Impacts of Arsenic on Rad18 and Translesion Synthesis

Abstract

Arsenite interferes with DNA repair protein function resulting in the retention of UV-induced DNA damage. Accumulated DNA damage promotes replication stress which is bypassed by DNA damage tolerance pathways such as translesion synthesis (TLS). Rad18 is an essential factor in initiating TLS through PCNA monoubiquitination and contains two functionally and structurally distinct zinc fingers that are potential targets for arsenite binding. Arsenite treatment displaced zinc from endogenous Rad18 protein and mass spectrometry analysis revealed arsenite binding to both the Rad18 RING finger and UBZ domains. Consequently, arsenite inhibited Rad18 RING finger dependent PCNA monoubiquitination and polymerase eta recruitment to DNA damage in UV exposed keratinocytes, both of which enhance the bypass of cyclobutane pyrimidine dimers during replication. Further analysis demonstrated multiple effects of arsenite, including the reduction in nuclear localization and UV-induced chromatin recruitment of Rad18 and its binding partner Rad6, which may also negatively impact TLS initiation. Arsenite and Rad18 knockdown in UV exposed keratinocytes significantly increased markers of replication stress and DNA strand breaks to a similar degree, suggesting arsenite mediates its effects through Rad18. Comet assay analysis confirmed an increase in both UV-induced single-stranded DNA and DNA double-strand breaks in arsenite treated keratinocytes compared to UV alone. Altogether, this study supports a mechanism by which arsenite inhibits TLS through the altered activity and regulation of Rad18. Arsenite elevated the levels of UV-induced replication stress and consequently, single-stranded DNA gaps and DNA double-strand breaks. These potentially mutagenic outcomes support a role for TLS in the cocarcinogenicity of arsenite.

1. Introduction

Arsenic is an established human skin carcinogen and acts as a cocarcinogen by enhancing the DNA damage, mutagenicity, and tumor formation of ultraviolet radiation (UV). A proposed mechanism underlying arsenic cocarcinogenicity is the disruption of DNA repair leading to the retention of UV-induced DNA damage (Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021). Arsenic inhibits DNA repair through multiple mechanisms including altered DNA repair protein expression, post translational modifications, subcellular localization, recruitment to DNA damage, and enzymatic activity. Many of these mechanisms are dependent upon the function of zinc finger domains for protein:protein and protein:DNA interactions (Holcomb et al., 2017; Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021). DNA repair proteins poly(ADP-ribose) polymerase 1 (PARP-1) and xeroderma pigmentosum complementation group A (XPA) are involved in base and nucleotide excision repair respectively, and contain zinc finger DNA binding domains which have been well characterized for arsenite binding and zinc displacement (Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021).

The inhibition of DNA repair by arsenic increases UV-induced DNA damage retention and the likelihood of unrepaired DNA lesions present during replication (Holcomb et al., 2017; Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021). These lesions are circumvented through the activation of DNA damage tolerance (DDT) pathways such as translesion synthesis (TLS) (Ma et al., 2020; Vaziri et al., 2016). When replication forks encounter DNA damage, the replicative polymerase and helicase become uncoupled. Consequently, a stretch of single-stranded DNA (ssDNA) is exposed and is subsequently bound with replication protein A (RPA). Regions of ssDNA and RPA assist in the recruitment of the E3 ubiquitin ligase RAD18 in complex with the E2 ubiquitin conjugating enzyme human Rad6 (homologs Rad6A/B) (Figure 1A). The Rad18 RING (Really Interesting New Gene) finger facilitates the monoubiquitination of proliferating cell nuclear antigen (PCNA), a processivity factor necessary for coordinating DNA replication (Figure 1A). Monoubiquitinated PCNA enhances the recruitment and retention of alternative polymerases with larger, more accommodating active sites to bypass the damage by inserting nucleotides opposite the DNA lesion. Bypass of UV-induced cyclobutane pyrimidine dimers (CPDs) by TLS polymerase eta (Polη) is viewed as an error free process; however, other TLS polymerases are more error prone (Lou et al., 2021; Ma et al., 2020; Vaziri et al., 2016). An alternative DDT pathway, template switching, involves the extension of monoubiquitinated PCNA to polyubiquitinated PCNA via a Rad5 complex. Template switching is relatively error free and is mediated by strand invasion of the sister chromatid. Both TLS and template switching depend on the E3 ubiquitin ligase activity of the Rad18 RING finger for the monoubiquitination of PCNA (Gaillard et al., 2015; Ma et al., 2020; Vaziri et al., 2016).

Figure 1: Arsenite alters TLS initiation.

(A) Rad18 Schematic. Dashed arrows represent Rad18 zinc finger mediated post translational modifications. Solid arrows represent Rad18 domain binding activity. BD = Binding domain. NLS = Nuclear Localization Signal. (B) HEKn cells were treated without or with 1 μM (1A) or 5 μM arsenite (5A) for 1, 24, and 48 h. Whole cell lysates were collected and analyzed via western blot as described in Methods. Monoubiquitinated Rad18 (Rad18-mUb) was normalized to total protein stain and no treatment control (0 h). N≥3. ****p<0.0001 compared to no treatment control. (C) HEKn cells were treated without (0A) or with 1 μM (1A) or 5 μM arsenite (5A) for 24 h prior to exposure without (No Damage: ND) or with 2.8 kJ/m² UVB. Whole cell lysates were collected 4 h post UV exposure and analyzed via western blot. Monoubiquitinated PCNA (Lys164; PCNA-mUb) was normalized to total protein stain and no treatment control. N≥6. ****p<0.0001, **p<0.01, *p<0.05 compared to corresponding 0A control. ^δδδδp<0.0001 comparing ND to UV for arsenite treatment matched groups. (D) HEKn cells were transfected with 40 nM SCR (UV+SCR) or 20 nM each of R18A and R18B (UV+R18AB; 40 nM total) as described in Methods, then exposed to 2.8 kJ/m² UVB 2 days later. Whole cell lysates were collected 4 h post UV exposure and analyzed via western blot. Monoubiquitinated PCNA was normalized to total protein stain and UV+SCR control. N=3. *p<0.05 compared to UV+SCR. Representative western blots for Figure 1B–D are shown in Supplemental Figure 7. All values represent mean ± SEM.

One mechanism by which arsenic may alter TLS is through the binding and disruption of the Rad18 zinc fingers. Several studies have demonstrated arsenite targeting of RING finger containing proteins resulting in protein degradation or inhibition (Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021). The loss of Rad18 or Rad18 RING function substantially reduces UV-induced PCNA monoubiquitination and thus the initiation of TLS (Huang et al., 2009; Inagaki et al., 2011a; Miyase et al., 2005; Shiomi et al., 2007; Watanabe et al., 2004). In contrast, no UBZ domains have been evaluated as potential arsenite targets. The Rad18 UBZ domain prevents the overactivation of TLS by facilitating the monoubiquitination of Rad18 and the subsequent formation of the Rad18 homodimer (Figure 1A) (Inagaki et al., 2011a; Miyase et al., 2005; Zeman et al., 2014). High levels of Rad18 are known to overactivate TLS leading to mutagenesis and the development of treatment resistant cancers, demonstrating the importance of proper TLS regulation to prevent mutagenesis (Ma et al., 2020; Vaziri et al., 2016).

The impact of arsenic on TLS is unknown and may represent an avenue for the cocarcinogenicity of arsenic as both the overactivation and underactivation of TLS can promote genomic instability (Gaillard et al., 2015; Ma et al., 2020; Vaziri et al., 2016). In the context of UV exposure, TLS is well characterized for inducing mutations when error prone TLS polymerases replace the function of Polη. Indeed, the debilitating disorder xeroderma pigmentosum is associated with mutations in the gene coding for Polη. Though error prone in certain conditions, TLS is necessary for bypassing unresolved lesions during replication to prevent mutagenic replication gaps and fork collapse events (Gaillard et al., 2015; Ma et al., 2020; Vaziri et al., 2016). The loss of TLS is associated with an increase in DNA damage induced s-phase arrest, DNA strand breaks, chromosomal aberrations, and >4 bp genomic deletions (Bi et al., 2005; Lou et al., 2021; Ma et al., 2020; Saberi et al., 2007; Sasatani et al., 2015; Shiomi et al., 2007; Tateishi et al., 2003; Vaziri et al., 2016). If arsenic disrupts TLS, these adverse effects on genomic integrity may be exacerbated due to the arsenic-induced retention of DNA damage and suppression of alternative pathways for remediating stalled replication forks (Gaillard et al., 2015; Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021). Although several in vitro, in vivo, and ex vivo studies have reported an increase in DNA strand breaks, chromosome aberrations, and genomic deletions with arsenic exposure (Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021), the mechanisms have not been well-defined and arsenic disruption of TLS may be involved.

Results from this study support the conclusion that arsenite targets the functions of Rad18 and thereby interferes with TLS. Functional analysis of the Rad18 RING finger and UBZ domains showed a reduction in PCNA and Rad18 monoubiquitination in normal human neonatal epidermal keratinocyte (HEKn) cells exposed to arsenite, indicating zinc finger inhibition. Exposure of keratinocytes to arsenite resulted in zinc loss from endogenous Rad18 and mass spectrometry analysis revealed arsenite binding to each of the Rad18 zinc fingers. The nuclear localization of several TLS factors including Rad18 and Rad6 were significantly decreased in keratinocytes exposed to arsenite. In addition, arsenite reduced chromatin recruitment of Rad18, Rad6, RPA, and Polη in UV exposed keratinocytes. These findings were supported by an arsenite-induced reduction of Rad18 and Polη colocalization with PCNA. Both arsenite treatment and Rad18 knockdown significantly decreased post-UV DNA replication and elevated the levels of phospho-histone H2A.X (Ser139) (PH2AX), a marker of replication stress and DNA strand breaks (Cleaver, 2011; Mognato et al., 2021; Sirbu et al., 2011). Additionally, arsenite increased the presence of both ssDNA and DNA double-strand breaks (DSBs) as measured by neutral single-cell gel electrophoresis (comet assay). These findings demonstrate that Rad18 is a target for arsenite and the observed downstream consequences of arsenic exposure support disruption of TLS.

2. Materials and Methods

2.1. Reagents

Chemicals

Sodium arsenite (>99%) was purchased from Fluka Chemie (Buchs, Switzerland); zinc chloride (>98%) from Sigma (St. Louis, MO); and 16% Formaldehyde Solution from Thermo Scientific (Waltham, MA).

Small interfering RNAs (siRNAs) & transfection reagent

The Rad18 Human siRNA Oligo Duplex kit was obtained from Origene (Rockville, MD) and contained 3 unique Rad18 27mer siRNA duplexes A-C #SR311213 (Locus ID 56852; RefSeq NM_020165), Trilencer-27 Universal Scrambled Negative Control siRNA Duplex #SR30004, and RNAse free siRNA Duplex Resuspension Buffer. The Trilencer-27 Fluorescent-labeled transfection control siRNA duplex #SR30002 and siTran 2.0 siRNA Transfection Reagent kit #TT320001 containing siTran 2.0 Transfection Reagent and 5x Transfection Buffer were purchased from Origene. siRNA Duplex Resuspension Buffer was added to siRNA duplexes to a final concentration of 10 μM according to manufacturer’s instructions and aliquots frozen at −20 °C.

Antibodies

Primary antibodies obtained from Cell Signaling (Danvers, MA) include DNA Polymerase η (E1I7T) Rabbit mAb #13848, HR6A/HR6B Rabbit Antibody #4944, PCNA (D3H8P) XP^® Rabbit mAb #13110, PCNA (PC10) Mouse mAb #2586, Phospho-Histone H2A.X (Ser139) Antibody #2577, Phospho-Histone H2A.X (Ser139) (D7T2V) Mouse mAb #80312, Phospho-Rad18 (Ser403) Rabbit Antibody #8393, Rad18 (D2B8) XP^® Rabbit mAb #9040, and RPA70/RPA1 Rabbit Antibody #2267. HR6B/UBE2B Mouse Antibody (PCRP-UBE2B-1C7) #NBP3-07169 was obtained from Novus Biologicals (Centennial, CO). POLH Rabbit Polyclonal Antibody #PA5-76055 was obtained from Thermo Fisher Scientific.

Secondary antibodies used in near-infrared western blotting were obtained from LI-COR Biosciences (Lincoln, NE) and include IR Dye 680RD Goat anti-Mouse #925-68070, IR Dye 680RD Goat anti-Rabbit #925-68071, IR Dye 800CW Goat anti-Mouse #925-32210, and IR Dye 800CW Goat anti-Rabbit #925-32211. Invitrogen secondary antibodies used in immunocytochemistry were obtained through Thermo Fisher Scientific and include Goat anti-Mouse IgG (H+L) Alexa Fluor Plus 488 #A32723, Goat anti-Mouse IgG (H+L) Alexa Fluor Plus 647 #A32728, Goat anti-Rabbit IgG (H+L) Alexa Fluor Plus 488 #A32731, and Goat anti-Rabbit IgG (H+L) Alexa Fluor Plus 647 #A32733.

Peptides

The following peptides were ordered from Genemed Synthesis, Inc. (San Antonio, TX): TKVDCPVCGVNIPESHINKHLDSCLS (UBZ), TKVDCPVCGVNIPESHINKHLDSHLS (UBZ_C2H2), LRCGICFEY-FNIAMIIPQCSHNYCSLCIRKFLSYKTQCPTCCV (RING), LRHGIHFEYFNIAMIIPQHSHNYCSLCIRKFLSYKT-QCPTCCV (RING_H2C2). Peptides were validated by the manufacturer using HPLC and purity was determined to be >95%.

2.2. Ultraviolet Radiation (UV)

UV exposures were conducted with two FS40 UVB broadband bulbs emitting wavelengths of light between 280 nm and 400 nm with a peak near 300 nm. The proportion and intensity of UVA/UVB is measured annually using an ILT2400 radiometer equipped with UVA (SED033) and UVB (SED240) detectors (International Light Technologies; Peabody, MA). The spectral emission profile for exposures described herein is 91% UVB, 9% UVA and <1% UVC. In experiments involving UV, HEKn cells in medium were exposed to 2.8 kJ/m² UVB. The dose of UVB was sufficient to induce CPDs (Supplemental Figure 1A), PH2AX (Supplemental Figure 1B), and replication stress (Supplemental Figure 1C) 4 h post UV without significantly altering viability 4 h or 24 h post UV (Supplemental Figure 1D).

2.3. Cell Culture and Treatment

Cell culture

Pooled normal human neonatal epidermal keratinocytes (HEKn; lot #05461) and single donor HEKn cells (lot #00263; used only in zinc release assay) were purchased from Lifeline Cell Technologies (Oceanside, CA) and cultured in Lifeline DermaLife K culture medium and supplements without the addition of antibiotics. HEKn cells were subcultured for no more than 10 passages using Lifeline trypsin 0.05% EDTA 0.02%, Lifeline trypsin neutralization solutions, and Sigma Dulbecco’s Phosphate Buffered Saline (DPBS). HEKn stocks were stored in Lifeline FrostaLife freezing medium in liquid nitrogen. Cells were maintained at 37°C in a 95% air/5% CO₂ humidified incubator. All PBS buffers used in cell culture and in the following methods were calcium and magnesium free.

Cell treatment

One mM stock solutions of arsenite were prepared in milliQ water. Stock solutions were sterilized using a 0.22-μm syringe filter and aliquots stored at −20°C. Working solutions were prepared by diluting the stock with complete cell culture medium. For experiments involving cell exposures, cells were placed in complete medium containing arsenite at the concentrations and times indicated in the figures and figure legends. Treatment of HEKn cells for 24 h with 1 μM or 5 μM arsenite did not significantly alter viability (Supplemental Figure 1E).

Rad18 knockdown

HEKn cells were transfected with Trilencer-27 Fluorescent-labeled transfection control (TYE563 labeled siRNA), universal scrambled control (SCR), Rad18 siRNA-A (R18A), Rad18 siRNA-B (R18B), or Rad18 siRNA-C (R18C) according to the manufacturer’s instructions. Briefly, cells were plated in Thermo Nunc Lab-Tek II 4-well slides at 2.5×10⁴ cells per well, Greiner Bio-One (Frickenhausen, Germany) CELLSTAR 100 mm cell culture plates at 4×10⁵ cells per plate, or Greiner Bio-One black 96-well plates at 5×10³ cells per well. After 1 day of growth, medium was replaced with fresh medium 30–60 min prior to transfection. siRNA was diluted in 1x Transfection Buffer, then siTran 2.0 Reagent added, mixed, and incubated at room temperature for 15 min. The transfection mixture was added dropwise to cells, and slides or plates gently rocked. Final concentrations were 20 or 40 nM siRNA as reported in figure legends, 9% Transfection Buffer, and 0.055% siTran 2.0 Reagent. Medium containing transfection complex was removed 5–6 h later and replaced with fresh medium. Cells were treated as described in figure legends and fixed for immunocytochemistry, collected for western blot, or read for viability 2 days post transfection.

Transfection with 20 nM of TYE563 labeled siRNA demonstrates efficient uptake of siRNA with the transfection conditions described above (Supplemental Figure 2A). HEKn cells transfected with 20 nM of R18A-C resulted in a 31% knockdown of Rad18 with R18A and 35% knockdown with R18B compared to 20 nM SCR control as measured by immunocytochemistry. No significant changes were observed with R18C and therefore this siRNA was excluded from the study (Supplemental Figure 2B). Western blot analysis demonstrated a 30% and 45% knockdown with R18A and R18B respectively (Supplemental Figure 2C), comparable to the findings obtained by immunocytochemistry. HEKn cells transfected with 20 nM each of R18A and R18B (R18AB; 40 nM total) led to a 62% and 71% knockdown of Rad18 compared to 40 nM SCR control as measured by immunocytochemistry (Supplemental Figure 2D) or western blot (Supplemental Figure 2E), respectively. Knockdown of Rad18 in HEKn cells did not significantly alter viability (Supplemental Figure 2F).

2.4. Western Blot

HEKn cells were plated in 100 mm plates at 1.4×10⁵ cells per plate or 4×10⁵ cells per plate (siRNA experiments only) and treated as described in figure legends. Plates were washed 2 times with ice-cold 1xPBS, then scraped and collected in non-denaturing lysis buffer (20 mM TRIS pH 7.5, 150mM NaCl, 1% Triton X-100) with 1:100 Thermo Halt Protease and Phosphatase Inhibitor Cocktail on ice. Lysates were sonicated for 15 pulses (3 output control, 30% duty cycle) with a Branson Sonifier Cell Disruptor 200 and centrifuged at 14000xg for 15 min at 4°C. Supernatant was transferred to fresh tube and frozen at −80°C. Protein concentration was determined using Thermo Scientific Pierce BCA Protein Assay Kit. Equal amounts of lysate (20–60μg protein) in laemmli buffer (5x: 10% SDS, 500mM DTT, 50% glycerol, 250mM Tris HCl, 0.5% Bromophenol Blue, pH 6.8) were loaded into a 10–15% SDS-PAGE gel along with Bio-Rad (Hercules, CA) Precision Plus Protein All Blue Prestained Protein Standards. Electrophoresis was performed in a Bio-Rad Mini-PROTEAN Tetra Cell or a C.B.S Scientific (San Diego, CA) Double-Wide Mini-Blotter at 120v for 1.5–2.5 h with a Bio-Rad PowerPac Basic Power Supply. Protein was transferred to LI-COR Odyssey Nitrocellulose Membrane at 30v overnight at 4°C. Membranes were allowed to dry for at least one h and then rehydrated in 1xTBS (20 mM Tris base, 150 mM NaCl, pH 7.6).

Blots were stained with LI-COR Revert 700 Total Protein Stain according to manufacturer’s instructions and imaged with LI-COR Odyssey Fc or 9120 Imager or Bio-Rad ChemiDoc MP Imaging System. Blots were destained and blocked for 1 h at room temperature in LI-COR Intercept Blocking Buffer TBS. Primary antibodies were diluted (according to manufacturer’s recommendations for western blot) in Intercept Blocking Buffer with 0.2% Tween 20 and incubated overnight at 4°C on a rocker. Three washes were performed with 1xTBS 0.1% Tween-20 (TBST) before the addition of LI-COR IR Dye secondary antibody (1:15000) for 1 h at room temperature. Three washes were performed with TBST followed by 1 wash with TBS prior to imaging on a LI-COR Odyssey Fc or 9120 Imager or Bio-Rad ChemiDoc MP Imaging System. No saturation was present in images collected for analysis. Target quantification was performed using LI-COR Image Studio Lite Ver 5.2. To calculate target signal, the product of background and area was subtracted from the sum of the pixel intensity values (Total). Signal = Total − (Background × Area). Results were normalized to Revert 700 Total Protein Stain. Blots were stripped with LI-COR NewBlot IR Stripping Buffer according to manufacturer’s instructions. Stripping efficiency was assessed prior to experimentation. Depending on the experiment, blots were cut prior to imaging to fit on the imager tray, but blot sections were imaged at the same time and under the same imaging conditions. In experiments involving multiple blots, normalization controls were included. All western blot analysis was performed with ≥3 independent experimental replicates. Uncropped western blot images for each figure are provided in Supplemental Figure 3. Brightness and contrast were adjusted uniformly to enhance the visualization of representative blots.

2.5. Immunocytochemistry

Slide preparation

HEKn cells were plated in 4-well chamber slides at 1.25×10⁴ cells per well or 2.5×10⁴ cells per well (siRNA experiments only) and treated as described in figure legends. Cells were fixed in 3.7–4% paraformaldehyde for 15 min, washed 3 times in 1xPBS, then permeabilized with dry methanol at −20°C for 10 min. After 3 washes, cells were blocked in 1xPBS 5% normal goat serum (Thermo Scientific) 0.15% Triton X-100 (Fisher BioReagents) buffer for 1 h at room temperature. Slides were placed in humidity chambers and incubated with primary antibody in 1xPBS 1% bovine serum albumin (BSA; Sigma) 0.15% Triton X-100 overnight at 4°C. Primary antibody concentrations were diluted according to manufacturer’s recommendations for immunofluorescence (immunocytochemistry), except 1:200–1:400 was used for Rad18 (D2B8) XP^® Rabbit mAb #9040 and 1:200 was used for RPA70/RPA1 Rabbit Antibody #2267. Slides were washed 3 times, then incubated with secondary antibodies (1:1000) in 1xPBS 1%BSA 0.15% Triton X-100 buffer for 1 h at room temperature. After 3 washes, slides were incubated with 0.5 μg/mL 4’,6-Diamidino-2-Phenylindole Dihydrochloride (DAPI; Invitrogen) in 1xPBS for 10 min. After 3 more washes, slides were mounted with Invitrogen Prolong Glass Antifade Mountant, allowed to cure overnight at room temperature, then sealed with nail polish and stored at 4°C. All washes (5 min each) and incubations (except for the primary antibody incubation) were performed at room temperature on a rotator. For the CPD immunocytochemistry (Supplemental Figure 1A), cells were denatured in 2 N HCl for 30 min, then washed 5 times with 1xPBS prior to block.

Confocal microscopy

Sub-Airy unit (0.6AU) pinhole confocal microscopy was performed as previously described (Noureddine et al., 2021; Saha et al., 2022) with a Leica TCS-SP8 (Leica Microsystems; Wetzlar, Germany) controlled by LASX software version 3.5.7 and followed by computational image restoration (deconvolution) with Huygens Essential version 20.10 (Scientific Volume Imaging; Hilversum, Netherlands) utilizing a constrained maximum likelihood estimation algorithm. All 3D confocal scanning parameters (Z-stacks; 1.5 μm physical length, 11 slices) were obtained with a 63×/1.4NA apochromat oil objective in sequential scanning mode, a sub-Airy pinhole size of 0.6 AU, and image size of 92 μm × 92 μm. Both lateral and axial voxel dimensions were acquired at ideal Nyquist sampling rates (https://svi.nl/NyquistCalculator) using spectral hybrid detectors where fluorophore emission spectra bands-widths were precisely set in combination with sequential scanning to avoid any possibility of fluorophore cross-talk.

As stated above, quality control measures were undertaken to avoid fluorophore cross-talk by systematically validating no direct laser excitation of adjacent (red shifted) channels were observed during parameter optimization. Secondary only controls were included in each experiment to minimize background and determine appropriate values for background subtraction during image restoration and subsequent analysis. Replicate control wells were included to test the well-to-well variability of each antibody and analysis, and no significant differences were detected. Nuclear staining was utilized to select groups of cells (≥5 cells per image) as to avoid bias from antibody staining.

Maximum intensity projection (MIP) analysis

All post-processed (deconvolved) confocal images were converted to MIPs with LAS X software. Raw MIPs were imported into Intelligent Imaging Innovations (Denver, CO) Slidebook 6 version 6.0.6 for analysis. Masks were utilized to define whole cells, as well as the nuclear and cytoplasmic regions. Sum of intensity per nuclei was measured for CPD, PH2AX, EdU (5-Ethynyl-2’-deoxyuridine), and phospho-p53 (Ser15) targets. The remaining targets were analyzed by calculating the whole cell, nuclear, or cytoplasmic target sum of intensity and dividing by the number of cells per image. Results were pooled from ≥3 independent experimental replicates. Channel threshold and brightness were adjusted uniformly to enhance the visualization of representative images.

Colocalization analysis

Deconvolved confocal images were analyzed for the spatial overlap between two targets using the Huygens Essentials Colocalization Analyzer. First, background was estimated from control images using the Costes method and the resulting channel thresholds were applied equally to all images in each experiment. Next, the Pearson’s correlation coefficient was calculated for each image and colocalization maps in the form of iso-colocalization surfaces were generated. Colocalization maps were processed with MIP Renderer using false color to highlight the lowest to highest pixel values using the following color scheme: purple (lowest) < blue < green < yellow < orange < red (highest). Results were pooled from 3 independent experimental replicates. Brightness was adjusted uniformly to enhance the visualization of representative colocalization map MIPs.

2.6. Zinc Release Assay

Single donor HEKn cells were cultured in 150 mm plates to approximately 50% of confluence then treated with or without 1 μM arsenite for 24 h. Total protein was collected, proteins of interest immunoprecipitated, and zinc content determined using 4,(2-pyridylazo)-resorcinol as previously described (Cooper et al., 2013; Zhou et al., 2011). Antibodies utilized for the immunoprecipitation (IP) and/or immunoblotting (IB) of target proteins and their final concentrations are listed in Supplemental Table 1. Zinc release assay was performed with ≥3 independent experimental replicates.

2.7. Peptide Analysis

Peptide stocks

The RING finger peptides were resuspended in 70% acetonitrile, 10% acetic acid, 20% milliQ water, and 250 μM dithiothreitol to a concentration of 1 mM and sonicated for 10 pulses. The UBZ domain peptides were resuspended in 50% acetonitrile, 50% milliQ water, and 250 μM dithiothreitol to a concentration of 1 mM and gently vortexed. Peptide stocks were stored in a desiccation chamber at −20°C.

Liquid Chromatography (LC) and Electrospray Ionization (ESI) Mass Spectrometry (MS)

Stock peptides were added to 10 mM ammonium acetate (pH 7.2) and 250 μM dithiothreitol buffer to a final concentration of 100 μM. Aliquots of the 100 μM peptides were mixed with 400 μM arsenite and incubated at room temperature for ≥30 min. Coincubation experiments were conducted by incubating the peptide with arsenite and zinc at different molar ratios (ratio 1:2, 1:1, 2:1, respectively).

Both LC-MS experiments and ESI–MS direct infusion (Supplemental Figure 10) were performed on Q-exactive orbitrap classic mass spectrometer (Thermo Fisher Scientific; San Jose, CA) equipped with a HESI source (Thermo Fisher Scientific; San Jose, CA). For ESI-MS, samples from the metal binding assay were diluted in 10-fold with 10 mM ammonia acetate and introduced into the MS source at 5 μL/min. The ESI source spray voltage, capillary temperature, sheath gas, auxiliary gas, S-lens RF level and mass resolution were maintained at 3.5 kV, 320°C, 5 psi, 0 psi, 50 V, and 17,500, respectively. For the LC-MS experiment, 1 μL of undiluted samples were injected using a Vanquish Flex Binary UHPLC (Thermo Fisher Scientific; San Jose, CA) equipped with a Biobasic 4 5 μm 50 × 2.1 mm column (Thermo Fisher Scientific; San Jose, CA). Peptides were eluted using mobile phase A of 0.1% formic acid in water, mobile phase B of 0.1% formic acid in acetonitrile, and a 10 min gradient of 1) 3%–97% B over 6.5 min, 2) then a hold of B for 1.5 min, 3) return to A in 0.5 min and 4) a final hold of A for 1.5 min. MS data were acquired in the mass range of 400 – 4,000 m/z. For data analysis, intact peptide masses were deconvoluted using Unidec1 Version 5.0.2 (Marty et al., 2015).

2.8. Chromatin Fractionation

HEKn cells were plated in 100 mm plates at 1.2×10⁵ cells per plate and cultured for 3 days, then treated with 1 μM or 5 μM arsenite for 24 h prior to UV exposure. Cells were lysed 4 h post UV exposure and proteins were fractionated as described previously (Lake et al., 2010). Briefly, cells were rinsed with 1xPBS, collected in 200 μl fractionation buffer (150 mM NaCl, 0.5 mM MgCl2, 20 mM HEPES pH 8.0, 10% glycerol, 0.5% Triton X-100, 1 mM DTT) on ice and centrifuged at 15,000xg for 20 min at 4°C. Resulting supernatant (roughly 300 μl) was removed and added to 75 μl of 5x laemmli buffer (soluble fraction), and 125 μl of 2x laemmli buffer was added to the pellet (chromatin fraction). All samples were sonicated continuously for 10 seconds. The chromatin-enriched fraction was 3 times more concentrated than the soluble fraction. Equal volumes of samples were loaded onto 13–14% SDS-PAGE gels and western blot analysis performed as described in 2.4. Representative Revert 700 Total Protein Stain of samples is provided in Supplemental Figure 4A. Fractionation efficiency was confirmed with histone H3 and histone H4 for the insoluble fraction and GAPDH for the soluble fraction (Supplemental Figure 4B). The recruitment of PARP-1 to chromatin in arsenite and UV treated cells was similar to previous findings (Supplemental Figure 4C) (Ding et al., 2017). Chromatin fractionation was performed with 3 independent experimental replicates.

2.9. DNA Synthesis Assay

Click-iT Plus EdU Cell Proliferation Kit for Imaging (Alexa Fluor^™ 555 dye) was obtained from Thermo Fisher Scientific and reagents prepared according to manufacturer’s instructions. HEKn cells were plated in 4-well chamber slides at 1.25×10⁴ cells per well or 2.5×10⁴ cells per well (siRNA experiments only) and treated as described in figure legends. Ten μM EdU was added 1 h prior to UV exposure and cells were fixed 4 h post UV for a total EdU treatment of 5 h. Cells not exposed to UV were also treated with EdU for 5 h. EdU detection was performed according to manufacturer’s instructions. Briefly, cells were fixed in 3.7% paraformaldehyde for 15 min, washed 2 times with 3% BSA in 1xPBS, then permeabilized with 0.5% Triton^® X-100 in 1xPBS for 20 min. Cells were washed 2 times with 3% BSA in 1xPBS, then incubated with Click-iT Plus reaction cocktail for 20 min. Cells were washed once with 3% BSA in 1xPBS, then 2 times with 1xPBS prior to a 30 min incubation with Hoechst^® 33342 in 1xPBS. Cells were washed 3 times with 1xPBS and mounted with Invitrogen Prolong Glass Antifade Mountant. All incubations were performed on a rotator at room temperature. Slides were allowed to cure overnight at room temperature, then sealed with nail polish and stored at 4°C. Imaging was performed as described in 2.5.

2.10. Comet Assay

Neutral comet assay was performed according to instructions included with Trevigen (Gaithersburg, MD) CometAssay #4250-050-K with the addition of an S1 nuclease treatment as previously described (Quinet et al., 2016). Briefly, HEKn cells were plated in 100 mm plates at 1.4×10⁵ cells per plate and cultured for 3 days. Cells were treated with 1 μM or 5 μM arsenite for 24 h prior to exposure with or without UV. Cells were collected 4 h post UV. For positive controls (as described in Supplemental Materials and Methods), HEKn cells were treated with 25 μM etoposide, 15 μM Mitomycin C, and serum starvation (Lifeline DermaLife medium without K supplements) for 4 h. Cells were collected by trypsinization, washed, and resuspended in ice cold 1xPBS, then counted and adjusted to 1×10⁵ cells per mL in ice cold 1xPBS. Cells were combined with 37°C molten LMAgarose (Trevigen) at a ratio of 1:10 (v/v), and 50 μl of the mixture was immediately pipetted onto each well of a Trevigen 20-well CometSlide. The slides were placed at 4°C for 30 min, then immersed in prechilled Trevigen Lysis Buffer at 4°C for 1 h. Slides with S1 nuclease treatment were washed 3 times with Promega (Madison, WI) 1x S1 Nuclease Reaction Buffer, then incubated for 30 min at 37°C in humidity chamber with 40 U/mL of Promega S1 Nuclease in 1X Reaction Buffer. All slides were washed with prechilled 1x Neutral Electrophoresis Buffer (0.1 M Tris Base plus 0.3 M sodium acetate) for 30 minutes at 4°C

The slides were transferred to the CometAssay ES tank with prechilled 1x Neutral Electrophoresis Buffer. Electrophoresis was performed at 21v for 1 h at 4°C. Excess Neutral Electrophoresis Buffer was drained and slides immersed in DNA precipitation solution (1 M ammonium acetate in 95% ethanol) for 30 min at room temperature. Slides were then immersed in 70% ethanol for 30 min at room temperature. Slides were dried overnight at room temperature in the dark. Diluted Invitrogen SYBR Gold (1:10000 in TE buffer) was pipetted onto each well and incubated for 5 min at 4°C. Excess SYBR Gold was removed, slides dried at 37°C and wells imaged using an Olympus IX83 fluorescence microscope equipped with cellSens Dimension (Olympus; v 1.9) imaging software and a DP80 digital camera. Comets (≥22 per treatment) were analyzed for percent tail DNA using Comet Analysis Software (Trevigen) version 1.2. Comet assay was performed with ≥3 independent experimental replicates. Controls were performed. Etoposide demonstrated DNA DSBs captured by the neutral comet assay. DSBs were increased with S1 nuclease treatment of the replication stress positive controls mitomycin C and serum starvation (Supplemental Figure 5).

2.11. Statistical Analysis

All graph values represent mean ± standard error of the mean (SEM). Statistical comparisons were performed by nonparametric, unpaired Student’s 2-sample t-test using GraphPad Prism 5.0 & 8.0 (GraphPad Software Inc.; San Diego, CA). p<0.05 was considered to be statistically significant.

3. Results

3.1. Arsenite alters Rad18 function required for TLS initiation

TLS is a mechanism to bypass DNA damage during replication and Rad18 performs multiple functions in TLS that are mediated by distinct domains (Figure 1A). The initiation of TLS through Rad18 is tightly regulated due to the negative implications of error prone DNA damage bypass. Alterations in Rad18 expression and posttranslational modifications, including phosphorylation and ubiquitination, in response to DNA damage are important means by which Rad18 modulates TLS (Ma et al., 2020; Vaziri et al., 2016; Zeman et al., 2014). Arsenite treatment of HEKn cells for 24 h revealed a modest increase in the levels of Rad18 and its binding partner Rad6 but did not change protein levels of two additional TLS proteins (RPA and PCNA) (Supplemental Figure 6A). TLS efficiency is increased by UV-induced Rad18 phosphorylation (Ma et al., 2020; Vaziri et al., 2016); however, neither basal nor UV-induced Rad18 phosphorylation at the Polη binding site (serine 403) was significantly altered by arsenite treatment (Supplemental Figure 6B). Because Rad18 harbors two zinc binding domains, further analysis was performed to determine if arsenite modifies zinc finger-dependent ubiquitination in TLS regulation (Figure 1A).

The monoubiquitination of Rad18 by Rad6 stimulates Rad18 homodimerization, which sequesters Rad18 from initiating TLS in undamaged cells. Rad18 monoubiquitination is reduced in response to some, but not all DNA damaging agents, freeing Rad18 to bind to other factors in order to promote DDT pathways (Zeman et al., 2014). Knockout or mutations in the Rad18 UBZ domain prevents monoubiquitination and impairs Rad18 homodimerization (Inagaki et al., 2011a; Miyase et al., 2005; Zeman et al., 2014). The effects of arsenite on Rad18 monoubiquitination has not been reported previously. Rad18 monoubiquitination was significantly reduced in HEKn cells exposed to 5 μM arsenite for 24 and 48 h suggesting impairment of the UBZ domain function (Figure 1B).

Rad18 itself is an E3 ubiquitin ligase and monoubiquitinates PCNA through its RING finger in response to DNA damage-induced stalled replication forks (Ma et al., 2020; Vaziri et al., 2016). The monoubiquitination of PCNA enhances the recruitment and retention of TLS polymerases. Treatment of HEKn cells with arsenite for 24 h significantly reduced both basal and UV-induced PCNA monoubiquitination (Lys164) in a concentration-dependent manner (Figure 1C). These results were confirmed with a general PCNA antibody (Supplemental Figure 8A). Knockdown of Rad18 (Supplemental Figure 2E) significantly reduced UV-induced PCNA monoubiquitination in HEKn cells (Figure 1D), confirming the importance of Rad18 in this response.

Both the RING and UBZ domains of Rad18 are zinc binding motifs (Inagaki et al., 2011a). Treatment of HEKn cells with the membrane-permeable zinc chelator TPEN (N,N,N′,N′-tetrakis(2-pyridinylmethyl)-1,2-ethanediamine) significantly reduced both PCNA monoubiquitination (Supplemental Figures 8A–B) and Rad18 monoubiquitination (Supplemental Figure 8C). These findings suggest arsenite may disrupt the zinc-binding domains of Rad18 as has been reported for other zinc finger proteins (Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021).

3.2. Rad18 is a target of arsenite

Arsenite binds and preferentially disrupts zinc fingers containing ≥3 zinc coordinating cysteine residues (Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021). Based on this feature, the UBZ and RING zinc fingers of Rad18 are candidate targets for arsenite (Figure 2B). The Rad18 RING finger is important in the initiation of TLS through the monoubiquitination of PCNA and is composed of two interdigitated zinc-binding sites, a C4 and a CHC2 (Huang et al., 2011). The Rad18 UBZ domain is involved in Rad18 homodimerization and coordinates one zinc ion with a C2HC zinc-binding site (Rizzo et al., 2014). A zinc displacement assay was used to determine if Rad18 is a target of arsenite. HEKn cells were treated with 1 μM arsenite for 24 h and zinc finger targets were isolated by immunoprecipitation. Zinc content was detected by incubation with a colorimetric zinc chelator, 4,(2-pyridylazo)-resorcinol). Arsenite exposure resulted in nearly 80% loss of zinc from endogenous Rad18, which was significantly greater than the loss from the C2H2 zinc finger control, aprataxin (ATPX) (Figure 2A) and comparable to zinc loss from previously identified arsenite targets PARP-1 and XPA (Supplemental Figure 9) (Zhou et al., 2011).

Figure 2: Arsenite binds Rad18 zinc finger domains.

(A) Single donor HEKn cells were treated with or without 1 μM arsenite for 24 h. Each target was isolated by immunoprecipitation and free zinc detected by 4,(2-pyridylazo)-resorcinol chelation assay as described in Methods. Results were normalized to no treatment controls defined as 100% for each target. Values represent mean ±SEM. N≥3. ****p<0.0001 compared to aprataxin (APTX). (B) Rad18 zinc finger (ZF) peptide sequence, corresponding peptide mass, and results are shown. (C-F) In an ammonium acetate buffered solution, 100 μM of each peptide was incubated with 400 μM arsenite (As(III)) for a minimum of 30 min at room temperature prior to LC-MS analysis as described in Methods. LC-MS results were obtained for (C) Rad18 RING, (D) RING_H2C2, (E) UBZ, and (F) UBZ_C2H2 peptides, respectively.

More detailed analysis using mass spectrometry was conducted to determine which of the Rad18 zinc fingers may be a target of arsenite. Peptides were synthesized containing the amino acid sequence for the Rad18 RING finger and UBZ domains, as well as mutants with only two cysteines per zinc-binding site (Figure 2B). Resuspended Rad18 zinc finger peptides were incubated with or without arsenite or zinc for LC-MS analysis. Arsenite incubation with the RING peptide revealed a +144 Dalton (Da) peak shift indicating peptide bound to two arsenic ions (Figure 2C). The shift was evident even when peptide was coincubated with arsenite and excess zinc (Supplemental Figure 10A). Interestingly, the H2C2 mutant RING peptide was still capable of binding to one arsenic ion (giving +72 Da shift in the observed mass-to-charge value) (Figure 2D), perhaps due to the availability of ≥3 zinc coordinating cysteine residues retained between the two binding sites. The +72 Da shift is due to binding of an arsenic ion and removal of two protons.

Incubation of the Rad18 UBZ peptide with arsenite resulted in a +72 Da peak shift corresponding to one bound arsenic ion (Figure 2E), which was absent with the C2H2 mutant UBZ peptide (Figure 2F). These findings were consistent with the matrix-assisted laser desorption/ionization - time of flight (MALDI-TOF) MS data (Supplemental Figure 11). Arsenite binding to the UBZ domain was not evident when coincubated with zinc (Supplemental Figure 10B). ESI-MS results showed peaks corresponding to free peptide and zinc bound peptide, but no peak corresponding to arsenic bound peptide (Supplemental Figure 10C). Altogether, these data suggest greater sensitivity of the Rad18 RING finger for disruption by arsenic in comparison to the UBZ domain.

3.3. Arsenite decreases the nuclear localization of Rad18 and Rad6 and colocalization of the Rad18-Rad6 complex

Alterations in TLS factor localization is another means of TLS regulation. Rad18 is predominantly found in the nucleus with a smaller fraction in the cytoplasm. Many factors influence Rad18 localization including cell cycle phase and monoubiquitination (Inagaki et al., 2009; Masuyama et al., 2005; Miyase et al., 2005). Nuclear localization of Rad18 was significantly decreased by 16% (Figure 3A) and the cytoplasmic fraction increased by over 3-fold (Supplemental Figure 12A) following treatment of HEKn cells with 5 μM arsenite. The reduction of Rad18 in the nucleus and increase in the cytoplasm was evident regardless of exposure to UV.

Figure 3: Arsenite disrupts the nuclear localization and colocalization of Rad18 and Rad6B.

(A-B) HEKn cells were treated without (0A) or with 1 μM (1A) or 5 μM arsenite (5A) for 24 h prior to exposure without (ND) or with 2.8 kJ/m² UVB. Cells were fixed 4 h post UV exposure and fractional nuclear localization of (A) Rad18 and (B) Rad6B was measured by immunocytochemistry as described in Methods. MIPs were analyzed for protein nuclear sum of intensity normalized to total sum of intensity and no treatment control with Slidebook 6. Nuclei are outlined in representative images and secondary only controls (CONTROL) are included. N≥15 images (≥5 cells per image). ****p<0.0001, **p<0.01, *p<0.05 compared to corresponding 0A control. ^δδp<0.01, ^δp<0.05 comparing ND to UV for arsenite treatment matched groups. (C) HEKn cells were treated without (NT) or with 1 μM (1A) or 5 μM arsenite (5A) for 24 h then fixed and immunocytochemistry performed. Pearson’s correlation coefficients and colocalization maps for Rad18 and Rad6B were obtained with Huygens Colocalization Analyzer as described in Methods. Results were normalized to NT control. Cells from representative colocalization map MIPs are provided and display false coloring to show pixel intensity from purple (lowest) < blue < green < yellow < orange < red (highest). N≥12 z-stacks (≥5 cells per z-stack). ****p<0.0001 compared to NT. All values represent mean ± SEM.

It is proposed that Rad6 localization is dependent on Rad18 binding and nuclear translocation via the Rad18 nuclear localization signal (Hedglin and Benkovic, 2015). The interaction of Rad18 with Rad6 is dependent on the Rad18 RING finger and Rad6 binding domain (Figure 1A) (Inagaki et al., 2011a). Knockout or mutations in the Rad18 RING finger decrease the amount of Rad6 coimmunoprecipitated with Rad18 (Inagaki et al., 2011a). The nuclear localization of the Rad6 human homolog Rad6B was significantly decreased (Figure 3B), and the cytoplasmic fraction increased (Supplemental Figure 12A) in response to arsenite with or without UV. Since the Rad18 RING finger is a target of arsenite (Figure 2), RING binding activity may be disrupted. To assess the impact of arsenite on the interactions between Rad18 and Rad6B, colocalization analysis was performed in arsenite treated HEKn cells. Results demonstrate a concentration dependent decrease in Rad18-Rad6B colocalization (Figure 3C), suggesting a disruption in the interaction between Rad18 and Rad6.

RPA and PCNA are predominantly located in the nucleus with little detected in the cytoplasm (Supplemental Figure 13). Arsenite caused minor decreases in the nuclear localization of these factors (Supplemental Figure 13), with corresponding increases in the cytoplasmic fraction (Supplemental Figure 12A). The nuclear fraction of Polη was significantly increased by UV exposure (Supplemental Figure 13), which was substantially suppressed by arsenite coexposure. Altogether, localization analysis revealed a decrease in the nuclear fraction of several TLS factors and an arsenite-induced loss in Rad18-Rad6B colocalization which suggests disruption of the Rad18 RING finger binding activity and suppression of TLS initiation.

3.4. Arsenite alters TLS factor recruitment in response to UV

Rad18 recruitment to chromatin is heavily regulated by multiple mechanisms to ensure TLS is initiated only when necessary. RPA and single-stranded DNA are critical factors controlling Rad18 recruitment to stalled replication forks (Li et al., 2020; Ma et al., 2020). Chromatin fractionation of UV exposed HEKn cells revealed a substantial loss in chromatin bound Rad18, Rad6, and RPA upon coexposure to arsenite (Figure 4A). In contrast, 5 μM arsenite significantly increased the amount of chromatin bound PCNA by 66% in UV exposed cells.

Figure 4: Arsenite alters TLS factor recruitment in response to UV.

(A) HEKn cells were treated without (UV) or with 1 μM (UV+1A) or 5 μM (UV+5A) arsenite for 24 h prior to exposure with 2.8 kJ/m² UVB. Chromatin fractionation was performed 4 h post UV exposure as described in Methods. TLS proteins in the chromatin fraction were normalized to total protein stain and UV only control. N=3. ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05 compared to UV. Representative western blots are shown in Supplemental Figure 14. (B-C) HEKn cells were treated without (UV) or with 1 μM (UV+1A) or 5 μM (UV+5A) arsenite for 24 h prior to exposure with 2.8 kJ/m² UVB. Cells were fixed 4 h post UV exposure and immunocytochemistry was performed. Pearson’s correlation coefficients and colocalization maps for (B) Rad18 and (C) Polη colocalization with PCNA were obtained with Huygens Colocalization Analyzer. Results were normalized to UV only control. Cells from representative colocalization map MIPs are provided and display false coloring to represent pixel intensity from purple (lowest) < blue < green < yellow < orange < red (highest). N≥18 z-stacks (≥5 cells per z-stack). ****p<0.0001, *p<0.05 compared to UV. All values represent mean ± SEM.

PCNA monoubiquitination by the Rad18-Rad6 complex enhances the recruitment of Polη to stalled replication forks (Ma et al., 2020; Vaziri et al., 2016). Arsenite treatment or Rad18 knockdown significantly decreased PCNA monoubiquitination (Figure 1C–D). Therefore, it is expected that arsenite exposure will negatively impact the recruitment of Polη to chromatin. Indeed, arsenite treatment decreased the amount of chromatin bound Polη (Figure 4A). The decreased recruitment of critical TLS factors Rad18 and Polη in response to UV was confirmed by colocalization analysis. Colocalization results demonstrated an arsenite-induced reduction in the interaction of Rad18 (Figure 4B) and Polη (Figure 4C) with PCNA in UV exposed cells. Both the inhibition of the Rad18 RING finger domain through arsenite binding and the reduction in nuclear localization and chromatin recruitment of Rad18 could negatively impact TLS initiation, thus impeding the recruitment of Polη to stalled replication forks and increasing replication stress.

3.5. Arsenite and Rad18 knockdown induce replication stress

If not resolved or bypassed, the formation of bulky adducts from UV exposure may induce replication fork stalling, replication gaps, and fork collapse. These outcomes of replication stress can be measured as a decrease in DNA synthesis, increase in ssDNA, and increase in DSBs respectively. Consequently, these events promote apoptosis or mutagenesis. Failure of TLS has been demonstrated to exacerbate replicative stress (Gaillard et al., 2015; Ma et al., 2020; Vaziri et al., 2016).

In order to test the net effect of arsenite on replication stress, HEKn cells were treated with arsenite then UV exposed and the effects on DNA replication (EdU incorporation) and DNA strand breaks and ssDNA gaps were measured. Arsenite treatment of HEKn cells led to a pronounced reduction in EdU incorporation post-UV exposure (Figure 5A). In the absence of UV, 5 μM, but not 1 μM, arsenite decreased EdU incorporation (Supplemental Figure 15A). Knockdown of Rad18 led to a 46% decrease in DNA synthesis post-UV (Figure 5A) as has been reported previously (Tateishi et al., 2003, 2000; Yamashita et al., 2002), and a reduction in EdU incorporation in unexposed cells (Supplemental Figure 15B). The addition of arsenite to Rad18 knockdown cells did not significantly alter EdU incorporation when compared to the SCR siRNA control cells, suggesting that the arsenite effect is largely mediated through Rad18.

Figure 5: Arsenite and Rad18 knockdown increase replication stress.

(A-B) HEKn cells were transfected with 40 nM of universal scrambled control (SCR) or 20 nM each of R18A and R18B (R18AB). One day post transfection, cells were treated without (UV) or with 1 μM (UV+1A) or 5 μM arsenite (UV+5A) for 24 h prior to 2.8 kJ/m² UVB. (A) EdU (10 μM) was added 1 h prior to UV. Cells were fixed 4 h post UV and EdU detection was performed as described in Methods. MIPs were analyzed for EdU sum of intensity per nucleus normalized to UV+SCR control. Nuclei are outlined in representative images and secondary only controls (CONTROL) are included. N≥300 nuclei. ****p<0.0001 compared to corresponding UV control. ^δδδδp<0.0001 comparing SCR to R18AB for arsenite treatment matched groups. (B) HEKn cells were treated without (UV) or with 1 μM (UV+1A) or 5 μM arsenite (UV+5A) for 24 h prior to 2.8 kJ/m² UVB. Cells were fixed 4 h post UV exposure and immunocytochemistry was performed. MIPs were analyzed for PH2AX (Ser139) sum of intensity per nucleus normalized to UV+SCR control. N≥119 nuclei. ****p<0.0001, ***p<0.001 compared to corresponding UV control. ^δδδδp<0.0001 comparing SCR to R18AB for arsenite treatment matched groups. (C) HEKn cells were treated without (UV) or with 1 μM (UV+1A) or 5 μM arsenite (UV+5A) for 24 h prior to 2.8 kJ/m² UVB. Neutral comet assay was performed 4 h post UV exposure with or without the addition of S1 nuclease as described in Methods and representative images are shown. Comets were analyzed for percent tail DNA using Trevigen Comet Analysis Software and normalized to UV only control. N≥3. ***p<0.001, **p<0.01, *p<0.05 compared to corresponding UV control. ^δδp<0.01, ^δp<0.05 comparing No S1 to +S1 for arsenite treatment matched groups. All values represent mean ± SEM.

Prolonged stalling of replication forks can lead to replication fork collapse resulting in the formation of mutagenic DNA DSBs (Gaillard et al., 2015; Ma et al., 2020; Vaziri et al., 2016) and PH2AX serves as both a marker of replication stress and DNA strand breaks (Cleaver, 2011; Mognato et al., 2021; Sirbu et al., 2011). In the absence of UV, the levels of PH2AX were not significantly increased by arsenite (Supplemental Figure 15C) or Rad18 siRNA treatment (Supplemental Figure 15D). However, UV-induced PH2AX was increased with arsenite exposure in a concentration dependent manner (Figure 5B, SCR). Knockdown of Rad18 led to a nearly 3-fold increase in UV-induced PH2AX in the absence of arsenite indicating involvement of Rad18 in the response. Arsenite modestly increased PH2AX in Rad18 knockdown cells, possibly due to inhibition of residual Rad18 (Supplemental Figure 2D–E) or through another mechanism. These findings indicate that arsenite treatment and Rad18 knockdown each promote UV-induced replication stress.

Neutral comet assays measure DNA DSBs, and the addition of the S1 nuclease allows for the detection of ssDNA gaps. To capture mutagenic ssDNA gaps and fork collapse events as a consequence of replication stress, UV exposed HEKn cells were tested using a neutral comet assay +/− S1 nuclease (Quinet et al., 2016). Both 1 μM and 5 μM arsenite increased percent tail DNA (DSBs) in UV exposed HEKn cells (Figure 5C) but not arsenite alone (Supplemental Figure 15E). The addition of S1 nuclease to arsenite exposed cells further enhanced percent tail DNA with (Figure 5C) or without UV (Supplemental Figure 15E) indicating the presence of ssDNA. Altogether, these findings are supportive of an arsenite-induced inhibition of TLS leading to replication stress, ssDNA gaps, and fork collapse events.

Replication stress activates ATM/ATR mediated phosphorylation of p53 (phospho-p53) promoting apoptosis, which is a characteristic of TLS failure (Gaillard et al., 2015; Huang et al., 2009). Indeed, the knockdown of Rad18 in HEKn cells significantly increased UV-induced phospho-p53 (Supplemental Figure 16A) and cleaved caspase-3, a marker of apoptosis (Supplemental Figure 16B). Interestingly, 5 μM arsenite suppressed the UV-induced phosphorylation of p53 in Rad18 deficient cells (Supplemental Figure 16A), as well as decreased cleaved caspase-3 in UV exposed HEKn cells (Supplemental Figure 16C). These findings support previous studies demonstrating the ability of environmentally relevant levels of arsenite to inhibit apoptosis, such as through the inhibition of PARP-1 or p53 (Chen et al., 2005; Qin et al., 2012; Sun et al., 2011; Wu et al., 2005; Zhou et al., 2017).

4. Discussion

In this study we provide evidence for Rad18 as an arsenic target and for arsenic disruption of Rad18 functions in TLS. Arsenic targeting of sulfhydryl groups is known to alter various signaling pathways leading to changes in growth, differentiation, and cell survival in addition those involved in DNA repair (Hubaux et al., 2013; Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021). A consequence of arsenic inhibition of nucleotide and base excision repair pathways is the retention of UV-induced DNA damage (Figure 6.1) (Holcomb et al., 2017; Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021). Unrepaired damage increases replication stress and the burden on DDT pathways including TLS (Figure 6.2). Rad18 regulates TLS initiation and either the under- or over-activation of TLS have deleterious consequences on genomic integrity (Ma et al., 2020; Vaziri et al., 2016).

Figure 6: Schematic summarizing the impact of arsenic on TLS.

(1) Arsenic inhibition of base and nucleotide excision repair pathways increases the retention of UV-induced DNA damage and consequently, (2) increasing the burden on TLS. (3) Mass spectrometry analysis revealed the Rad18 RING finger and UBZ domains as targets of arsenite. (4) Disruption of the Rad18 RING domain may lead to the observed arsenite-induced decrease in PCNA monoubiquitination. (5) Arsenite exposure decreased UV-induced chromatin recruitment and PCNA colocalization of Rad18 and Polη, further disrupting TLS. (6) Rad18 knockdown and arsenite treatment led to a significant decrease in DNA replication and increase in PH2AX post UV, signifying replication stress and potential fork collapse events. Neutral comet assay analysis with S1 nuclease confirmed the presence of arsenite-induced ssDNA and DSBs in UV exposed HEKn cells. Altogether, these data support the inhibition of Rad18-mediated TLS by arsenite.

We find that arsenite binds to both zinc finger domains of Rad18 (Figure 6.3). The predominant actions of arsenite on Rad18 and its functions include the decrease of RING finger dependent PCNA monoubiquitination which is the critical initiation event for TLS (Figure 6.4) and the recruitment of Polη to chromatin (Figure 6.5), which support an underactivation of TLS leading to replication stress. Indeed, arsenite reduced post-UV DNA replication and increased post-UV strand breaks as detected by PH2AX and comet assay analysis (Figure 6.6). Both arsenic exposure and Rad18 ablation are associated with rate of sister chromatid exchange (Tateishi et al., 2003; Yamashita et al., 2002; Zhou et al., 2021), chromosomal aberrations (Despras et al., 2016; Saberi et al., 2007; Smith et al., 2004; Zhou et al., 2021), micronuclei formation (Sasatani et al., 2015; Zhou et al., 2021), and genomic deletions (Lou et al., 2021; Zhou et al., 2021). Thus, arsenite not only enhances UV-induced DNA lesions but through inhibition of Rad18 impairs a major mechanism to bypass the damage and prevent replication stress, mutagenic fork collapse, and genomic instability.

Two distinct zinc finger domains, RING and UBZ (UBZ4-type), mediate the functions of Rad18 and are capable of binding to arsenite. Based on zinc competition, the RING peptide displayed higher affinity for arsenite than the UBZ peptide (Supplemental Figure 10). This is further supported by the concentration dependence for arsenite-induced suppression of PCNA monoubiquitination (Figure 1C) versus Rad18 monoubiquitination (Figure 1B). Arsenite binding to the Rad18 zinc finger domains may have broader implications since both domains are implicated in other pathways including homologous recombination (Huang et al., 2009; Inagaki et al., 2011b, 2011a; Kermi et al., 2015; Ma et al., 2020; Song et al., 2010; Vaziri et al., 2016; Williams et al., 2011). In addition, arsenite binding to the UBZ domain is a novel finding and other UBZ4-type proteins important in the maintenance of genome integrity may be targets, such as TLS polymerase kappa (Ma et al., 2020; Vilas et al., 2018).

Arsenite treatment times and concentrations utilized in this study were chosen to gain mechanistic insight into the actions of arsenite through Rad18 in HEKn cells but do not closely reflect typical human exposures. As many studies have demonstrated considerable differences in the observed effects of arsenite when variables such as treatment time and concentration are altered (Muenyi et al., 2015; Nail et al., 2022; Osmond et al., 2010; Tam et al., 2020; Zhou et al., 2021), we performed a 7-day arsenite exposure in HEKn cells to test the effects of lower arsenite concentrations on PCNA monoubiquitination as the key Rad18 dependent initiation step for TLS. We found inhibition of PCNA monoubiquitination at sub micromolar concentrations (Supplemental Figure 17), as well as a greater impact with extended 1 μM arsenite treatment compared to 24 h (Supplemental Figure 17, 25% reduced; Figure 1C, 15% reduced). This suggests that chronic human exposures, especially in areas of higher endemic arsenic levels, may affect Rad18 activity and TLS (Bustaffa et al., 2014; Focazio et al., 2000; Ingram et al., 2020; Welch et al., 2000).

Results from this study support the arsenite-induced inhibition of Rad18 through altered activity (Figures 1–2) and regulation (Figures 3–4) ultimately leading to replication stress. These findings are consistent with those of Rad18 knockout demonstrating increased replication stress in UV exposed cells (Ma et al., 2020; Tateishi et al., 2003, 2000; Vaziri et al., 2016; Yamashita et al., 2002). UV-induced replication stress was confirmed in arsenite treated cells, measured as a reduction in post-UV DNA replication (Figure 5A), and increase in ssDNA and DNA strand breaks (Figure 5B–C). Similarities between the findings from arsenite and Rad18 siRNA treated cells suggests arsenite mediates its effects through Rad18. Interestingly, both 5 μM arsenite and Rad18 knockdown in HEKn cells had an impact on the normal replication of HEKn cells. This may be due to recent findings linking PCNA monoubiquitination to DNA replication origin activation or as a consequence of replication stress (Gaillard et al., 2015; Leung et al., 2022).

The two dominant DDT pathways are the more error prone TLS and template switching which is considered error free; both are dependent on Rad18 mediated PCNA monoubiquitination (Gaillard et al., 2015; Ma et al., 2020; Vaziri et al., 2016). Therefore, arsenite is predicted to impinge on both DDT pathways involved in replication fork maintenance. With severe replication stress, cells trigger apoptosis to avoid propagation of DNA damage (Gaillard et al., 2015). The loss of Rad18 is associated with enhanced UV-induced apoptosis (Saberi et al., 2007; Tanoue et al., 2018; Tateishi et al., 2003; Tian et al., 2013; Watanabe et al., 2004; Yamashita et al., 2002) yet arsenic suppressed this response (Supplemental Figure 16), supporting previous studies demonstrating the arsenite-induced inhibition of apoptosis in keratinocytes (Chen et al., 2005; Qin et al., 2012; Sun et al., 2011; Wu et al., 2005; Zhou et al., 2017). Arsenite inhibition of apoptosis may promote carcinogenesis by allowing the survival of DNA damaged cells defective in TLS.

TLS is associated with different mutagenic pathways when the equilibrium of TLS polymerases is disrupted or Rad18 is overexpressed (Ma et al., 2020; Vaziri et al., 2016). There are limited studies on the consequences of Rad18 knockout in UV-induced skin carcinogenesis. Because Rad18 null mice have compensatory signaling by DNA damage response factor checkpoint kinase 2 (Chk2), studies were conducted in Rad18−/−, Chk2−/− and double knockout mice (Tanoue et al., 2018). The number of mice with well-differentiated squamous cell carcinoma (SCC) after chronic UV exposure was similar in wild-type (6/20), Rad18−/− (6/19), and Chk2−/− (7/19) mice with significantly more mice bearing tumors in the double knockout (14/20) and a trend toward earlier tumor development in Rad18−/− Chk2−/− mice compared to Chk2−/− mice.

A more detailed in vivo analysis reported that Rad18 determines levels of single nucleotide variations (SNVs) versus insertion/deletion (INDEL) mutations during 7,12- dimethylbenz[a]anthracene (DMBA)-induced skin carcinogenesis (Lou et al., 2021). Skin tumor genomes from Rad18+/+ mice revealed that these tumors display mutational signatures with high levels of A(T)>T(A) SNVs. Rad18 expression was also strongly associated with high SNV burdens in human tumor data from The Cancer Genome Atlas. In contrast, tumors from Rad18−/− mice had a mutation pattern characterized by increased numbers of deletions >4 bp and an increased contribution of COSMIC signature 3 also associated with BRCA-mutant tumors (Lou et al., 2021). A study of DMBA-induced hematological malignancies reported that the number of mice with B Cell Lymphoma was significantly higher in the Rad18−/− group (11/23) compared to the Rad18+/+ group (1/20) (Yang et al., 2016). These studies are consistent with other evidence that alterations in the balance of TLS leads to differences in genomic outcomes (Figure 6.2).

The actions of arsenic in UV-induced skin carcinogenesis include increased DNA damage retention, mutations, and tumors (Muenyi et al., 2015; Tam et al., 2020; Zhou et al., 2021). Our findings demonstrate an impact of arsenic on Rad18 and TLS which in many aspects resembles Rad18 knockout, including enhanced replication stress and DNA damage-induced strand breaks (Figure 5) (Shiomi et al., 2007; Tateishi et al., 2003; Yamashita et al., 2002). However, outcomes of Rad18 inhibition by arsenic are expected to differ from those obtained by Rad18 knockout alone, as has been reported for PARP1 inhibition versus silencing (Godon et al., 2008). It is possible that Rad18 may still interact with other factors and facilitate functions independent of its zinc finger domains with arsenic exposure. Though perturbed by arsenic, the recruitment of inhibited Rad18 to stalled replication forks and gaps may block the actions of compensatory E3 ubiquitin ligases, such as RNF8 (ring finger protein 8), CRL4Cdt2 (Cullin4-RING ligase (CRL4)-Ddb1-Cdt2), and HLTF (Helicase Like Transcription Factor) (Ma et al., 2020; Tanoue et al., 2018). These E3 ligases also contain RING domains which may be vulnerable to inhibition by arsenic. The reduction in the activity of these redundant mechanisms may shunt cells to more error prone pathways, particularly in the context of UV-induced CPDs where TLS is the predominant tolerance pathway and remarkably accurate (Cohen et al., 2015; Ma et al., 2020). Future studies to define the effects of arsenic on UV-induced tumor mutational spectrum should shed light on the broader impact of arsenite-induced TLS inhibition.

In conclusion, results from this study support the novel finding that arsenite exposure in keratinocytes suppresses Rad18 function in TLS. Rad18 is highly conserved in eukaryotes emphasizing the importance of this protein in preventing TLS imbalance and associated loss of genomic integrity (Gaillard et al., 2015; Ma et al., 2020; Vaziri et al., 2016). Arsenite tips the balance by binding and inhibiting the Rad18 RING finger, which may underly the arsenite-induced reduction in PCNA monoubiquitination, Polη recruitment, post-UV replication, and increase in ssDNA and DNA strand breaks in UV exposed keratinocytes. Overall, results from this study exemplify the complexity of arsenite on DNA damage repair and mutagenic pathways and reveal for the first time a potential role of TLS in the cocarcinogenicity of arsenic.

Highlights.

• Arsenite binds and disrupts the Rad18 zinc finger domains

• UV-induced PCNA monoubiquitination is decreased by arsenite

• Arsenite decreases TLS factor localization and chromatin recruitment

• Arsenite and Rad18 deficiency promote replication stress

Abbreviations

BSA

bovine serum albumin

CPD

cyclobutane pyrimidine dimer

Da

Dalton

DAPI

4’,6-Diamidino-2-Phenylindole Dihydrochloride

DDT

DNA damage tolerance

EdU

5-Ethynyl-2’-deoxyuridine

DSB

double-strand break

HEKn

normal human neonatal epidermal keratinocytes

ESI

electrospray ionization

LC

liquid chromatography

MALDI-TOF

matrix-assisted laser desorption/ionization - time of flight

MIP

max intensity projection

MS

mass spectrometry

ND

no damage

NT

no treatment

PARP

poly(ADP-ribose) polymerase

PCNA

proliferating cell nuclear antigen

PH2AX

phospho-histone H2A.X (Ser139)

Polη

DNA polymerase eta

RING

really interesting new gene

RPA

replication protein A

SCR

scrambled

SEM

standard error of the mea

siRNA

small interfering RNA

ssDNA

single-stranded DNA

TLS

translesion synthesis

TPEN

(N,N,N′,N′-tetrakis(2-pyridinylmethyl)-1,2-ethanediamine)

UBZ

ubiquitin-binding zinc finger

UV

ultraviolet radiation

XPA

Xeroderma Pigmentosum Complementation group A