Rev1 Employs a Novel Mechanism of DNA Synthesis Using a Protein Template

See allHide authors and affiliations

Science  30 Sep 2005:
Vol. 309, Issue 5744, pp. 2219-2222
DOI: 10.1126/science.1116336


The Rev1 DNA polymerase is highly specialized for the incorporation of C opposite template G. We present here the crystal structure of yeast Rev1 bound to template G and incoming 2′-deoxycytidine 5′-triphosphate (dCTP), which reveals that the polymerase itself dictates the identity of the incoming nucleotide, as well as the identity of the templating base. Template G and incoming dCTP do not pair with each other. Instead, the template G is evicted from the DNA helix, and it makes optimal hydrogen bonds with a segment of Rev1. Also, unlike other DNA polymerases, incoming dCTP pairs with an arginine rather than the templating base, which ensures the incorporation of dCTP over other incoming nucleotides. This mechanism provides an elegant means for promoting proficient and error-free synthesis through N2-adducted guanines that obstruct replication.

Rev1, a member of the eukaryotic Y family DNA polymerases, is highly specific for incorporating a C opposite template G (1, 2). In this respect, Rev1 differs not only from the replicative and repair polymerases (Pols), which incorporate the correct nucleotide opposite all four template bases with nearly equivalent catalytic efficiencies, but it differs also from the other three eukaryotic Y family Pols—η, ι, and κ (3). Of these, Pols η and κ form the four Watson-Crick base pairs with similar catalytic efficiencies (46), whereas Polι incorporates the correct nucleotide opposite template purines with a much higher efficiency than opposite template pyrimidines (711).

Rev1 incorporates a C opposite template G with an efficiency (kcat/Km) of 1.2 μM–1 min–1, and it misincorporates a G, an A, or a T opposite template G with efficiencies that are lower by a factor of about 103 to 105 than for the incorporation of C (2). Rev1 also incorporates a C opposite templates A, T, and C with efficiencies that are lower by a factor of about 102 to 103 than for the incorporation of C opposite template G; moreover, Rev1 shows no propensity for incorporating the correct nucleotide opposite these template residues (2). Rev1 incorporates a C opposite an abasic site also, albeit with a lowered efficiency compared with that opposite template G (2). Rev1, thus, is specific not only for template G, but also for the incoming 2′-deoxycytidine 5′-triphosphate (dCTP) (2). Furthermore, Rev1 can proficiently incorporate a C opposite an N2-adducted G, for example, a γ-hydroxy-1,N2-propano-2′-deoxyguanosine (γ-HOPdG) adduct generated from the reaction of acrolein (produced by the peroxidation of lipids in cells) with the N2 of a G (12).

The specificity of Rev1 poses two questions. First, what is the chemical basis for Rev1's G-template specificity? Second, what is the chemical nature of Rev1's specificity for incorporation of a C nucleotide, even opposite an abasic site?

We report here the structure of the Rev1 catalytic core (residues 297 to 746), which exhibits the same nucleotide incorporation specificity and efficiency as the wild-type protein, in ternary complex with a template-primer presenting G in the active site and with incoming dCTP (table S1). The structure, determined at 2.3 Å resolution (fig. S1), reveals Rev1 embracing the template-primer with its palm (residues 356 to 365, 438 to 536), fingers (366 to 437), and thumb (537 to 603) domains, and the PAD (polymerase-associated domain; residues 621 to 738) unique to Y family polymerases (1318) (Figs. 1 and 2). The palm carries the active site residues (Asp362, Asp467, and Glu468) that catalyze the nucleotidyl transfer reaction. The fingers domain lies over the replicative end of the template-primer, but, unlike other DNA polymerases it makes very few contacts with the templating base; instead, interactions are primarily with incoming dCTP and the unpaired nucleotides at the 5′ end of the template. The thumb and the PAD approach the template-primer from opposite sides, connected by a long linker, which is mostly helical rather than extended as in other Y family polymerases (Figs. 1 and 2). Incoming dCTP binds with its triphosphate moiety interlaced between the fingers and palm domains, making hydrogen bonds with Ser402, Tyr405, and Arg408 from the fingers domain and Lys525 from the palm domain (Fig. 3A). The catalytic residues, Asp362, Asp467, and Glu468 are arrayed between the dCTP triphosphate moiety and the primer terminus, and two Mg2+ ions—analogous to metals “A” and “B” in replicative DNA polymerases (1921)—complete the Rev1 active site (Fig. 3A). Overall, Rev1 is well poised for dCTP insertion, with the putative 3′ oxygen (at the primer terminus) located ∼4 Å from the dCTP α-phosphate and aligned with respect to the Pα-O3′ bond (angle of about 160°).

Fig. 1.

Structure of the Rev1-DNA-dCTP ternary complex. The palm, fingers, and thumb domains and the PAD are shown in cyan, yellow, orange, and green, respectively. The linker joining the thumb to the PAD is shown in pink. The N-digit in Rev1 is shown in dark blue. DNA is in gray, template G and incoming dCTP are in red, and the putative Mg2+ ions are in dark blue. Black dashed lines depict hydrogen bonds between dCTP and Arg324, and between template G and a loop in the PAD. Cyan dashed line indicates an unstructured loop in the palm domain.

Fig. 2.

Comparison between Rev1 and Polι. (A) The juxtaposition of structurally equivalent domains in Rev1 and Polι. The N-digit is unique to Rev1 and is highlighted in dark blue. Also highlighted in dark blue is an extended loop in the Rev1 PAD (termed the G loop) that interacts with template G. (B) Secondary structure and domain topologies of Rev1 and Polι. The coloring scheme is the same as in Fig. 1.

Fig. 3.

Rev1-DNA-dCTP interactions. (A) A close-up view of the Rev1 active site region. The N-digit, fingers, and palm domains and the PAD are shown in dark blue, yellow, cyan, and green, respectively. The DNA is colored gray, and template G and incoming dCTP are shown in full atom coloring. The putative Mg2+ ions are dark blue, and the two displayed water molecules are colored magenta. Highlighted and labeled are the catalytic residues (27) (D362, D467, and E468), residues that interact with the triphosphate moiety of incoming dCTP (S402, Y405, R408, and K525), R324 that makes hydrogen bonds with dCTP base, L325 that pushes template G out of the DNA helix, and residues that interact with the extrahelical template G (M685, G686. K681, and W417). Note that template G and incoming dCTP partner with segments of Rev1. (B) Close-up views of template G–G-loop (left) and incoming dCTP-Arg324 (right) interactions. Dashed lines depict the network of direct and water-mediated hydrogen bonds (with distances in angstroms above the bonds).

The right-handed grip of palm, fingers, thumb, and the PAD on the template-primer is augmented in Rev1 by an “N-digit” at the N terminus (305 to 355) that interacts with incoming dCTP. The N-digit is composed of a loose α-loop substructure that fits into a shallow depression at the confluence of the palm, fingers, and PAD, connected to a long α helix that travels between the palm and the PAD and joins at the base of the palm (Figs. 1 and 2). The Rev1 PAD is exceptional in having an extra-long loop, a “G loop,” which interacts with template G. Compared with Polι-DNA-dNTP (2′-deoxynucleoside 5′-triphosphate) ternary complexes (18, 22), the Rev1 PAD as a whole is shifted toward the 3′ end of the template, which both creates a “route” for the N-digit helix and helps to position the G loop (670 to 688) (Fig. 2A).

The templating G and incoming dCTP do not pair with each other. The templating G is evicted from the DNA helix by Leu325 (from the N-digit) jutting into the DNA (Fig. 3), reminiscent of the flipping of DNA bases in DNA methyltransferases and glycosylases (23). Leu325 takes up much of the space vacated by templating G, whereas an adjoining residue, Arg324, makes a set of complementary hydrogen bonds with incoming dCTP (Fig. 3). The dCTP base tilts toward Arg324, whereas the dC base at the primer terminus tilts in the opposite direction to accommodate Leu325 (and to a lesser extent Leu328) within the DNA helix. The N-digit is indispensable for Rev1 function, as the protein encompassing residues 329 to 746 is completely inactive. This protein lacks residues 305 to 328 of the N-digit, among which are the residues Arg324 and Leu325.

The templating G swings out of the DNA helix (at ∼90°) and two hydrogen bonds are established between N7 and O6 at its “Hoogsteen edge” and the main-chain amides of Met685 and Gly686 on the G loop (Fig. 3). In addition, the O6 of G hydrogen bonds to a water molecule linked to Lys681, and the N3 makes an out-of-plane hydrogen bond with a water molecule that is fixed in position by Asp399, Trp417, and Lys681. The pattern of hydrogen bonding is such that only a G can optimally pair with the G loop. If the templating G were to be replaced by A, for example, the N6 of A would be incapable of making a hydrogen bond with the NH of Gly686, and there would also be electrostatic and steric repulsion from the positioning of two hydrogen bond donors opposite each other. Similarly, T or C in place of G would again lead to a loss of hydrogen bonds; and in the case of T, the C5 methyl group would also sterically clash with the G loop. In view of these steric constraints, it is not surprising therefore that Rev1 prefers an abasic site at the templating position rather than an A, C, or T.

In addition to hydrogen bonds with the G-loop main chain, the extruded templating G is also fixed in its extrahelical position by a set of van der Waals contacts. The guanine slips into a small hydrophobic pocket delineated by the Met685 side chain on one side and the aliphatic portion of Lys681 on the other (Fig. 3). It is also noteworthy that Gly686 occupies a region of the Ramachandran plot (ϕ = 66.0 and ψ = –163.1) more easily accessible to a glycine, which suggests that a glycine at this position is important in configuring the G loop. Taken together, the chemical basis for Rev1's G-template specificity is that the extrahelical G makes optimal hydrogen bonds with a segment of Rev1 instead of the incoming nucleotide.

Also, unlike other DNA polymerases, incoming dCTP pairs with an arginine rather than the templating base. Two hydrogen bonds are established between N3 and O2 at the Watson-Crick edge of dCTP and the Nη2 and Nϵ donor groups of Arg324, respectively (Figs. 3 and 4). Arg324 extends from the N-digit and is buttressed as the surrogate templating residue by a network of hydrogen bonds with Asp399 and the 5′ phosphate of the ejected template G. The pattern of hydrogen bonding between dCTP and Arg324 isagainsuchthatsubstitutionbyany other incoming nucleotide would lead to loss of hydrogen bonds, as well as unfavorable electrostatic and steric intrusion. If dCTP were to be replaced by 2′-deoxythymidine 5′-triphosphate (dTTP), it would position two hydrogen bond donors opposite each other [N3-H(T)-H-Nη2(Arg324)], whereas substitution by dGTP or dATP would be even more severe, with the larger purine colliding with the guanidinium group of Arg324. By pairing dCTP with an arginine, Rev1 maintains specificity for dCTP over other incoming nucleotides, even opposite an abasic site.

Fig. 4.

Comparison between Arg:dCTP pairing in Rev1 (top), Hoogsteen dG:dCTP base-pairing in Polι (middle), and standard Watson-Crick dG:dCTP base-pairing in replicative DNA polymerases (bottom). The atoms and hydrogen bonding distances are labeled.

The structure suggests a specific mechanism for Rev1's ability to promote replication through N2-adducted guanines that obstruct replication. The N2 group of G can conjugate with a variety of endogenously formed adducts. Indeed, Rev1 has recently been shown to promote replication through an N2-adducted G, derived from acrolein. Acrolein, an α,β-unsaturated aldehyde, is generated in vivo as the end product of lipid peroxidation and during oxidation of polyamines. The reaction of acrolein with the N2 of G in DNA followed by ring closure at N1 leads to the formation of the cyclic adduct γ-HOPdG (fig. S2), which presents a strong block to synthesis by DNA polymerases. Rev1, however, incorporates a C opposite this lesion as efficiently as opposite an undamaged G (12). The exclusion of template G from the DNA helix places the N2 of G in a large (solvent-filled) void between the PAD and the fingers domain (Fig. 3), where an adduct such as γ-HOPdG would be sterically unhindered (fig. S2). Indeed, one major role of Rev1 DNA synthetic activity based on an extrahelical G would be to promote replication through a variety of N2-guanine adducts that sterically impinge on the minor groove. The incorporation of a correct nucleotide opposite an N2-adducted guanine is ensured by the pairing of dCTP with an arginine. The structure also correlates well with the inhibitory effect that lesions such as O6-methylguanine and 8-oxoguanine have on Rev1's ability to incorporate C (2). Unlike the N2 group, the O6 and C8 atoms are relatively buried when template G is evicted from the DNA helix, and almost any adduct at these positions will invariably clash with Rev1.

In the transfer RNA (tRNA) CCA-adding enzyme, both the tRNA backbone and the protein contribute to the specificity of the incoming nucleotide (24). The Rev1 structure presents a mechanism for DNA polymerization in which specificity for both the templating and the incoming nucleotide is provided by the protein rather than the DNA. Eukaryotic translesion synthesis polymerases thus use a variety of means of DNA polymerization, which include Watson-Crick base-pairing by Pols η (25) and κ (26), Hoogsteen base-pairing by Polι (18, 22), and protein template–directed synthesis by Rev1 (Fig. 4).

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Tables S1

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article