DMESH: A Structure-Preserving Diffusion Model for 3-D Mesh Denoising.

Denoising diffusion models have shown a powerful capacity for generating high-quality image samples by progressively removing noise. Inspired by this, we present a diffusion-based mesh denoiser that progressively removes noise from mesh. In general, the iterative algorithm of diffusion models attempts to manipulate the overall structure and fine details of target meshes simultaneously. For this reason, it is difficult to apply the diffusion process to a mesh denoising task that removes artifacts while maintaining a structure. To address this, we formulate a structure-preserving diffusion process. Instead of diffusing the mesh vertices to be distributed as zero-centered isotopic Gaussian distribution, we diffuse each vertex into a specific noise distribution, in which the entire structure can be preserved. In addition, we propose a topology-agnostic mesh diffusion model by projecting the vertex into multiple 2-D viewpoints to efficiently learn the diffusion using a deep network. This enables the proposed method to learn the diffusion of arbitrary meshes that have an irregular topology. Finally, the denoised mesh can be obtained via refinement based on 2-D projections obtained from reverse diffusion. Through extensive experiments, we demonstrate that our method outperforms the state-of-the-art mesh denoising methods in both quantitative and qualitative evaluations.

Role of the CCL20/CCR6 axis in tubular epithelial cell injury: Kidney-specific translational insights from acute kidney injury to chronic kidney disease.

This study investigated the role of the axis involving chemokine receptor 6 (CCR6) and its ligand chemokine (C-C motif) ligand 20 (CCL20) in acute kidney disease (AKD) using an ischemia-reperfusion injury (IRI) model. The model was established by clamping the unilateral renal artery pedicle of C57BL/6 mice for 30 min, followed by evaluation of CCL20/CCR6 expression at 4 weeks post-IRI. In vitro studies were conducted to examine the effects of hypoxia and H2 O2 -induced oxidative stress on CCL20/CCR6 expression in kidney tissues of patients with AKD and chronic kidney disease (CKD). Tubular epithelial cell apoptosis was more severe in C57BL/6 mice than in CCL20 antibody-treated mice, and CCR6, NGAL mRNA, and IL-8 levels were higher under hypoxic conditions. CCL20 blockade ameliorated apoptotic damage in a dose-dependent manner under hypoxia and reactive oxygen species injury. CCR6 expression in IRI mice indicated that the disease severity was similar to that in patients with the AKD phenotype. Morphometry of CCL20/CCR6 expression revealed a higher likelihood of CCR6+ cell presence in CKD stage 3 patients than in stage 1-2 patients. Kidney tissues of patients with CKD frequently contained CCL20+ cells, which were positively correlated with interstitial inflammation. CCL20/CCR6 levels were increased in fibrotic kidneys at 4 and 8 weeks after 5/6 nephrectomy. These findings suggest that modulating the CCL20/CCR6 pathway is a potential therapeutic strategy for managing the progression of AKD to CKD.

VISGRADER: Automatic Grading of D3 Visualizations.

Manually grading D3 data visualizations is a challenging endeavor, and is especially difficult for large classes with hundreds of students. Grading an interactive visualization requires a combination of interactive, quantitative, and qualitative evaluation that are conventionally done manually and are difficult to scale up as the visualization complexity, data size, and number of students increase. We present VISGRADER, a first-of-its kind automatic grading method for D3 visualizations that scalably and precisely evaluates the data bindings, visual encodings, interactions, and design specifications used in a visualization. Our method enhances students' learning experience, enabling them to submit their code frequently and receive rapid feedback to better inform iteration and improvement to their code and visualization design. We have successfully deployed our method and auto-graded D3 submissions from more than 4000 students in a visualization course at Georgia Tech, and received positive feedback for expanding its adoption.

Genotoxic effects of the major alkylation damage N7-methylguanine and methyl formamidopyrimidine.

Various alkylating agents are known to preferentially modify guanine in DNA, resulting in the formation of N7-alkylguanine (N7-alkylG) and the imidazole ring opened alkyl-formamidopyrimidine (alkyl-FapyG) lesions. Evaluating the mutagenic effects of N7-alkylG has been challenging due to the instability of the positively charged N7-alkylG. To address this issue, we developed a 2'-fluorine-mediated transition-state destabilization approach, which stabilizes N7-alkylG and prevents spontaneous depurination. We also developed a postsynthetic conversion of 2'-F-N7-alkylG DNA into 2'-F-alkyl-FapyG DNA. Using these methods, we incorporated site-specific N7-methylG and methyl-FapyG into pSP189 plasmid and determined their mutagenic properties in bacterial cells using the supF-based colony screening assay. The mutation frequency of N7-methylG was found to be less than 0.5%. Our crystal structure analysis revealed that N7-methylation did not significantly alter base pairing properties, as evidenced by a correct base pairing between 2'-F-N7-methylG and dCTP in Dpo4 polymerase catalytic site. In contrast, the mutation frequency of methyl-FapyG was 6.3%, highlighting the mutagenic nature of this secondary lesion. Interestingly, all mutations arising from methyl-FapyG in the 5'-GGT(methyl-FapyG)G-3' context were single nucleotide deletions at the 5'-G of the lesion. Overall, our results demonstrate that 2'-fluorination technology is a useful tool for studying the chemically labile N7-alkylG and alkyl-FapyG lesions.

Characterization of relativistic electron-positron beams produced with laser-accelerated GeV electrons.

The characterization of an electron-positron beam generated from the interaction of a multi-GeV electron beam with a lead plate is performed using GEANT4 simulations. The dependence of the positron beam size on driver electron beam energy and lead converter thickness is investigated in detail. A pancake-like positron beam structure is generated with a monoenergetic multi-GeV driver electron beam, with the results indicating that a 5 GeV driver electron beam with 1 nC charge can generate a positron beam with a density of 1015-1016 cm-3 at one radiation length of lead. In addition, we find that electron-positron beams generated using above-GeV electron beams have neutralities greater than 0.3 at one radiation length of lead, whereas neutralities of 0.2 are observed when using a 200 MeV electron beam. The possibility of observing plasma instabilities in experiments is also examined by comparing the plasma skin depth with the electron-positron beam size. A quasi-neutral electron-positron plasma can be produced in the interaction between a 1 nC, 5 GeV electron beam and lead with a thickness of five radiation lengths. Our findings will aid in analyzing and interpreting laser-produced electron-positron plasma for laboratory astrophysics research.

Reproductive performance and sex ratio adjustment of the wild boar (Sus scrofa) in South Korea.

The wild boar (Sus scrofa), a polygynous species, rapidly expanded its geographical range and increased its population size in South Korea following the extinction of large carnivores and changes to rural environments. Understanding wild boar reproductive traits and strategies is essential for their effective management; however, studies in this area are lacking. Using samples collected from hunting bags, the relationships between 1) litter size and female weight and 2) fetal sex ratio and female body condition were examined to understand wild boar life-history strategies. Wild boars showed a seasonal breeding pattern that maximized reproduction. Litter size (mean = 5.7 ± 1.7) was correlated with female weight, whereas fetal sex ratio was not explained by female body condition. However, the heaviest ranked fetuses within the litters were male-biased. Wild boars aged three years or less accounted for 90% of the total population, and sexual dimorphism developed from two years of age. Considering that their reproductive strategy is more effective (i.e., early gestation and large litter size) than that of other polygynous species, the Trivers-Willard model was not supported for the wild boars in this study. Instead, females adjusted the sex of the heaviest fetus in the litter to maximize lifetime reproductive success.

Structure of the major oxidative damage 7,8-dihydro-8-oxoguanine presented into a catalytically competent DNA glycosylase.

If left unrepaired, the major oxidative DNA lesion 7,8-dihydro-8-oxoguanine (oxoG) promotes G-to-T transversions by favorably adopting a syn conformation and base pairing with dATP during replication. The human oxoG DNA glycosylase hOGG1 senses and removes oxoG amid millions-fold excess of guanine, thereby counteracting the genotoxic effects of the major oxidative damage. Crystal structures of hOGG1 in complex with oxoG-containing DNA have provided key insights into the lesion recognition and catalysis mechanisms of the enzyme. These lesion-recognition complex (LRC) structures typically involve a catalytically inactive hOGG1 mutant, where one of the catalytic-site amino acid residues is mutated to prevent the cleavage of oxoG. The use of a catalytically incompetent hOGG1 mutant has thus precluded understanding of unscathed interactions between oxoG and hOGG1 catalytic site as well as interactions among catalytic-site amino acid residues. As an orthogonal approach to visualize such interactions, we have co-crystallized a catalytically competent hOGG1 bound to 2'-fluoro-oxodG-containing DNA, a transition state destabilizing inhibitor that binds hOGG1 but is not processed by the enzyme. In this fluorinated lesion-recognition complex (FLRC), the 8-oxo moiety of oxoG is recognized by Gly42 and the Watson-Crick edge of oxoG is contacted by Gln315 and Pro266. The previously observed salt bridge between Lys249 and Cys253 is lacking in the FLRC, suggesting Lys249 is primed by Cys253 and poised for nucleophilic attack on C1' of oxodG. Overall, hOGG1 FLRC marks the first structure of oxoG presented into an intact catalytic site of hOGG1 and provides complementary insights into the glycosylase mechanisms of the enzyme.

Technical Consideration towards Robust 3D Reconstruction with Multi-View Active Stereo Sensors.

It is possible to construct cost-efficient three-dimensional (3D) or four-dimensional (4D) scanning systems using multiple affordable off-the-shelf RGB-D sensors to produce high-quality reconstructions of 3D objects. However, the quality of these systems' reconstructions is sensitive to a number of factors in reconstruction pipelines, such as multi-view calibration, depth estimation, 3D reconstruction, and color mapping accuracy, because the successive pipelines to reconstruct 3D meshes from multiple active stereo sensors are strongly correlated with each other. This paper categorizes the pipelines into sub-procedures and analyze various factors that can significantly affect reconstruction quality. Thus, this paper provides analytical and practical guidelines for high-quality 3D reconstructions with off-the-shelf sensors. For each sub-procedure, this paper shows comparisons and evaluations of several methods using data captured by 18 RGB-D sensors and provide analyses and discussions towards robust 3D reconstruction. Through various experiments, it has been demonstrated that significantly more accurate 3D scans can be obtained with the considerations along the pipelines. We believe our analyses, benchmarks, and guidelines will help anyone build their own studio and their further research for 3D reconstruction.

A Comparison and Evaluation of Stereo Matching on Active Stereo Images.

The relationship between the disparity and depth information of corresponding pixels is inversely proportional. Thus, in order to accurately estimate depth from stereo vision, it is important to obtain accurate disparity maps, which encode the difference between horizontal coordinates of corresponding image points. Stereo vision can be classified as either passive or active. Active stereo vision generates pattern texture, which passive stereo vision does not have, on the image to fill the textureless regions. In passive stereo vision, many surveys have discovered that disparity accuracy is heavily reliant on attributes, such as radiometric variation and color variation, and have found the best-performing conditions. However, in active stereo matching, the accuracy of the disparity map is influenced not only by those affecting the passive stereo technique, but also by the attributes of the generated pattern textures. Therefore, in this paper, we analyze and evaluate the relationship between the performance of the active stereo technique and the attributes of pattern texture. When evaluating, experiments are conducted under various settings, such as changing the pattern intensity, pattern contrast, number of pattern dots, and global gain, that may affect the overall performance of the active stereo matching technique. Through this evaluation, our discovery can act as a noteworthy reference for constructing an active stereo system.

Effects of N7-Alkylguanine Conformation and Metal Cofactors on the Translesion Synthesis by Human DNA Polymerase η.

Non-enzymatic alkylation on DNA often generates N7-alkyl-2'-deoxyguanosine (N7alkylG) adducts as major lesions. N7alkylG adducts significantly block replicative DNA polymerases and can be bypassed by translesion synthesis (TLS) polymerases such as polymerase η (polη). To gain insights into the bypass of N7alkylG by TLS polymerases, we conducted kinetic and structural studies of polη catalyzing across N7BnG, a genotoxic lesion generated by the carcinogenic N-nitrosobenzylmethylamine. The presence of templating N7BnG in the polη catalytic site decreased the replication fidelity by ∼9-fold, highlighting the promutagenicity of N7BnG. The catalytic efficiency for dCTP incorporation opposite N7BnG decreased ∼22-fold and ∼7-fold compared to the incorporation opposite undamaged guanine in the presence of Mg2+ and Mn2+, respectively. A crystal structure of the complexes grown with polη, templating N7BnG, incoming dCTP, and Mg2+ ions showed the lack of the incoming nucleotide and metal cofactors in the polη catalytic site. Interestingly, the templating N7BnG adopted a syn conformation, which has not been observed in the published N7alkylG structures. The preferential formation of syn-N7BnG conformation at the templating site may deter the binding of an incoming dCTP, causing the inefficient bypass by polη. In contrast, the use of Mn2+ in place of Mg2+ in co-crystallization yielded a ternary complex displaying an anti-N7BnG:dCTP base pair and catalytic metal ions, which would be a close mimic of a catalytically competent state. We conclude that certain bulky N7-alkylG lesions can slow TLS polymerase-mediated bypass by adopting a catalytically unfavorable syn conformation in the replicating base pair site.

Genotoxic C8-Arylamino-2'-deoxyadenosines Act as Latent Alkylating Agents to Induce DNA Interstrand Cross-Links.

DNA interstrand cross-links (ICLs) are extremely deleterious and structurally diverse, driving the evolution of ICL repair pathways. Discovering ICL-inducing agents is, thus, crucial for the characterization of ICL repair pathways and Fanconi anemia, a genetic disease caused by mutations in ICL repair genes. Although several studies point to oxidative stress as a cause of ICLs, oxidative stress-induced cross-linking events remain poorly characterized. Also, polycyclic aromatic amines, potent environmental carcinogens, have been implicated in producing ICLs, but their identities and sequences are unknown. To close this knowledge gap, we tested whether ICLs arise by the oxidation of 8-arylamino-2'-deoxyadenosine (ArNHdA) lesions, adducts produced by arylamino carcinogens. Herein, we report that ArNHdA acts as a latent cross-linking agent to generate ICLs under oxidative conditions. The formation of an ICL from 8-aminoadenine, but not from 8-aminoguanine, highlights the specificity of 8-aminopurine-mediated ICL production. Under the influence of the reactive oxygen species (ROS) nitrosoperoxycarbonate, ArNHdA (Ar = biphenyl, fluorenyl) lesions were selectively oxidized to generate ICLs. The cross-linking reaction may occur between the C2-ArNHdA and N2-dG, presumably via oxidation of ArNHdA into a reactive diiminoadenine intermediate followed by the nucleophilic attack of the N2-dG on the diiminoadenine. Overall, ArNHdA-mediated ICLs represent rare examples of ROS-induced ICLs and polycyclic aromatic amine-mediated ICLs. These results reveal novel cross-linking chemistry and the genotoxic effects of arylamino carcinogens and support the hypothesis that C8-modified adenines with low redox potential can cause ICLs in oxidative stress.

Multi-EPL: Accurate multi-source domain adaptation.

Given multiple source datasets with labels, how can we train a target model with no labeled data? Multi-source domain adaptation (MSDA) aims to train a model using multiple source datasets different from a target dataset in the absence of target data labels. MSDA is a crucial problem applicable to many practical cases where labels for the target data are unavailable due to privacy issues. Existing MSDA frameworks are limited since they align data without considering labels of the features of each domain. They also do not fully utilize the target data without labels and rely on limited feature extraction with a single extractor. In this paper, we propose Multi-EPL, a novel method for MSDA. Multi-EPL exploits label-wise moment matching to align the conditional distributions of the features for the labels, uses pseudolabels for the unavailable target labels, and introduces an ensemble of multiple feature extractors for accurate domain adaptation. Extensive experiments show that Multi-EPL provides the state-of-the-art performance for MSDA tasks in both image domains and text domains, improving the accuracy by up to 13.20%.

Unsupervised multi-source domain adaptation with no observable source data.

Given trained models from multiple source domains, how can we predict the labels of unlabeled data in a target domain? Unsupervised multi-source domain adaptation (UMDA) aims for predicting the labels of unlabeled target data by transferring the knowledge of multiple source domains. UMDA is a crucial problem in many real-world scenarios where no labeled target data are available. Previous approaches in UMDA assume that data are observable over all domains. However, source data are not easily accessible due to privacy or confidentiality issues in a lot of practical scenarios, although classifiers learned in source domains are readily available. In this work, we target data-free UMDA where source data are not observable at all, a novel problem that has not been studied before despite being very realistic and crucial. To solve data-free UMDA, we propose DEMS (Data-free Exploitation of Multiple Sources), a novel architecture that adapts target data to source domains without exploiting any source data, and estimates the target labels by exploiting pre-trained source classifiers. Extensive experiments for data-free UMDA on real-world datasets show that DEMS provides the state-of-the-art accuracy which is up to 27.5% point higher than that of the best baseline.

AUBER: Automated BERT regularization.

How can we effectively regularize BERT? Although BERT proves its effectiveness in various NLP tasks, it often overfits when there are only a small number of training instances. A promising direction to regularize BERT is based on pruning its attention heads with a proxy score for head importance. However, these methods are usually suboptimal since they resort to arbitrarily determined numbers of attention heads to be pruned and do not directly aim for the performance enhancement. In order to overcome such a limitation, we propose AUBER, an automated BERT regularization method, that leverages reinforcement learning to automatically prune the proper attention heads from BERT. We also minimize the model complexity and the action search space by proposing a low-dimensional state representation and dually-greedy approach for training. Experimental results show that AUBER outperforms existing pruning methods by achieving up to 9.58% better performance. In addition, the ablation study demonstrates the effectiveness of design choices for AUBER.

Insights into the substrate discrimination mechanisms of methyl-CpG-binding domain 4.

G:T mismatches, the major mispairs generated during DNA metabolism, are repaired in part by mismatch-specific DNA glycosylases such as methyl-CpG-binding domain 4 (MBD4) and thymine DNA glycosylase (TDG). Mismatch-specific DNA glycosylases must discriminate the mismatches against million-fold excess correct base pairs. MBD4 efficiently removes thymine opposite guanine but not opposite adenine. Previous studies have revealed that the substrate thymine is flipped out and enters the catalytic site of the enzyme, while the estranged guanine is stabilized by Arg468 of MBD4. To gain further insights into the mismatch discrimination mechanism of MBD4, we assessed the glycosylase activity of MBD4 toward various base pairs. In addition, we determined a crystal structure of MBD4 bound to T:O6-methylguanine-containing DNA, which suggests the O6 and N2 of purine and the O4 of pyrimidine are required to be a substrate for MBD4. To understand the role of the Arg468 finger in catalysis, we evaluated the glycosylase activity of MBD4 mutants, which revealed the guanidinium moiety of Arg468 may play an important role in catalysis. D560N/R468K MBD4 bound to T:G mismatched DNA shows that the side chain amine moiety of the Lys stabilizes the flipped-out thymine by a water-mediated phosphate pinching, while the backbone carbonyl oxygen of the Lys engages in hydrogen bonds with N2 of the estranged guanine. Comparison of various DNA glycosylase structures implies the guanidinium and amine moieties of Arg and Lys, respectively, may involve in discriminating between substrate mismatches and nonsubstrate base pairs.

Insights into the mismatch discrimination mechanism of Y-family DNA polymerase Dpo4.

Nucleobases within DNA are attacked by reactive oxygen species to produce 7,8-dihydro-8-oxoguanine (oxoG) and 7,8-dihydro-8-oxoadenine (oxoA) as major oxidative lesions. The high mutagenicity of oxoG is attributed to the lesion's ability to adopt syn-oxoG:anti-dA with Watson-Crick-like geometry. Recent studies have revealed that Sulfolobus solfataricus P2 DNA polymerase IV (Dpo4) inserts nucleotide opposite oxoA in an error-prone manner and accommodates syn-oxoA:anti-dGTP with Watson-Crick-like geometry, highlighting a promutagenic nature of oxoA. To gain further insights into the bypass of oxoA by Dpo4, we have conducted kinetic and structural studies of Dpo4 extending oxoA:dT and oxoA:dG by incorporating dATP opposite templating dT. The extension past oxoA:dG was ∼5-fold less efficient than that past oxoA:dT. Structural studies revealed that Dpo4 accommodated dT:dATP base pair past anti-oxoA:dT with little structural distortion. In the Dpo4-oxoA:dG extension structure, oxoA was in an anti conformation and did not form hydrogen bonds with the primer terminus base. Unexpectedely, the dG opposite oxoA exited the primer terminus site and resided in an extrahelical site, where it engaged in minor groove contacts to the two immediate upstream bases. The extrahelical dG conformation appears to be induced by the stabilization of anti-oxoA conformation via bifurcated hydrogen bonds with Arg332. This unprecedented structure suggests that Dpo4 may use Arg332 to sense 8-oxopurines at the primer terminus site and slow the extension from the mismatch by promoting anti conformation of 8-oxopurines.

DNA interstrand cross-links induced by the major oxidative adenine lesion 7,8-dihydro-8-oxoadenine.

Oxidative damage to DNA generates 7,8-dihydro-8-oxoguanine (oxoG) and 7,8-dihydro-8-oxoadenine (oxoA) as two major lesions. Despite the comparable prevalence of these lesions, the biological effects of oxoA remain poorly characterized. Here we report the discovery of a class of DNA interstrand cross-links (ICLs) involving oxidized nucleobases. Under oxidative conditions, oxoA, but not oxoG, readily reacts with an opposite base to produce ICLs, highlighting a latent alkylating nature of oxoA. Reactive halogen species, one-electron oxidants, and the myeloperoxidase/H2O2/Cl- system induce oxoA ICLs, suggesting that oxoA-mediated cross-links may arise endogenously. Nucleobase analog studies suggest C2-oxoA is covalently linked to N2-guanine and N3-adenine for the oxoA-G and oxoA-A ICLs, respectively. The oxoA ICLs presumably form via the oxidative activation of oxoA followed by the nucleophilic attack by an opposite base. Our findings provide insights into oxoA-mediated mutagenesis and contribute towards investigations of oxidative stress-induced ICLs and oxoA-based latent alkylating agents.

Fabrication of Ag nanorods on micropost array for a metal-enhanced fluorescence substrate with a high signal-to-background ratio.

Selective fabrication of metallic nanostructures at the spotting area is required to increase the signal-to-background noise ratio (SBR) of the metal-enhanced fluorescence (MEF) substrate. As a simple and cost-effective fabrication method for MEF substrate with high SBR, a glancing angle deposition (GLAD) process of Ag material on the UV-imprinted micropost array (50 µm in height, 300 µm in diameter, and 600 µm in pitch) was proposed to selectively fabricate Ag nanorods on the top of micropost structure (spotting area). Ag nanorod formation at the bottom of the micropost decreased as the deposition angle in Ag GLAD increased. A deposition angle of 89° and deposition thickness of 500 nm were selected as the optimum GLAD conditions to maximize the SBR. The optimum Ag nanorods on micropost array (AgNMPA) MEF substrate provided 71-fold fluorescence signal enhancement and 25-times higher SBR than the bare glass substrate. It also provided 7-times higher SBR than the Ag nanorod MEF substrate, which has a similar Ag nanorod structure but is not selectively formed. The detection limit of AgNMPA was 16- and 4-times lower than that of the amine-functionalized glass substrate and commercial epoxy slide, respectively. Although the fluorescence signal of AgNMPA was similar to that of Ag nanorod substrate, the detection limit was 2-times lower because of the low signal standard deviation caused by the low background noise and clear spot shape.

Fabrication of a Plasmonic Nanoantenna Array Using Metal Deposition on Polymer Nanoimprinted Nanodots for an Enhanced Fluorescence Substrate.

A simple and cost-effective method is proposed herein for a plasmonic nanoantenna array (PNAA) for the fabrication of metal-enhanced fluorescence (MEF) substrates in which fluorophores interact with the enhanced electromagnetic field generated by a localized surface plasmon to provide a higher fluorescence signal. The PNAA is fabricated by the deposition of a silver (Ag) layer on an ultraviolet (UV) nanoimprinted nanodot array with a pitch of 400 nm, diameter of 200 nm, and height of 100 nm. During deposition, raised Ag nanodisks and a lower Ag layer are, respectively, formed on the top and bottom of the imprinted nanodot array, and the gap between these Ag layers acts as a plasmonic nanoantenna. Since the thickness of the gap within the PNAA is influenced by the thickness of Ag deposition, the effects of the latter upon the geometrical properties of the fabricated PNAA are examined, and the electromagnetic field intensity distributions of PNAAs with various Ag thicknesses are simulated. Finally, the fluorescence enhancement factor (FEF) of the fabricated PNAA MEF substrate is measured using spotted Cy5-conjugated streptavidin to indicate a maximum enhancement factor of ~22× for the PNAA with an Ag layer thickness of 75 nm. The experimental results are shown to match the simulated results.

Structural insights into the bypass of the major deaminated purines by translesion synthesis DNA polymerase.

The exocyclic amines of nucleobases can undergo deamination by various DNA damaging agents such as reactive oxygen species, nitric oxide, and water. The deamination of guanine and adenine generates the promutagenic xanthine and hypoxanthine, respectively. The exocyclic amines of bases in DNA are hydrogen bond donors, while the carbonyl moiety generated by the base deamination acts as hydrogen bond acceptors, which can alter base pairing properties of the purines. Xanthine is known to base pair with both cytosine and thymine, while hypoxanthine predominantly pairs with cytosine to promote A to G mutations. Despite the known promutagenicity of the major deaminated purines, structures of DNA polymerase bypassing these lesions have not been reported. To gain insights into the deaminated-induced mutagenesis, we solved crystal structures of human DNA polymerase η (polη) catalyzing across xanthine and hypoxanthine. In the catalytic site of polη, the deaminated guanine (i.e., xanthine) forms three Watson-Crick-like hydrogen bonds with an incoming dCTP, indicating the O2-enol tautomer of xanthine involves in the base pairing. The formation of the enol tautomer appears to be promoted by the minor groove contact by Gln38 of polη. When hypoxanthine is at the templating position, the deaminated adenine uses its O6-keto tautomer to form two Watson-Crick hydrogen bonds with an incoming dCTP, providing the structural basis for the high promutagenicity of hypoxanthine.