Enhancement of NEIL1 Protein-initiated Oxidized DNA Base Excision Repair by Heterogeneous Nuclear Ribonucleoprotein U (hnRNP-U) via Direct Interaction*

Background: Oxidized bases in mammalian genome are repaired via base excision repair (BER) process that utilizes both common repair and noncanonical proteins. Results: hnRNP-U stimulates NEIL1-initiated BER via direct interaction and is required for enhancing oxidative stress-induced repair. Conclusion: NEIL1's interaction with hnRNP-U is critical for regulating oxidized genome damage repair after oxidative stress. Significance: The RNA-binding protein hnRNP-U plays a role in maintaining genomic integrity. Repair of oxidized base lesions in the human genome, initiated by DNA glycosylases, occurs via the base excision repair pathway using conserved repair and some non-repair proteins. However, the functions of the latter noncanonical proteins in base excision repair are unclear. Here we elucidated the role of heterogeneous nuclear ribonucleoprotein-U (hnRNP-U), identified in the immunoprecipitate of human NEIL1, a major DNA glycosylase responsible for oxidized base repair. hnRNP-U directly interacts with NEIL1 in vitro via the NEIL1 common interacting C-terminal domain, which is dispensable for its enzymatic activity. Their in-cell association increases after oxidative stress. hnRNP-U stimulates the NEIL1 in vitro base excision activity for 5-hydroxyuracil in duplex, bubble, forked, or single-stranded DNA substrate, primarily by enhancing product release. Using eluates from FLAG-NEIL1 immunoprecipitates from human cells, we observed 3-fold enhancement in complete repair activity after oxidant treatment. The lack of such enhancement in hnRNP-U-depleted cells suggests its involvement in repairing enhanced base damage after oxidative stress. The NEIL1 disordered C-terminal region binds to hnRNP-U at equimolar ratio with high affinity (Kd = ∼54 nm). The interacting regions in hnRNP-U, mapped to both termini, suggest their proximity in the native protein; these are also disordered, based on PONDR (Predictor of Naturally Disordered Regions) prediction and circular dichroism spectra. Finally, depletion of hnRNP-U and NEIL1 epistatically sensitized human cells at low oxidative genome damage, suggesting that the hnRNP-U protection of cells after oxidative stress is largely due to enhancement of NEIL1-mediated repair.

Repair of oxidized base lesions in the human genome, initiated by DNA glycosylases, occurs via the base excision repair pathway using conserved repair and some non-repair proteins. However, the functions of the latter noncanonical proteins in base excision repair are unclear. Here we elucidated the role of heterogeneous nuclear ribonucleoprotein-U (hnRNP-U), identified in the immunoprecipitate of human NEIL1, a major DNA glycosylase responsible for oxidized base repair. hnRNP-U directly interacts with NEIL1 in vitro via the NEIL1 common interacting C-terminal domain, which is dispensable for its enzymatic activity. Their in-cell association increases after oxidative stress. hnRNP-U stimulates the NEIL1 in vitro base excision activity for 5-hydroxyuracil in duplex, bubble, forked, or single-stranded DNA substrate, primarily by enhancing product release. Using eluates from FLAG-NEIL1 immunoprecipitates from human cells, we observed 3-fold enhancement in complete repair activity after oxidant treatment. The lack of such enhancement in hnRNP-U-depleted cells suggests its involvement in repairing enhanced base damage after oxidative stress. The NEIL1 disordered C-terminal region binds to hnRNP-U at equimolar ratio with high affinity (K d ‫؍‬ ϳ54 nM). The interacting regions in hnRNP-U, mapped to both termini, suggest their proximity in the native protein; these are also disordered, based on PONDR (Predictor of Naturally Disordered Regions) prediction and circular dichroism spectra. Finally, depletion of hnRNP-U and NEIL1 epistatically sensitized human cells at low oxidative genome damage, suggesting that the hnRNP-U protection of cells after oxidative stress is largely due to enhancement of NEIL1-mediated repair.
The genomes of aerobic organisms are constantly exposed to reactive oxygen species (ROS), 2 which are formed endogenously, mostly as byproducts of cellular respiration, and also induced by environmental agents and during inflammatory responses (1)(2)(3). ROS induce oxidatively modified bases (e.g. 8-oxoguanine, dihydrouracil, thymine glycol, 5-hydroxyuracil (5-OHU), etc.), abasic (AP) sites, and DNA single-strand breaks; these have been implicated in the etiology of various diseases including cancer, degenerative neurological disorders, and arthritis, and in aging (4 -9). If left unrepaired, most of these lesions are mutagenic and/or cytotoxic and are repaired via the evolutionarily conserved base excision/single-strand break repair (BER/SSBR) pathway. Repair of oxidized bases in the mammalian genome is initiated with their excision by one of five DNA glycosylases (DGs): OGG1, NTH1, and NEIL 1-3 (10,11). Oxidized base-specific DGs have intrinsic AP lyase activity, as a result of which the AP site produced after damaged base excision is cleaved via ␤ or ␤␦ elimination to generate 3Ј-phospho ␣,␤-unsaturated aldehyde and 3Ј-phosphate, respectively (10). NEIL1 and -2 belonging to the Nei family of DGs are distinct from OGG1/NTH1 in the Nth family because of their ␤␦-lyase activity and ability to excise base lesions from single-stranded DNA substrate in vitro (12). As we had shown earlier, the 3Ј-phosphate terminus generated by NEIL1 and -2 is removed by polynucleotide kinase/phosphatase in the second step of BER, whereas the 3Ј-phospho ␣,␤-unsaturated aldehyde phosphate produced by OGG1/NTH1 is removed by AP-endonuclease (APE1) (12). Subsequent repair steps, which occur via the one-nucleotide gap filling (named SN-BER), are common for both families of DGs in most cells. Alternatively, long patch repair with incorporation of 2-8 nucleotides could occur at the lesion site in replicating cells where the replication enzymes are co-opted for the repair (13)(14)(15).
Recent evidence indicates the involvement of noncanonical proteins in BER, in particular, RNA-binding proteins that lack known repair-related activity (16 -18). These were shown to modulate the repair activity of DGs via direct interaction. For example, NEIL2 interacts functionally with Y-box-binding protein (YB-1), and also with RNA binding activity (17), under conditions of oxidative stress. The immunoprecipitates (IPs) of both NEILs contain hnRNP-U. We previously characterized the involvement of NEIL2 in the preferential repair of transcribed genes, where NEIL2 interacts functionally with both RNA polymerase II and hnRNP-U (19). Unlike NEIL2, NEIL1 is up-regulated during the S-phase and interacts functionally with DNA replication proteins, including proliferating cell nuclear antigen (PCNA), replication protein A (RPA), FEN-1, and Werner helicase (WRN), etc. Based on these results, we have proposed that NEIL1 is preferentially involved in replication-coordinated BER (20 -23). Because of the differential functions and mechanisms of NEIL1 versus NEIL2, we decided to investigate the specific role of hnRNP-U in NEIL1-mediated repair.
In addition to its role in mRNA processing and transport, hnRNP-U has some unique functions, as also observed for other members of the hnRNP family (24,25). hnRNP-U was independently identified and named scaffold attachment factor (SAF)-A on the basis of its strong binding to nuclear scaffold/ matrix with affinity for A-T-rich DNA (26). With a molecular mass of 90 kDa, it is the largest member of the abundant hnRNP family (24). Interestingly, hnRNP-U was also found to be associated with WT1 (Wilms tumor 1) protein and suggested to be a potential Wilms tumor gene (27). hnRNP-U interacts with MDM2, an E3 ubiquitin ligase involved in degradation of p53, a central player in DNA damage response and also in DNA repair (28). The hnRNP-U interaction with some proteins may be regulated via its phosphorylation by DNA-protein kinase, which is activated by double-strand breaks in the genome (29,30). In this study, we have documented the physical and functional interaction of NEIL1 with hnRNP-U, which enhances cell survival after oxidative stress by increasing NEIL1-mediated oxidized base repair.

EXPERIMENTAL PROCEDURES
Oligonucleotide and Plasmid Substrates-A 51-mer oligonucleotide containing 5-OHU, a major oxidative deamination product of C in DNA, at position 26 from the 5Ј-end, the complementary oligonucleotide containing G opposite the lesion (Sigma), was used for generating duplex oligonucleotides and partial duplexes containing a bubble or fork structure (supplemental Table S1). To produce 5Ј-32 P-labeled substrates, the single-stranded (ss) oligonucleotides were labeled at their 5Ј termini with [␥-32 P]ATP using T 4 -polynucleotide kinase prior to annealing. The labeled substrates were separated from unincorporated radioactivity by chromatography on Sephadex G25. The pUC19CPD plasmid substrate containing a single 5-OHU for complete repair assay was generated as described previously (31,32).
Expression and Purification of Recombinant Proteins-Recombinant wild type (WT) NEIL1, its truncated polypeptides, polynucleotide kinase/phosphatase, DNA polymerase ␤, and DNA ligase III␣ were purified to homogeneity from Escherichia coli as described previously (12,(32)(33)(34). His-tagged WT human hnRNP-U (the plasmid was a kind gift from Gourisankar Ghosh, University of California, San Diego) was expressed in BL-21 Codon-Plus E. coli (Stratagene) and purified from the bacterial extract by affinity chromatography on a Ni-NTA-agarose column (Qiagen). It was further purified to homogeneity by elution from Q-Sepharose (Amersham Biosciences) using 100 -750 mM NaCl gradient. The most purified fraction was eluted at 400 mM NaCl and stored after dialysis in PBS (pH 7.4) containing 450 mM NaCl, 50% glycerol, and 1 mM DTT. The hnRNP-U domains containing 1-100, 1-188, 1-235, 236 -645, and 646 -808 residues were cloned for GST fusion between the BamHI/EcoRI restriction sites in the pGEX-2T E. coli expression plasmid. The GST-tagged polypeptides were also expressed in E. coli, purified by glutathione-Sepharose affinity chromatography (GE Healthcare), and then eluted using 20 mM reduced glutathione. The GST fusion peptides were dialyzed and further purified by cation exchange chromatography.
Cell Culture and Co-immunoprecipitation (co-IP) Assays-The human embryonic kidney cell line HEK293 was grown in DMEM containing 10% FBS, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM glutamine, at 37°C, 5% CO 2 , using reagents from Invitrogen. For co-IP experiments, we generated stable transfectants of HEK293 cells expressing FLAG-NEIL1 or empty-FLAG as follows. Mammalian expression plasmids for C-terminally FLAG-tagged NEIL1 were previously described (33,35). Log-phase HEK293 cells (in 6-cm plates) were transfected with 1 g of NEIL1-FLAG or the empty vector plasmid (19). After 24 h, the cells were trypsinized and plated in fresh medium containing Zeocin (100 g/ml) to select clones carrying stably integrated plasmid DNA. After 10 days, Zeocinresistant clones were individually expanded and screened for expression of FLAG-NEIL1; only the clones with ectopic NEIL1 expression comparable with that of endogenous NEIL1 were used in this study. For IP analysis, the cells were lysed in lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, and protease inhibitor mixture (Roche Applied Science)) at 48 h after transfection. After digesting an aliquot with 500 units/ml each of DNase I and RNase A (Ambion) at 37°C for 1 h followed by centrifugation as described previously (22,23), the lysates were immunoprecipitated for 3 h at 4°C with FLAG M2 antibody (Ab)-bound agarose beads (Sigma). The beads were collected by centrifugation and washed three times with cold Tris-buffered saline (TBS), and the FLAG immunocomplex was eluted from the beads by adding SDS loading buffer and separated by SDS-PAGE followed by immunoblotting with the appropriate antibodies.
To analyze the cell cycle dependence of the NEIL1 interaction with hnRNP-U, HEK293 cells were synchronized by double thymidine block (36). Briefly, at ϳ30% confluency, the cells were treated with 2 mM thymidine in DMEM for 18 h and then transferred to fresh DMEM for 6 h. This was followed by a second thymidine treatment for 17 h. The cells were finally washed twice with PBS, allowed to grow in DMEM, and harvested at 0, 1.5, 3, 4.5, and 6 h. The cells were fixed in 70% prechilled (Ϫ20°C) ethanol, stained with the staining solution (containing 100 g/ml RNase A, 40 g/ml propidium iodide in PBS), and analyzed by flow cytometry.
In Vitro Pulldown Assay-GST pulldown assays were performed as described previously (20). Briefly, the GST fusion hnRNP-U polypeptides (20 pmol) were bound to glutathione-Sepharose beads (20 l) and then mixed with full-length NEIL1 or its deletion polypeptides (10 pmol) in 0.5 ml of TBS buffer and incubated with constant rocking for 4 h at 4°C. After extensive washing, the bound proteins were separated by SDS-PAGE for immunoblotting analysis.
Far Western Analysis-The proteins (40 pmol) transferred to the nitrocellulose membrane after SDS-PAGE were denatured in situ with 6 M guanidine-HCl and renatured by successive incubation with serially diluted guanidine-HCl in PBS ϩ 1 mM DTT (23). The membrane was then incubated with hnRNP-U or NEIL1 (10 pmol/ml) in PBS supplemented with 0.5% nonfat dried milk, 0.05% Tween 20, 10 mM trimethylamine N-oxide, and 1 mM DTT for 3 h followed by washing. Stable binding of interacting protein was confirmed by immunoblotting.
For kinetics studies, the fork substrate (2.5-120 nM) was incubated with 20 nM NEIL1 or NEIL plus hnRNP-U (10 nM) for 4 min at 37°C after initiating the reaction by adding NEIL1. K m , V max , and k cat values were calculated from the linear range of the reaction by regression analysis using SigmaPlot (22).
Complete Repair Assay-NEIL1-initiated repair was assayed in reconstituted in vitro system using a 5-OHU-containing duplex oligonucleotide substrate (22) or 5-OHU-containing plasmid substrate (31). Briefly, after individual protein levels in the reaction mixture were optimized for maximum product formation, the repair reaction (20 l) containing 50 fmol each of NEIL1, polynucleotide kinase/phosphatase, and DNA polymerase ␤, 100 fmol of DNA III␣, 1 mmol of ATP, 10 mol of [␣-32 P]dCTP, and unlabeled dNTPs (25 mmol) was incubated for 30 min at 37°C. Appropriate controls lacking various components were included. For repair using the FLAG-NEIL1 IP, lysates of HEK293 cells stably expressing FLAG-NEIL1, untreated or treated with glucose oxidase (GO) (50 ng/ml) were incubated with beads attached with FLAG Ab and normalized for the FLAG level as before (31). Appropriately washed beads were then mixed with the repair reaction mixture containing the plasmid substrate, 1 mM ATP, and [␣-32 P]dATP and incubated with intermittent mixing for 30 min at 37°C. The products were then analyzed in a PhosphorImager after separation in a denaturing gel (22).
In Situ Proximity Ligation Assay (PLA)-HEK293 cells grown overnight in 16-well chamber slides were fixed with 4% paraformaldehyde and then permeabilized with 0.2% Tween 20 followed by incubation with a primary Ab for NEIL1 (rabbit (33)) or hnRNP-U (mouse monoclonal; Santa Cruz Biotechnology). The PLA assay was performed using the Duolink PLA kit (Olink Bioscience catalogue number LNK-92101-KI01, Uppsala, Sweden) per the manufacturer's instructions. The nuclei were counterstained with DAPI, and the PLA signals were visualized in a fluorescence microscope (Olympus) at 200ϫ magnification.
Clonogenic Survival Assay-Log phase HEK293 cultures were treated with NEIL1 siRNA (80 nM; targeting the 3Ј-UTR region of the NEIL1 gene; sense sequence, 5Ј-CCGUGAU-GAUGUUUGUUUAUU-3Ј; antisense sequence, 5Ј-UAAA-CAAACAUCAUCACGGUU-3Ј, Sigma) or hnRNP-U siRNA (Dharmacon, catalogue number J-013501-05), individually or together for 48 h. Scrambled siRNA was used as a control. NEIL1 or hnRNP-U down-regulation was confirmed by immunoblotting of the cell extracts at 48 h after transfection. Cells were trypsinized and transferred to 60-mm dishes (400 cells/ dish). Twelve hours after plating, the cells were treated with GO (0 -200 ng/ml) for 15 min. After allowing the cells to grow in fresh medium for 10 days, the colonies were counted after staining with crystal violet for calculating the surviving fraction (37).
PONDR Modeling-Secondary structures of NEIL1 and hnRNP-U were predicted using PONDR (Predictor of Naturally Disordered Regions) modeling software (38,39). The protein sequences were obtained from the National Center for Biotechnology Information (NCBI) database. The PONDR score of 0.5 and higher indicate a disordered structure.
Circular Dichroism (CD) Spectroscopy-The CD spectra (195-260 nm) were recorded for the hnRNP-U terminal polypeptides 1-100 and 651-806 (2 M) in 25 mM Tris-HCl buffer (pH 7.4), 50 mM NaCl in a JASCO J-720 spectropolarimeter (31), using a 1-mm cuvette, and four repetitions of the CD spectra were used for averaging and then corrected for the contribution from the buffer alone. The secondary structures of the enzyme were calculated from the spectra using the K2d software (40).

RESULTS
In-cell Association of NEIL1 with hnRNP-U and Its Enhancement after Oxidative Stress-After initial identification of hnRNP-U as one of the NEIL1 interacting partners by mass spectrometry of proteins in NEIL1 IP from HEK293 cells (data

hnRNP-U Enhances Oxidized Base Repair in Human Genome
not shown), we decided to confirm in-cell association of endogenous NEIL1 with hnRNP-U using the in situ PLA in HEK293 cells. In this analysis, two proteins are immunostained with distinct species-specific secondary Abs that are linked to complementary oligonucleotides. When two different Ab molecules bind in close proximity (Ͻ40 nm), the linked DNA can be linearly amplified via the rolling circle mechanism and visualized as distinct foci with a fluorescent probe. This assay was shown to be highly specific for physically interacting endogenous proteins in the cell (41)(42)(43). We detected a significant number of foci with NEIL1-hnRNP-U, whereas such signal was essentially absent when the primary Abs for NEIL1 or hnRNP-U were replaced with normal IgGs (Fig. 1A). Interestingly, the association was significantly enhanced (ϳ4-fold) in oxidatively stressed cells after treatment with 50 ng/ml GO for 30 min. We then tested for the presence of hnRNP-U in NEIL1 IP from HEK293 cells stably transfected with NEIL1-FLAG or empty-FLAG expression plasmid. The levels of ectopic NEIL in these lines are comparable with that of the endogenous enzyme, and the 30-min GO exposure does not change the level of expression of endogenous NEIL1 or hnRNP-U (data not shown). The presence of hnRNP-U in the NEIL1-FLAG IP, but not in the vector control IP, and the hnRNP-U-enhanced association with NEIL1 in GO-treated cells (Fig. 1B), confirm the PLA results. Because both hnRNP-U and NEIL1 bind to DNA, we considered the possibility that the interaction is indirect, mediated by their independent binding to DNA. That prior treatment of the cell lysates with DNase I ϩ RNase A (500 units/ml each) before co-IP did not affect hnRNP-U level in the NEIL1 IP strongly supports direct interaction between these polypeptides (Fig. 1C), which was also confirmed by similar analysis in HCT116 cells (data not shown). To confirm the specificity of oxidative stress dependence of enhancement of NEIL1-hnRNP-U association, we performed PLA analysis in HEK293 cells after irradiation with UV-C (254 nm), which pre-dominantly induces cyclobutane dimers and (6 -4) photoproducts in cellular genomes (44). No difference in NEIL1-hn-RNP-U association was observed with low UV dose (5 J/m 2 ) in control versus irradiated cells (supplemental Fig. S1). A moderate increase in PLA signal observed at higher dose could be due to secondary generation of ROS (45). Taken together, these data demonstrate the stable association of NEIL1 with hnRNP-U in human cells, which is enhanced by oxidative stress.
S-phase Activation of NEIL1 Contributes to Its Increased Association with hnRNP-U-Because NEIL1 may be unique among oxidized base-specific DGs in transcriptional activation during the S-phase (34), we tested the possibility of cell cycledependent modulation of the NEIL1-hnRNP-U complex. Using double thymidine block (36), we synchronized HEK293 cells (supplemental Fig. S2), confirmed enrichment of S-phase cells (ϳ65%) 3 h after release, and monitored the level of hnRNP-U and NEIL1 polypeptides. As expected, the NEIL1 level was significantly higher (ϳ5-fold) in S-phase enriched cells when compared with G 1 cells, but the hnRNP-U level was not significantly different in the two cell populations. However, the amount of the hnRNP-U in NEIL1 in IP from the S-phase cells was significantly higher (ϳ4-fold) than in the IP isolated from G 1 cells. The comparable increase in NEIL1 level and its association with hnRNP-U in S-phase cells suggest that the enhanced association of NEIL1 with hnRNP-U in the S-phase was due to the increased level of NEIL1. This also predicts enhanced NEIL1mediated repair during the S-phase.
Direct Binding of NEIL1 to hnRNP-U and the Mapping of Their Interaction Domains-To further examine whether hnRNP-U interacts with NEIL1 directly or is mediated by other proteins, we examined in vitro interaction using far Western and GST pulldown analysis. Fig. 2A shows that hnRNP-U interacted with NEIL1 and its deletion polypeptides containing residues 1-349, 289 -389, 289 -349, 312-389, and 312-349, but  OCTOBER 5, 2012 • VOLUME 287 • NUMBER 41 not with truncated polypeptides 1-311 or 1-288 or with BSA used as a control. hnRNP-U binding to the NEIL1 C-terminal peptide 312-349 (minimum interaction interface) was further confirmed by GST pulldown assays using GST fusion of the NEIL1 C-terminal peptides (Fig. 2, B and C).

hnRNP-U Enhances Oxidized Base Repair in Human Genome
The binding of hnRNP-U to the NEIL1 C-terminal peptide was further analyzed in-cell by PLA and IP studies. HEK293 cells stably expressing FLAG-NEIL1 or FLAG-NEIL1(1-311) were subjected to PLA using FLAG Ab (mouse; Sigma) and hnRNP-U Ab (rabbit; Santa Cruz Biotechnology). Interaction indicated by PLA signals was observed with FLAG-NEIL1/hn-RNP-U, but not with FLAG-NEIL1(1-311)/hnRNP-U or IgG/ hnRNP-U (Fig. 3A). Consistent with this, co-IP studies showed the presence of hnRNP-U in FLAG-NEIL1 but not in FLAG-NEIL1(1-311) IP (Fig. 3B).
We then characterized the regions in the hnRNP-U polypeptide required for NEIL1 interaction. GST pulldown analysis showed that the interaction interface of hnRNP-U is localized to two segments present within both the N-terminal (residues 1-100) and the C-terminal regions (residues 651-806; Fig. 4, A  and B). PONDR analysis predicts that both these terminal segments in hnRNP-U are intrinsically disordered. This was further confirmed by CD spectroscopy of purified polypeptides (residues 1-100 and 651-806 from the termini of hnRNP-U (supplemental Fig. S3), which showed 94 and 82% random coil conformation, respectively (40).
High Affinity Binding of hnRNP-U to NEIL1-Based on the change in intrinsic fluorescence of hnRNP-U in the presence of the NEIL1 C-terminal peptide (residues 312-349 lacking aro-  (22). B, co-elution of hnRNP-U with the NEIL1 GST-tagged deletion polypeptides bound to glutathione-Sepharose beads. C, the NEIL1 C-terminal residues (312-349) were identified as minimal interaction residues for hnRNP-U as shown schematically.

hnRNP-U Enhances Oxidized Base Repair in Human Genome
matic amino acids), we calculated the apparent dissociation constant (K d ) to be 54 nM for its binding to hnRNP-U (Fig. 5). Because our previous studies showed that this C-terminal peptide alone contributes to the NEIL1 interaction with all of its partners tested so far, and similar affinity was observed with the peptide versus full-length NEIL1 for FEN-1 (22), we are confident that the full-length NEIL1 interacts with hnRNP-U with comparable affinity. Further, a 1:1 molar stoichiometry for binding suggests simultaneous binding of the two terminal domains of hnRNP-U with NEIL1, implying the close proximity of the termini in the native conformation.
hnRNP-U Increases NEIL1 Turnover-We investigated the effect of hnRNP-U on the NEIL1 excision of 5-OHU and strand cleavage for various oligonucleotide substrates. hnRNP-U stimulated the NEIL1 activity up to ϳ4-fold in a concentrationdependent manner when the lesion was located in the singlestranded region or in close proximity to the primer-template junction of the partially duplex oligonucleotide (Fig. 6). The NEIL1 strand incision activity on a 5-OHU-containing duplex oligonucleotide substrate was stimulated ϳ2-fold by hnRNP-U at 1:1 molar ratio (Fig. 6). Bubble and forked oligonucleotides showed similar levels of stimulation (ϳ2.5-and 2.0-fold, respectively, at a 1:1 molar ratio), whereas ssDNA showed the highest stimulation (ϳ3.5-fold). Furthermore, the lack of stimulation of C-terminal truncation mutant NEIL1(1-311), which does not bind to hnRNP-U, confirmed the requirement of their interaction for stimulation (supplemental Fig. S4). We further observed that the N-or C-terminal peptide of hnRNP-U alone also stimulated NEIL1, albeit to a lesser extent when compared with WT hnRNP-U (supplemental Fig. S5).
We calculated the NEIL1 kinetic parameters using a forked substrate in the presence and absence of hnRNP-U ( Table 1). The ϳ2.5-fold decrease in K m indicated that hnRNP-U enhances the substrate affinity of NEIL1. However, ϳ8-fold increase in k cat suggests that the hnRNP-U stimulatory effect is primarily due to its role in product release.

. Mapping and correlation of NEIL1 interaction domains in hnRNP-U with predicted disordered regions.
A, co-elution of NEIL1 with GST-tagged hnRNP-U polypeptide fragments bound to glutathione-Sepharose beads and analysis of eluted proteins by Western analysis with NEIL1 Ab or GST Ab. B, schematic representation of the hnRNP-U N-(residues 1-100) and C-(residues 651-806) terminal segments for NEIL1 binding. C, PONDR modeling-based prediction of disordered structure of the interaction segments in NEIL1 (C terminus) and hnRNP-U (N and C termini), which was confirmed by CD spectroscopy (see supplemental Fig. S3).

hnRNP-U Enhances Oxidized Base Repair in Human Genome
We then analyzed repair of 5-OHU-containing plasmid with eluates from FLAG-NEIL1 IP isolated from HEK293 cells with or without prior GO treatment (50 ng/ml for 30 min). We normalized the FLAG level in the assay mixture for comparing repair activity (Fig. 7C). Complete repair by NEIL1 IP was reduced by 20% in hnRNP-U-depleted cells (siRNA-mediated, ϳ75% down-regulation; Fig. 7B, lane 5). Enhancement of repair level by about 3-fold was observed in GO-treated cells when compared with the control cells (Fig. 7B, lower panel, lane 6). However, this enhancement was squelched in hnRNP-U-depleted cells (Fig. 7C). Taken together, these results strongly suggest the significant contribution of hnRNP-U to the enhancement of NEIL1-initiated repair due to oxidative stress, which should increase the level of oxidized bases.
NEIL1 and hnRNP-U Protect HEK293 Cells from ROS Toxicity-The in vitro repair studies suggest the stimulatory role of hnRNP-U in repair of oxidized bases, particularly after oxidative stress, which may protect cells from ROS-induced toxicity. We tested this possibility by clonogenic survival assay of cells after GO treatment. NEIL1 deficiency significantly increased the sensitivity of cells to GO, implicating NEIL1 in repair of cytotoxic base lesions (Fig. 8). Moderate but definite increase in sensitivity to GO was also observed in hnRNP-Udeficient cells, which is consistent with the hnRNP-U stimulation of oxidized base damage repair (Fig. 7). However, ROS sensitivity was not significantly enhanced by combining hnRNP-U and NEIL1 deficiency, particularly at moderate GO level (Fig. 8), suggesting that the protection of cells after oxidative stress by hnRNP-U is primarily via activation of NEIL1mediated repair.  Table S1) for use as NEIL1 substrate.

DISCUSSION
Contrary to the common perception that the repair of oxidatively damaged bases in DNA via the BER pathway involves only a few conserved repair proteins, a more complex picture of BER is now emerging in that BER comprises several alternative subpathways that utilize distinct sets of repair enzymes and other proteins (16,19,21,35,46). We have been exploring the role of noncanonical proteins in oxidized base repair and have characterized the proteins present in immunoprecipitates of early BER enzymes including DGs. In addition to the BER proteins, we identified several additional proteins, including hnRNP-U, in the IPs of NEILs and APE1 (16,19,37,47). Many of the non-repair proteins in such IPs are RNA-binding proteins that are usually abundant. 3 To eliminate the possibility of their adventitious presence in the IPs, we examined their specific, binary interaction with NEIL1 and other DGs and mapped their interaction domains. Thus the studies described here provide evidence about the physiological relevance of the role of hnRNP-U in NEIL1-mediated repair. PLA results confirmed their association in the nucleus. In vitro studies show the hnRNP-U enhancement of NEIL1 activity via direct interaction. Definitive evidence for the role of hnRNP-U in oxidative damage repair in-cell was provided by the reduced survival of hnRNP-U-deficient cells caused by oxidative stress. Increased BER activity of the eluate from NEIL1 IP from ROS-treated cells was abrogated in hnRNP-U-depleted cells (Fig. 7). Because of the contribution of NEIL2 to oxidized base repair where hnRNP-U is also involved (19), it is likely that hnRNP-U provides cellular protection from ROS toxicity by activating both NEILs. The modest increase in ROS sensitivity due to hnRNP-U deficiency alone (Fig. 8) is consistent with its involvement in repair, primarily in oxidatively stressed cells. This is supported by the increase in its in-cell association with NEIL1 after GO treatment. Whether the enhanced affinity results from their covalent modifications remains to be tested. However, this may be logical because the DG is the ratelimiting enzyme in overall repair. Another common feature of the noncanonical proteins is their activation of DGs at the step of product release. A hallmark of DGs and other early BER enzymes is their product affinity, contrary to the notion of conventional enzyme kinetics. In fact, the affinity of OGG1 for the product AP site is higher than for the substrate 8-oxoG in DNA (48). It is possible that the noncanonical proteins such as hnRNP-U cause allosteric changes in the DGs to reduce their affinity for the cleaved DNA product (21,22).
As mentioned earlier, hnRNP-U (and many other hnRNP family members) have additional functions in transcriptional regulation (24,49), which may involve direct binding to sequence-specific or nonspecific DNA. Because of the presence of multiple domains and motifs in this large polypeptide, hnRNP-U interacts with diverse proteins (24). A recent study described the role of hnRNP-U-like proteins hnRNP-UL1 and -2 in DNA damage signaling and end resection at double-strand breaks, which may involve a distinct motif in these large polypeptides (50). Because of its nuclear scaffold attachment, hnRNP-U (and its homologues) may target the DNA repair site to the nuclear matrix where chromatin unfolding and refolding may occur. We have mapped the NEIL1-binding sequences to both the N termini and the C termini of hnRNP-U. The N-terminal region (ϳ160 residues) of hnRNP-U is rich in acidic (Asp and Glu) residues, and the C-terminal segment (ϳ112 amino acid residues) contains a cluster of RGG repeats. The function of the N terminus is not known, whereas the C-terminal RGG repeat motif is involved in RNA binding (24). Although the tertiary structures of hnRNP-U (and of other hnRNPs) have not been solved, the interaction mapping analysis strongly suggests close proximity of the hnRNP-U terminal segments in the native conformation (Figs. 4 and 5).
Our previous studies have shown that the disordered C-terminal extension of NEIL1 acts as the common interaction interface for diverse partners (12, 20 -23, 39). The absence of this extension in E. coli Nei, the prokaryotic orthologue of NEIL1, underscores the physiological significance of protein-protein interactions in mammalian BER (10,39), which may induce structure in the disordered region with multiple conformers. Interestingly, the NEIL1-hnRNP-U complex involves disorderdisorder interaction that would allow a large number of induced structures at the binding interface, which is common but poorly understood in mammalian cells (39).
In summary, these results together with past studies strongly suggest that BER for modified bases is tightly regulated in response to changes in cellular state and is facilitated by interaction with non-repair proteins such as hnRNP-U.