechanism, as are conserved carboxylate residues (Asp or Glu) (40). These polymerases have open pockets on the developing major groove side of the templating base. However, the minor groove side is tightly packed with amino acid residues; these protein–DNA interactions are essential for processivity and fidelity (41–43). The repair polymerase β (Pol β), a member of the X family, has overall structural and mechanistic features that are similar, although the domains differ (44).

When replicative polymerases encounter a lesion, they are frequently stalled and replaced by specialized lesion-bypass polymerases, many of which are from the Y family (45–48). One or more bypass polymerases are involved in transiting the lesion and extending several residues past it before the replicative polymerase resumes replication. Bypass polymerases are typically low fidelity on unmodified DNA and may be high or low fidelity in lesion bypass (43). Crystallographic characterizations of the Y family polymerase Dpo4, from the archaeon bacterium Sulfolobus solfataricus (14, 49–60), have provided structural understanding of this DinB homologue (37) of the human polymerase κ (61). Like replicative polymerases, Dpo4 is shaped like a right hand with an open pocket on the developing major groove side of the template and employs a two-metal ion mechanism for nucleotidyl transfer. However, the finger domain of Dpo4 is short with an additional flexible little finger domain holding the DNA backbone from the major groove side. The spacious active site, open and water-accessible on both the major and the minor groove side of the template, is capable of accommodating two templating residues. Entry of the correct dNTP partner to the template does not trigger an induced fit closing motion. Instead, a motion of translocation involving movement of the little finger and subsequently that of the thumb, relative to the DNA, is observed as the DNA is elongated by one residue in a replication cycle. This process produces a counter-clockwise rotation (when the ternary complex is viewed in the 5′- to 3′-direction of the template strand) and a concerted translation (in the 3′- to 5′-direction of the template) of the polymerase (57). A detailed chemical mechanism for nucleotidyl transfer in Dpo4 has recently been delineated (62). Recent crystal structures of human Y family polymerases Pol ι (63–66) and Pol κ (67) reveal significant similarities as well as individual structural features tailored to the specific enzyme function in nucleotide incorporation opposite lesions or extension beyond lesions.

4. Adduct Structures in DNA Polymerases

We now turn to structures of bulky adducts of interest within DNA polymerases. In this connection, the representative Pol I A family replicative polymerase Bacillus fragment (BF) from Bacillus stearothermophilus has provided valuable crystallographic insights. Crystal structures of BF without lesion have identified a series of positions in its active site region that the templating base traverses through a replicative cycle (68). The next-to-be replicated base occupies the preinsertion site, between the O and the O1 helices, in the open binary complex; Tyr 714 occupies the insertion position. At the next step, the template and its partner dNTP are placed in the insertion site in the closed ternary complex; the O and O1 helices have rotated, thus closing the active site of the polymerase, and Tyr 714 has vacated the active site. Following the nucleotidyl transfer reaction, the nascent base pair is translocated to the postinsertion site with the next base (5′ to the nascent base pair on the template strand) occupying the preinsertion site in the open polymerase binary complex, ready for another cycle of replication.

A crystal structure of the 10S-(+)-trans-anti-B[a]P-N²-dG adduct in the postinsertion site of BF (11) has been solved (PDB ID:1XC9). In this structure, the B[a]P rings are 5′-directed along the template strand on the minor groove side with the Watson–Crick pair intact at the lesion site (Figure 4A), as in the DNA duplex structure solved by NMR (Figure 2A, left panel), although the minor groove is not fully developed in the polymerase structure and detailed orientation of the B[a]P rings differs somewhat. However, an NMR solution structure of this adduct at a single-strand/double-strand junction with dC incorporated opposite the damaged guanine (PDB ID: 1AXO) (69) is very similar in orientational detail (Figure 4B). The B[a]P disrupts the critical polymerase minor groove contacts, hence accounting for the predominant blockage observed in primer extension studies of this adduct with BF. Small amounts of bypass are also experimentally noted and have been interpreted through modeling to entail a syn conformation of the damaged base with the B[a]P rings in the open pocket on the major groove side (70).

Figure 4.

Structures of the 10S-(+)-trans-anti-B[a]P-N²-dG adduct (A) in polymerase BF (PDB ID:1XC9) and (B) at a single-strand/double-strand junction in solution (PDB ID:1AXO).

Two different structures of dG–AF adducts in DNA–BF complexes have also been studied (12), with AF in the preinsertion site in one structure (PDB ID:1UA0) and in the postinsertion site in the other (PDB ID:1UA1). Interestingly, the adducted base is replicated by the polymerase, transiting from preinsertion to insertion and postinsertion sites. In the preinsertion site, the modified guanine is syn, and the AF rings are accommodated in the preinsertion site with extensive van der Waals interactions, with the syn-G stacked with the adjacent sugar (Figure 5A). While the O1 helix is disordered, the O helix remains ordered, open, and properly positioned for transfer of the adducted base to the insertion site. The polymerase is otherwise little disturbed. In transiting to the postinsertion site, however, the modified G is rotated to anti, and Watson–Crick pairing is intact with the AF rings in the major groove pocket (Figure 5C). The active site regions are markedly distorted; the next templating base is stacked with the fluorenyl rings and is blocked from entering the preinsertion site. The O and O1 helices are also distorted, and Tyr 714 is displaced. These structures explain replication of the modified G and also account for the observed predominant blockage of further primer extension. A surprising syn–anti rotation occurs in the polymerase, as was also observed in the NMR solution structures of this adduct. In the postinsertion site, the structure is very similar to the major groove NMR solution structure. The conformation in the preinsertion site where the modified G is single-stranded resembles the AF base-displaced intercalated and minor groove structures in that they share the syn conformation. The existence of a syn–anti glycosidic bond rotation in the polymerase, as in solution, is remarkable, as is the similarity between the NMR solution and the polymerase postinsertion site structures.

Figure 5.

Structures of dG-C8-AF in polymerase BF at (A) preinsertion (PDB ID:1UA0) and (C) postinsertion (PDB ID:1UA1) sites, respectively. The corresponding conformational themes in solution are shown in panels B and D, respectively. Tyr714 of the O helix is shown in cyan.

A crystal structure of a dG–AAF adduct in BF in the preinsertion site was also investigated and was found to have an open polymerase conformation with a well-ordered preinsertion site that was empty (12). However, the Pol I A family polymerase from bacteriophage T7 did yield a structure of the adduct (PDB ID: 1X9M) (13). The intent was to crystallize a ternary complex with templating dG–AAF and incoming ddCTP. However, no incoming nucleotide was observed in the crystal. The dG and adjacent sugar were located on the surface of the finger subdomain, flipped out of the polymerase active site in the syn conformation. The AAF rings were inserted into a hydrophobic pocket behind the O helix, which was rotated about 35° from the closed position that it adopts in the unmodified ternary complex. The site for binding the incoming nucleotide is blocked by the displaced O helix. This structure thus explains the predominant blockage one base before this lesion in T7 DNA polymerase. The strongly preferred syn conformation of this adduct in solution (29) is manifested in the polymerase.

A recent crystal structure has been solved for Pol β, in which a 1S-(−)-trans-anti-B[c]Ph-N²-dG adduct is the templating base in a DNA substrate with a one nucleotide gap (PDB ID: 2I9G) (15). The phenanthrenyl ring system is stacked over the base pair immediately 3′ to the modified deoxyguanosine, which is in the syn conformation (Figure 6A). This base is displaced downstream and prevents the polymerase from closing to form a catalytically active complex. Nonetheless, inefficient and error-prone bypass was observed with preferred misinsertion of purine opposite the lesion; it was suggested that the bypass might be untemplated and favored by the enlarged binding site and purine stacking with the aromatic B[c]Ph rings. Interestingly, a similar carcinogen–base linkage geometry, defined by torsion angles α′ [N1(G)–C2(G)–N²(G)–C1(B[c]Ph)] and β′ [C2(G)–N²(G)–C1(B[c]Ph)–C2(B[c]Ph)], was found in the NMR solution structure of this adduct (PDB ID:1HWV), although the modified nucleoside adopted the anti conformation and Watson–Crick pairing was observed in the NMR structure (Figure 6B) (71).

Figure 6.

Structures of the 1S-(−)-trans-anti-B[c]Ph-N²-dG adduct (A) in Pol β (PDB ID:2I9G) and (B) in duplex DNA in solution (PDB ID:1HWV).

In the Y family bypass polymerase, Dpo4, crystal structures have been solved for the 10R-(+)-cis-anti-B[a]P-N⁶-dA adduct (PDB ID:1S0M) (14), which has also been characterized by high-resolution NMR (PDB ID:1AXV) (72). In the crystal, the modified anti-adenine was in the postinsertion site opposite T. Two different conformers were found. In one conformation, the pyrenyl rings occupied the major groove open pocket with little perturbation to the polymerase; this conformer appeared to be fairly near-reaction ready with a distance between primer terminal O3′ and Pα of the dNTP < 5 Å (Figure 7A). The second conformer was nearly identical to the NMR solution structure, with pyrenyl rings intercalated on the 5′-side of the damaged base whose Watson–Crick pairing was perturbed but not ruptured. In this case, the O3′–Pα distance was >10 Å, which would preclude the nucleotidyl transfer reaction (Figure 7B). Ling et al. (14) suggest that poor extension beyond the adduct in Dpo4 occurs because the major groove conformer that appears most catalytically competent is in an energetically unfavorable domain in solution (73) and would occur less frequently. However, in Dpo4, the energetic advantage derived from the comfortable housing in the Dpo4 major groove pocket offsets the energy penalty.

Figure 7.

Structures of the 10R-(+)-cis-anti-B[a]P-N⁶-dA adduct (A, B) in polymerase Dpo4 (PDB ID:1S0M) and (C) in duplex DNA in solution (PDB ID:1AXV).

A recent series of crystal structures of Dpo4 containing three steps in the replicative cycle (a short binary complex, a ternary complex with dNTP, and a long binary complex following reaction, which is elongated by one nucleotide) have revealed the mechanism of translocation in this enzyme (57). The consecutive motion of the little finger and thumb, relative to the DNA, that was revealed provided insights into how bulky adducts may affect translocation in this enzyme. Modeling studies for the PhIP-C8-dG, AAF-C8-dG, and 10S-(+)-transanti-B[a]P-N²-dG adducts suggested that translocation would be much more impeded by the multiringed C8–dG adducts on the major groove side of the modified template as compared to the N²–dG adduct on the minor groove side (27, 74). The key feature is the steric blockage of the little finger rotation by the C8–dG major groove adducts but not the minor groove N²–dG adduct (Figure 8A). Primer extension studies in DinB family polymerases have shown more facile bypass of N²–dG than C8–dG adducts (Figure 8B) (37, 75, 76), in line with these modeling studies. Future investigations of translocation mechanisms in the face of bulky adducts have been facilitated by these Dpo4 crystal structures.

Figure 8.

Models of translocation blockage in polymerase Dpo4: (A) major groove positioned anti-dG-C8-PhIP adduct and (B) minor groove positioned 10S-(+)-trans-anti-B[a]P-N²-dG adduct in Dpo4. The Dpo4 little finger domain is shown in purple.

5. Conclusion

In conclusion, it appears on the basis of present evidence that NMR solution structural themes for polycyclic aromatic adducts are often observed in crystal structures of DNA polymerases. Striking similarities are observed for B[a]P-, B[c]Ph-, AAF-, and AF-derived adducts. However, polymerase interactions may at times override the intrinsic preferences in solution, even permitting an unfavorable solution structure to occur within the polymerase (14, 73). Favorable steric factors and hydrophobic interactions involving the bulky adduct are important considerations in this connection. Solvent exposure of the polycyclic aromatic ring system may be minimized through interactions with specific niches of the polymerase, as has recently been observed for a 10S-(+)-trans-anti-B[a]P-N²-dG adduct in the postinsertion site in Dpo4 (77, 78). Interestingly, in this case, too, the carcinogen–base linkage site adopts conformational domains similar to those in the NMR solution structure (19). The adduct structures in polymerases have begun to provide structural understanding of polymerase blockage and bypass of bulky lesions. In turn, modeling studies are beginning to take advantage of this knowledge to evaluate structural properties producing blockage or bypass by these types of lesions (70, 74, 79). Future structural studies within polymerases are eagerly awaited to provide further insights.