Arsenite Exposure Displaces Zinc from ZRANB2 Leading to Altered Splicing

Abstract

Exposure to arsenic, a class I carcinogen, affects 200 million people globally. Skin is the major target organ but the molecular etiology of arsenic-induced skin carcinogenesis remains unclear. As³⁺-induced disruption of alternative splicing could be involved, but the mechanism is unknown. Zinc finger proteins play key roles in alternative splicing. Arsenite (As³⁺) can displace zinc (Zn²⁺) from C3H1 and C4 zinc finger motifs (zfms), affecting protein function. ZRANB2, an alternative splicing regulator with two C4 zfms integral to its structure and splicing function was chosen as a candidate for this study. We hypothesized that As³⁺ could displace Zn²⁺from ZRANB2 altering its structure, expression and splicing function. As³⁺/Zn²⁺ binding and mutual displacement experiments were performed with synthetic apo-peptides corresponding to each ZRANB2 zfm, employing a combination of intrinsic fluorescence, UV spectrophotometry, zinc colorimetric assay and liquid chromatography-tandem mass spectrometry. ZRANB2 expression in HaCaT cells acutely exposed to As³⁺ (0 or 5 μM; 0-72 h, or 0-5 μM; 6 h) was examined by RT-qPCR and immunoblotting. ZRANB2-dependent splicing of TRA2B mRNA, a known ZRANB2 target, was monitored by RT-PCR. As³⁺ bound to, as well as displaced Zn²⁺ from, each zfm. Also, Zn²⁺ displaced As³⁺ from As³⁺-bound zfms acutely, albeit transiently. As³⁺ exposure induced ZRANB2 protein expression between 3-24 h and at all exposures tested, but not ZRANB2 mRNA expression. ZRANB2-directed TRA2B splicing was impaired between 3-24 h post-exposure. Furthermore, ZRANB2 splicing function was also compromised at all As³⁺ exposures, starting at 100 nm. We conclude that As³⁺ exposure displaces Zn²⁺ from ZRANB2 zfms, changing its structure and compromising splicing of its targets, and increases ZRANB2 protein expression as a homeostatic response both at environmental/toxicological exposures and therapeutically relevant doses.

Graphical Abstract

INTRODUCTION

About 200 million individuals in over 70 countries around the globe are chronically exposed to arsenic ¹. Most exposure is via contaminated groundwater, although, exposure from food, occupational and industrial sources are also known ². Acute iatrogenic exposure, as with anti-leukemia chemotherapy ³ or in traditional medicines ⁴ are also well documented. Chronic exposure results in the development of long term adverse health effects, including multi-organ cancers, non-cancerous skin lesions, as well as higher risk of diabetes, peripheral neuropathy, cardiovascular disorders, etc. ^{1, 2}. Skin cancers, including Bowen’s disease, basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cSCC) are the most common forms of arsenic-induced cancer. Squamous cell carcinoma is an insidious form of malignant skin cancer with high rates of recurrence ⁵. While ultraviolet (UV) radiation is the most well-known and characterized cause of cSCC, arsenic is the second most common. Interestingly, arsenic-induced cSCC appears almost exclusively on covered areas of the body, indicating a distinct molecular etiology compared to UV-induced cSCC. cSCC also can result from sub-chronic iatrogenic arsenic exposures ⁶.

How arsenic causes skin cancer at the molecular level has been fiercely debated for a considerable period of time. Several hypotheses have been propounded and questioned based on ambiguous experimental design and data that vary with wide disparity in dosage, period of exposure and model cell line/organism used ^7-9. Most common theories include induction of DNA damage, clastogenesis, aneuploidy, induction of reactive oxygen/nitrogen species, impairment of DNA damage response and repair, and miRNA dysregulation ^10-13. Another mechanism that has come to the forefront of cancer research in the last decade is alternative mRNA splicing. Alternative splicing is a highly regulated process that generates transcriptomic and proteomic diversity by producing multiple mRNA isoforms from a single gene. The process is tightly regulated and in association with nonsense mediated mRNA decay, maintains the cellular profile of canonical protein isoforms under normal conditions. However, dysregulation of alternative splicing can result in adverse health outcomes ^14-16. Mounting evidence suggests that dysregulated alternative splicing is a hallmark of carcinogenesis, as well as a potential therapeutic target ^15-20. There is also evidence suggesting that individual alternative splicing events are actually components of genome-wide coordinated change of multiple, co-regulated alternative splicing events rather than being isolated incidents ²¹. Although the mechanisms of such coordination are not well understood, it does put the spotlight onto regulators of alternative splicing as putative targets in carcinogenesis ²².

Many of the proteins involved in splicing contain zinc finger motifs (zfms) ²³. These zfms are absolutely critical for RNA binding and consequently splicing ^24-26. Recent research suggests As³⁺ exposure leads to dysregulation of critical DNA repair proteins which often contain zfms ^{27, 28-30}. As³⁺ binds to C3H1 (3 cysteine and 1 histidine residues) and C4 (four cysteine residues) zfms causing Zn²⁺ loss from these sites ^{28, 31, 32}. Such binding and consequent Zn²⁺ loss leads to experimentally demonstrable loss/alteration of protein function ^32-34. In fact, abrogation of the function of zfms in DNA repair proteins have been postulated to be a key mechanism in carcinogenesis ³⁵. It is thus reasonable to hypothesize that any splice factor containing C3H1 or C4 zfms is a potential target for As³⁺ mediated Zn²⁺ displacement, leading to functional abrogation. However, no study has ever looked at As³⁺ mediated Zn²⁺ displacement in any splicing factor. One study documented 104 differential alternative splicing events upon As³⁺ exposure in the BEAS-2B cell line ³⁶. This observation further suggests that exposure to As³⁺ could be altering the mechanism of splicing by hitherto unknown mechanisms. This gap in understanding prompted us to determine if As³⁺ is capable of displacing Zn²⁺ from a splicing regulatory protein, altering its structure and/or function.

In the present study, we chose ZRANB2 as a zfm containing splice regulator protein of interest. ZRANB2 is an evolutionarily conserved zfm protein with a well-validated role in the regulation of alternative splicing. The protein contains two C4 zfms, each of which recognizes an AGGUAA site and binds to single stranded RNA with high specificity ³⁷. ZRANB2 is responsible for splice site choice that drives alternative splicing and isoform specific gene expression regulation ^37-40. Identified splice targets of ZRANB2 include mRNAs encoding TRA2B, DRD2, WDR78 and ACAP1 among others ^39-41. Furthermore, ZRANB2 physically interacts with core spliceosomal components U2AF1 and SNRNP70 and this interaction is hypothesized to stabilize the splicing complex ^{38, 40}. Given its established role as a splice regulator, coupled with the presence of two C4 zfms, ZRANB2 is an ideal candidate for the present study.

Here we demonstrate that As³⁺ binds to each of the two ZRANB2 zfms displacing Zn²⁺ and altering the structure of these motifs. In addition, we demonstrate that acute As³⁺ treatment of HaCaT cells leads to changes in ZRANB2 expression as well as differential splicing of ZRANB2 target TRA2B (both at environmentally relevant and therapeutic doses), itself an alternative splicing factor ^{42, 43}. The present study provides a mechanistic understanding of how As³⁺ exposure can modulate the process of alternative splicing and protein isoform expression profiles.

MATERIALS AND METHODS

Caution: Sodium arsenite is a potent carcinogen and should be handled in accordance with NIH Guidelines for the Use of Chemical Carcinogens.

Chemicals.

Synthetic apo-peptides corresponding to the two ZRANB2 zfms (For sequences, see Table 1) were obtained from GenScript Corp. (Piscataway, NJ, USA) at a purity >95%. Zinc chloride (ZnCl₂; CAS 7646-85-7) was obtained from Millipore-Sigma (St. Louis, MO, USA). Fetal Bovine Serum (characterized) was obtained from Hyclone (Logan, UT, USA). Sodium arsenite (NaAsO₂; CAS 7784-0698), Tris(2-carboxyethyl)phosphine hydrochloride (TCEP HCl; CAS 5180545-9) and UltraPure™ DNase/RNase-Free Distilled Water, MEM alpha modification media, trypsin, ethylene diamine tetraacetic acid, penicillin/ streptomycin and L-glutamine were obtained from Thermo Fisher Scientific Inc. (Waltham, MA, USA), as were all other chemicals unless mentioned specifically.

Table 1.

Sequence of ZRANB2 Zfm Apo-peptides

Name	Peptide sequence	Length	Predicted MW (Da)	Predicted MW - Fully Reduced (Da)	Location in ZRANB2 Protein
ZNF1	MSTKNFRVSDGDWICPDKKCG NVNFARRTSCNRCGREKTT	40 a.a.	4583.19	4581.15	1-40
ZNF2	AEKSRGLFSANDWQCKTCSN VNWARRSECNMCNTPKYA	38 a.a	4370.89	4368.96	57-94

Preparation of Peptide and Metal Solutions.

Each lyophilized apo-peptide was dissolved to a final concentration of 1.25 mM in 0.25 mM TCEP prepared in ultrapure water. Peptide solutions were stored at −80 °C in single thaw aliquots until use. Stock solutions of As³⁺ and Zn²⁺ were prepared freshly in ultrapure water immediately before use and serially diluted using ultrapure water, as required.

Metal Binding/Displacement Assays.

In order to determine if one or both of the ZRANB2 zfms bind As³⁺ and/or Zn²⁺ and whether these metals can mutually displace each other from either or both of these zfms, we employed a combination of several biophysical techniques, as well as mass-spectrometry as described in the next sections. All these assays were performed using synthetic apo-peptides in cell free systems.

UV-Vis Spectrophotometry.

UV-Vis spectrophotometric experiments were performed to characterize the interaction of As³⁺ binding and mutual displacement of As³⁺/Zn²⁺ from ZRANB2 zfm apo-peptides as described elsewhere with minor modifications ^{32, 44}. Briefly, 100 μM of each ZRANB2 apo-peptide was incubated with increasing molar equivalents (Eq) of As³⁺ (0–2 Eq at 0.2 Eq intervals; 0-200 μM at 20 μM intervals) for 30 minutes at room temperature. For Zn²⁺ displacement by As³⁺, 100 μM of each ZRANB2 apo-peptide was initially incubated with 0.5 Eq Zn²⁺ (50 μM) for 30 minutes at room temperature, followed by titration with increasing molar equivalents of As³⁺ (0-2 Eq at 0.25 Eq intervals; 0-200 μM at 25 μM intervals) for 30 minutes at room temperature. At As³⁺ concentrations higher than 1 Eq following the addition of 1 Eq Zn²⁺ to the apo-peptides, particulates were seen to form in the reaction mixture, hindering the spectrophotometric detection. Consequently, for Zn²⁺ displacement reactions for UV-Vis spectrophotometric assay, 0.5 Eq of Zn²⁺ was used. For As³⁺ displacement by Zn²⁺, 100 μM of each ZRANB2 apo-peptide was initially incubated with 1 Eq As³⁺ (100 μM) for 30 minutes at room temperature, followed by titration with increasing molar equivalents of Zn²⁺ (0-2 Eq at 0.25 Eq intervals; 0-200 μM at 25 μM intervals) for 30 minutes at room temperature. All the reactions were performed in 20 mM Tris-HCI, pH 7.8 with 250 μM TCEP in a volume of 10 μL. Spectrophotometric measurements were performed using a NanoDrop™ One Microvolume UV-Vis Spectrophotometer (Thermo Fisher Scientific Inc.) in the wavelength range 230-400 nm ^{44, 45}. A buffer solution containing 20 mM Tris-HCI, pH 7.8 and 250 μM TCEP was used as the blank to set the baseline for the entire wavelength range. Absorbance values at 285 nm (A₂₈₅) were recorded in each case. Means of duplicate readings were taken for each sample. In another set of As³⁺ displacement by Zn²⁺ reactions, the experiments were performed with equimolar amounts of apo-peptide, As³⁺ and Zn²⁺ as described earlier in this section, but in a 20 μL reaction volume. Half of the reaction mixture (10 μL) was used for spectrophotometric measurements as described earlier. The remaining half was frozen at −20°C and spectrophotometric measurements were recorded 24 h post-freezing. Each experiment was performed independently three times.

Free Zn²⁺ Measurement:

Spectrophotometric experiments were performed to characterize the interaction of Zn²⁺ binding and mutual displacement of As³⁺/Zn²⁺ from ZRANB2 zfm apo-peptides as described previously with minor modifications³¹. Briefly, increasing amounts of each ZRANB2 apo-peptide (0-5 μM) was incubated with 10 μM Zn²⁺ for 30 minutes at room temperature. For Zn²⁺ displacement by As³⁺, 5 μM of each ZRANB2 apo-peptide was initially incubated with 10 μM Zn²⁺ for 30 minutes at room temperature, followed by titration with increasing concentration of As³⁺ (0-5 μM) for 30 minutes at room temperature. For As³⁺ displacement by Zn²⁺, 5 μM of each ZRANB2 apo-peptide was initially incubated with 5 μM As³⁺ for 30 minutes at room temperature, followed by titration with increasing concentration of Zn²⁺ (0-10 μM) for 30 minutes at room temperature. All the reactions were performed in 20 mM Tris-HCI, pH 7.0 with 25 μM TCEP. All the reactions were performed in a volume of 10 μL apart from the As³⁺ displacement by Zn²⁺ experiments, which were performed in 20 μL. After the incubations were over, 1 μL of 1 mM 4-(2-pyridylazoresorcinol) was added to each reaction mixture and spectrophotometric measurements were performed using a NanoDrop™ One Microvolume UV-Vis Spectrophotometer (Thermo Fisher Scientific Inc.) in the wavelength range 300-550 nm ⁴⁶. The absorbance values at 493 nm were recorded in each case as a measure of free Zn²⁺ signal in the samples. Means of duplicate readings were taken for each sample. Absorbance readings were converted to free Zn²⁺ concentration by comparison against a Zn²⁺ standard curve generated using A₄₉₃ values. The free Zn²⁺ concentration was subsequently subtracted from the input Zn²⁺ concentration to determine the concentration of apo-peptide bound Zn²⁺. For stability of Zn²⁺ binding (5 μM apo-peptide incubated with 10 μM Zn²⁺) and As³⁺ displacement by Zn²⁺ experiments, half of the reaction mixture (10 μL) was used for spectrophotometric measurements immediately, while the remaining half was frozen at −20°C and spectrophotometric measurements were recorded 24 h post-freezing. Each experiment was performed independently three times.

Mass Spectrometric Analyses.

Mass spectroscopy analyses were carried out for further characterization of As³⁺ binding and mutual displacement of As³⁺/Zn²⁺ from ZRANB2 zfm apo-peptides. For binding experiments, 1 mM of each ZRANB2 apo-peptide was incubated with As³⁺ (1 Eq; 1 mM) or Zn²⁺ (0.5 or 1 Eq; 0.5 or 1 mM) for 30 minutes at room temperature. For Zn²⁺ displacement by As³⁺, 1 mM of each ZRANB2 apo-peptide was initially incubated with 0.5 Eq Zn²⁺ (0.5 mM) for 30 minutes at room temperature, followed by titration with 2 Eq of As³⁺ (2 mM) for 30 minutes at room temperature. For As³⁺ displacement by Zn²⁺, 1 mM of each ZRANB2 apo-peptide was initially incubated with 1Eq As³⁺ (1 mM) for 30 minutes at room temperature, followed by titration with 1 Eq of Zn²⁺ (1 mM) for 30 minutes at room temperature. All the reactions were performed in 20 mM Tris-HCI, pH 7.8 with 2.5 mM TCEP in a volume of 10 μL. Samples were diluted with 2% v/v acetonitrile / 0.1% v/v formic acid to a final concentration of 1 pmol peptide/μL. Sample aliquots were dispensed into capped autosampler vials and loaded into a Waters M-Class Acquity UPLC system (Waters Corp., Milford, MA, USA) fitted with a Waters CSH C18 130 Å 1.7 μm 300 μm x 150 mm column, heated at 55 °C, at a flow rate of 10 μL/min. Data were acquired on the Synapt G2-Si Q-Tof (Waters Corp.) using the scan range 400-1500m/z with an MSe method. The .raw data were loaded into MassLynx4.1 (Waters Corp., Milford, MA, USA) and processed with MaxEnt3 (Waters Corp., Milford, MA, USA) for deconvolution and deisotoping to produce precursor and fragment ion mass lists for Skyline template generation. The .raw data were imported into Skyline v4.2.0.19072 using MS1 extracted ion chromatograms ⁴⁷. For ZNF1 (+5, +6, +7, and +8 charge states) and ZNF2 (+4, +5, +6, and +7 charge states) peptide data, precursor area was extracted for the top three theoretical isotopes of each monitored peptide using a TOF resolving power of 10,000. The area extracted for the fully and partially reduced forms was summed.

Intrinsic Fluorescence Assay.

Intrinsic fluorescence experiments were performed to characterize changes in the folding of the ZRANB2 zfm apo-peptides upon As³⁺/Zn²⁺ binding and mutual displacement of As³⁺/Zn²⁺, following previously described methods with minor modifications ^{46, 48}. All the reactions were performed in 20 mM Tris-HCI, pH 7.0 with 2.5 μM TCEP in a volume of 2.5 mL. Intrinsic fluorescence spectra were recorded using an excitation wavelength of 280 nm and an emission wavelength range of 290-500 nm (1 nm interval; 1 s integration time; 1 nm excitation bandpass; 5 nm emission bandpass) using the FluoroMax®-3 spectrofluorometer (Jobin Yvon Inc. Edson, NJ, USA). For metal binding experiments, 1 μM of each ZRANB2 apo-peptide was incubated with equimolar amounts of As³⁺ or Zn²⁺ for 30 minutes at room temperature followed by collection of intrinsic fluorescence spectra as described earlier. For Zn²⁺ displacement by As³⁺, 1 μM of As³⁺ was added to Zn²⁺-bound ZRANB2 apo-peptide and incubated for an additional 30 minutes at room temperature, followed by collection of intrinsic fluorescence spectra as described earlier. For As³⁺ displacement by Zn²⁺, 1 μM of Zn²⁺ was added to As³⁺-bound ZRANB2 apo-peptide and incubated for an additional 30 minutes at room temperature, followed by collection of intrinsic fluorescence spectra as described earlier. Total intrinsic fluorescence over the wavelength range was determined in each case by calculating the area under the curve (AUC). Each experiment was performed independently six times.

Cytotoxicity Assay.

Cytotoxicity in HaCaT cells due to As³⁺ exposure was quantitated using alamar blue assay. Alamar blue was obtained from Bio-Rad (Hercules, CA, USA) and the assay was performed using the manufacturer’s protocol with minor modifications⁴⁹. Briefly, 10⁴ cells were seeded per well in a 96 well plate in a volume of 100 μL. Cells were subsequently treated with increasing concentration of As³⁺ (final concentration 0-10 μM) 24 h post seeding in quadruplicates. Alamar blue was added at different time points (24, 48, 72 h post-treatment), allowed to incubate at 37 °C for 4 h and spectrophotometric readings were taken at 570 and 600 nm using Gen5™ microplate reader (Winooski, VT, USA). The readings were subsequently converted to percentage of live cells compared to untreated control using the equation and the molar extinction coefficient values provided by the manufacturer⁴⁹.

Cell Culture and As³⁺ Treatment.

HaCaT cells were the kind gift of Dr. Tai Hao Quan, University of Michigan. The identity of the cells as HaCaT was confirmed by STR mapping (Genetica DNA Laboratories/LabCorp, Burlington, NC, USA). The cells were cultured as described previously⁵⁰ with minor modifications. Briefly, cells were cultured in MEM alpha modification media supplemented with 10% fetal bovine serum, 100 U/mL penicillin/100 mg/mL streptomycin) and 2 mM L-glutamine. Cultures were maintained at 37°C in a humidified 5% CO₂ atmosphere. Twenty-four hours post-seeding, the cells were treated with NaAsO₂ by addition of 1000X stock to final [As³⁺] = 5 μM (or equal volume of ultrapure water as vehicle control). Cells were harvested for RNA and protein extraction 0, 1, 3, 6, 12, 24, 48 and 72 h post-treatment. For the concentration-response experiments, the HaCaT cells were treated with 0, 0.1, 1 and 5 μM As³⁺ for 6 h. Three independent replicates were obtained per condition per time point/concentration.

RNA Extraction and cDNA Generation.

Total RNA was purified from the cells employing the mirVana RNA Isolation Kit (Thermo Fisher Scientific Inc.) as described previously⁵⁰. RNA quality was determined using the Agilent RNA 6000 Pico Kit, Eukaryote, version 2.6 and the Agilent 2100 Bioanalyzer instrument (Agilent Technologies Inc., Santa Clara, CA, USA). All samples had RIN (RNA integrity number) > 9 ⁵⁰. cDNA was generated from total RNA using the PrimeScript RT Reagent Kit (TaKaRa Bio USA Inc., Mountain View, CA, USA) as per the manufacturer’s instructions. Briefly, 1 μg of total RNA was reverse transcribed into cDNA using Oligo (dT)₂₀ primer.

Quantitative PCR.

Quantitative PCR of cDNAs was performed using PowerUp SYBR Green Master Mix (Thermo Fisher Scientific) following the manufacturer’s recommended protocol. Briefly, 5 ng of cDNA was amplified using highly specific primers (200 nM each): ZRANB2 assay Hs.PT.58.40950918 (Integrated DNA Technologies, Coralville, IA, USA), and GAPDH forward 5’ ACCACAGTCCATGCCATCAC 3’ and reverse 5’ TCCACCACCCTGTTGCTGTA 3’ (Integrated DNA Technologies) in a final volume of 20 μL. The Viia 7 Real-Time PCR System (Thermo Fisher Scientific) was used for amplification and relative quantitation. All samples were run at least in triplicate, using GAPDH as the housekeeping gene. To determine the effect of arsenic exposure on the expression of ZRANB2 mRNA, the threshold cycle (Ct) values for GAPDH from each sample was subtracted from the Ct values for ZRANB2 from the respective sample to obtain a ΔCt value. An average for the ΔCt values was taken for the replicates. ΔCt values for controls at each time point were averaged and this average was subtracted from each ΔCt value for each sample at that time point to obtain a ΔΔCt value. The relative expression of ZRANB2 was calculated as fold change = 2^(−ΔΔCt). For each time point, the mean of control samples was taken as 1 fold and the As³⁺ exposed samples were expressed as fold change over control.

Analysis of Alternative Splicing.

RT-PCR was performed using cDNA generated from control or As³⁺ treated HaCaT cells, as described in the preceding sections. Primers were designed against flanking exons 1 (Forward; 5’- AAGGAAGGTGCAAGAGGTTG-3’) and 4 (Reverse; 5’- CGGCAATGGGACCATATTTA-3’) of TRA2B gene, a known splice target of ZRANB2, and obtained from Integrated DNA Technologies. The primers were designed to co-amplify β1 (502 bp) and β3 (368 bp) TRA2B mRNA isoforms ⁴⁰ in a single reaction. Each amplification reaction was performed using 400 ng cDNA for 36 cycles using 54 °C as the annealing temperature. The amplification products were resolved on 1.5% agarose gels pre-stained with RedSafe™ dye (Bulldog Bio, Portsmouth, NH, USA). Images were acquired with FOTO/Analyst FX (Thermo Fisher Scientific Inc.) and densitometric analysis by employing Image J software ⁵¹. Raw data from densitometric analyses were divided by the isoform size (fragment length in bp) to determine the number of events for the β1 and β3 isoforms. Percentage contribution of β3 isoform was determined in each sample under each condition using the following formula: β3 isoform (%) = β3*100/(β1+β3).

Immunoblotting.

Cell lysis and immunoblotting were performed as described previously⁵² with minor modifications. Briefly, cells were lysed with a solution of 10 mM Tris–HCI pH 7.4, 1 mM EDTA, 0.1% SDS, 180 μg/mL PMSF and 1X protease inhibitor cocktail (Thermo Fisher Scientific Inc.). Lysates were sonicated and centrifuged at 4° C for 15 min to remove insoluble debris. Protein concentrations were determined with BCA assay (Thermo Fisher Scientific Inc.). Proteins were resolved by electrophoresis in SDS polyacrylamide gels of appropriate percentage. The resolved proteins were electro-blotted onto PVDF membranes (Thermo Fisher Scientific Inc.). After staining with Coomassie Brilliant Blue R250 (Thermo Fisher Scientific Inc.) to ensure equal loading and transfer⁵³, membranes were blocked in 5% milk in TBST (10 mM Tris–HCI pH 7.4, 150 mM NaCI, 0.1% Tween 20) at room temperature for 1 h. Blots were subsequently probed with antibodies against Vinculin (Clone E19V; Cat No. 13901; Cell Signaling, Danvers, MA, USA; 1:1000 in 1% skimmed milk in TBST) and ZRANB2 (Clone B-5; Cat No. sc-514200; Santa Cruz Biotechnology, Dallas, TX, USA; 1:100 in 1% skim milk in TBST). Blots were incubated with HRP conjugated secondary antibody as required (in 1% skim milk in TBST), and subsequently with enhanced chemi-luminescent substrate (Thermo Fisher Scientific Inc.) or with SuperSignal™ West Atto Ultimate Sensitivity Substrate (Thermo Fisher Scientific Inc.), and images were acquired using FOTO/Analyst FX. Densitometric analyses were performed using Image J software⁵¹.

Statistical Analyses.

For UV-Vis spectrophotometric experiments to characterize As³⁺ binding to ZRANB2 zfm apo-peptides, blank subtracted A₂₈₅ values at each As³⁺ concentration were subtracted from the corresponding blank subtracted A₂₈₅ value of the native apo-peptide to calculate absorbance difference at 285 nm (ΔA₂₈₅) ΔA₂₈₅ values were plotted against As³⁺ concentration and fitted using one-site specific binding non-linear curve fitting function using GraphPad Prism 5. K_d values as well as r² values were determined for each curve fit. For mutual displacement of As³⁺/Zn²⁺ from ZRANB2 zfm apo-peptides experiments, blank subtracted A₂₈₅ values at each concentration was subtracted from the blank subtracted A₂₈₅ value of the native apo-peptide to calculate ΔA₂₈₅. ΔA₂₈₅ values were plotted against Zn²⁺/As³⁺ concentration and fitted using log(inhibitor) versus response non-linear curve fitting function using GraphPad Prism 5. For free Zn²⁺ measurements, the relationship between peptide bound Zn²⁺ and peptide/metal concentrations were fitted using linear regression using GraphPad Prism 5. For intrinsic fluorescence assays, the AUC values were log transformed, tested for normality by Shapiro-Wilk test, and analyzed initially by ANOVA with five levels followed by pairwise comparisons by Fisher’s least significant difference test (as the data were normally distributed). The displacement data for each peptide were subsequently added (ZNF1 Zn²⁺+As³⁺ Group: ZNF1+Zn+As combined with ZNF1+As+Zn; ZNF2 Zn²⁺+As³⁺ Group: ZNF2+Zn+As combined with ZNF2+As+Zn) and the four groups (No metal; Zn²⁺; As³⁺; Zn²⁺+As³⁺) were analyzed by two-way ANOVA. Densitometric data for time-course of ZRANB2 protein expression as well as percent of β3 isoform of TRA2B were analyzed using two-way ANOVA with Bonferroni multiple comparisons post-hoc tests (no normality or homoscedasticity assumption). Densitometric data for dose-response of ZRANB2 protein expression as well as percent of β3 isoform of TRA2B were analyzed using one-way ANOVA with Tukey’s multiple comparisons post-hoc tests. p<0.05 was considered to be significant.

RESULTS

UV-Vis Spectrophotometric Analysis of As³⁺/Zn²⁺ Interactions with ZRANB2 Zfm Apo-peptides.

Binding of As³⁺ to sulfhydryl groups of cysteine residues can be monitored by a change in the UV-Vis spectrum between 230-400 nm due to the formation of charge transfer electronic transition ^{44, 45}. We initially characterized the spectrum of each native ZRANB2 apo-peptide in this wavelength range (Supplementary Figure 1). Analysis of the spectrum showed that for each apo-peptide, the absorption maximum was at 285 nm (A₂₈₅), (Supplementary Figure 1). ΔA₂₈₅ values were used to quantify As³⁺ binding as well as mutual displacement of As³⁺/Zn²⁺ from ZRANB2 zfm apo-peptides in all subsequent UV-Vis experiments. As³⁺ binding experiments showed As³⁺ concentration-dependent increase in ΔA₂₈₅ (Figure 1A-B and Supplementary Figure 2A-B; Supplementary Table 1). For As³⁺ mediated Zn²⁺ displacement, we also observed a concentration-dependent increase in the ΔA₂₈₅, as expected, suggesting more and more As³⁺ was binding to each of the apo-peptides at the expense of Zn²⁺ (Figure 1C-D and Supplementary Figure 2C-D). Conversely, forZn²⁺ mediated As³⁺ displacement, we found a concentration-dependent decrease in the ΔA₂₈₅, suggesting reduction in As-S bonds as As³⁺ was progressively displaced by Zn²⁺ (Figure 1E-F and Supplementary Figure 2E-F).

Figure 1.

As³⁺ binding and mutual displacement of Zn²⁺/As³⁺ from ZRANB2 zfm apo-peptides using UV-Vis spectrophotometry. A, As³⁺ binding to ZNF1 (one-site specific binding non-linear curve fit). B, As³⁺ binding to ZNF2 (one-site specific binding non-linear curve fit). C, Zn²⁺ displacement by As³⁺ from ZNF1 (log(inhibitor) versus response non-linear curve fit). D, Zn²⁺ displacement by As³⁺ from ZNF2 (log(inhibitor) versus response non-linear curve fit). E, As³⁺ displacement by Zn²⁺ from ZNF1 (log(inhibitor) versus response non-linear curve fit). F, As³⁺ displacement by Zn²⁺ from ZNF2 (log(inhibitor) versus response non-linear curve fit). In each panel, the closed circle represents ΔA₂₈₅ values. Each point in each curve represents the Mean ± SD from three independent experiments.

Zinc Colorimetric Analysis of As³⁺/Zn²⁺ Interactions with ZRANB2 Zfm Apo-peptides.

While we could demonstrate As³⁺ binding and mutual displacement of As³⁺/Zn²⁺ from ZRANB2 zfm apo-peptides with UV-Vis assay, we could not quantify Zn²⁺ binding directly by that assay. In order to demonstrate that Zn²⁺ can bind to one or both ZRANB2 zfm apo-peptides, we performed a Zn²⁺ colorimetric assay. For Zn²⁺ binding experiments, the concentration of apo-peptide bound Zn²⁺ increased linearly with apo-peptide concentration (Figure 2A). Conversely, for As³⁺ mediated Zn²⁺ displacement experiments, the concentration of apo-peptide bound Zn²⁺ decreased linearly with As³⁺ concentration (Figure 2B). For both these experiments, a tangible difference was observed between the two apo-peptides with respect to how Zn²⁺ interacts with each of them. For Zn²⁺ mediated As³⁺ displacement, the concentration of apo-peptide bound Zn²⁺ was found to increase linearly with increase in Zn²⁺ concentration (Figure 2C), demonstrating Zn²⁺ was capable of displacing As³⁺ from each apo-peptide.

Figure 2.

Zn²⁺ Binding and displacement of Zn²⁺/As³⁺ from ZRANB2 zfm apo-peptides using spectrophotometric quantification of free Zn²⁺. A, Increasing amounts of each ZRANB2 apo-peptide (0-5 μM) was incubated with 10 μM Zn²⁺ for 30 minutes at room temperature. Increase in peptide bound Zn²⁺ with increase in peptide concentration shows Zn²⁺ binding to the peptides. B, 5 μM of each ZRANB2 apo-peptide was initially incubated with 10 μM Zn²⁺ for 30 minutes at room temperature, followed by titration with increasing concentration of As³⁺ (0-5 μM) for 30 minutes at room temperature. Decrease in peptide bound Zn²⁺ with increase in As³⁺ concentration shows displacement of Zn²⁺ by As³⁺ from each peptide. C, 5 μM of each ZRANB2 apo-peptide was initially incubated with 5 μM As³⁺ for 30 minutes at room temperature, followed by titration with increasing concentration of Zn²⁺ (0-10 μM) for 30 minutes at room temperature. Increase in peptide bound Zn²⁺ with increase in Zn²⁺ concentration shows displacement of As³⁺ by Zn²⁺ from each peptide. Open circles represent ZNF1 while open squares represent ZNF2 for all the panels. Each point in each curve represents the Mean ± SD from three independent experiments.

Mass Spectrometric Characterization of As³⁺/Zn²⁺ Interactions with ZRANB2 Zfm Apo-peptides.

The MS2 spectrum of each apo-peptide (+6 charge state) is shown in Figure 3 A-B. The experimentally determined mass at +6 charge state (764 for ZNF1 and 729 for ZNF2) closely matches the theoretical mass of the fully reduced apo-peptides at the same m/z state (763.8 for ZNF1 and 728.3 for ZNF2). Upon incubation with arsenic alone, the mass shift for each apo-peptide corresponds to As³⁺ bound form. Each peptide had a characteristic +12 m/z shift (776 for ZNF1 and 741 for ZNF2; Figure 3C-D) at +6 charge state. Similarly, upon incubation with Zn²⁺ followed by As³⁺, the same +12 m/z shift (Figure 3E-F) was seen for each peptide (at +6 charge state), characteristic of As³⁺-bound apo-peptides, suggesting that As³⁺ displaces Zn²⁺ from Zn²⁺-bound ZRANB2 apo-peptides. However, we could not show any mass shift for apo-peptides when incubated with Zn²⁺ alone (Supplementary Figure 3A-B). This is not entirely unexpected as LC-MS analysis of zfm peptides is known to extract Zn²⁺ from the peptides, leading to a significant reduction in the mass shift signal ⁵⁴. Interestingly, upon incubation with As³⁺ followed by Zn²⁺, +12 m/z shift was seen (at +6 charge state) for each peptide (Supplementary Figure 3C-D), characteristic of As³⁺-bound peptides. This was surprising, because, we expected to see mass signature of the native apo-peptides, but not of the As³⁺-bound peptides, had it been an outcome of Zn²⁺ extraction from the peptides during ESI-MS/MS analysis.

Figure 3.

Mass spectrometric (ESI-MS/MS) analysis of As³⁺ binding and displacement of Zn²⁺ by As³⁺ from ZRANB2 Zfm Apo-peptides. A, Native ZNF1 apo-peptide mass signature was detected at m/z = 764. B, Native ZNF2 apo-peptide mass signature was detected at m/z = 729. C, As³⁺ binding to ZNF1 was detected at m/z = 776, which represents a +12 m/z shift compared with the native ZNF1 apo-peptide (m/z = 764). D, As³⁺ binding to ZNF2 was detected at m/z = 741, which represents a +12 m/z shift against the native ZNF2 apo-peptide (m/z = 729). E, Zn²⁺ displacement by As³⁺ from ZNF1 was detected at m/z = 776, which represents a +12 m/z shift against the native ZNF1 apo-peptide (m/z = 764). F, Zn²⁺ displacement by As³⁺ from ZNF2 was detected at m/z = 741, which represents a +12 m/z shift against the native ZNF2 apo-peptide (m/z = 729). All the figures represent +6 charge state. Insets represent maginified view of the highest peak in each case.

Intrinsic Fluorescence Analysis of Structural Changes in ZRANB2 Apo-peptides upon Interaction with As³⁺/Zn²⁺.

Having demonstrated that each ZRANB2 zfm apo-peptide can bind both As³⁺ and Zn²⁺, and that each can displace the other, we wanted to examine if such interactions could affect the tertiary structure of these apo-peptides. For this purpose, change in intrinsic fluorescence of the apo-peptides was monitored ⁴⁸. The data indicate that Zn²⁺ binding expectedly promotes folding in each peptide as evidenced by an increase in fluorescence intensity (Figure 4A-B), supporting the notion that the Zn²⁺-bound form represents the natural folded conformation under physiological conditions ⁴⁶ that is necessary for RNA binding. As³⁺ addition, interestingly, promoted folding for ZNF1 but unfolding for ZNF2 (Figure 4A-B). Zn²⁺ also promoted folding of each peptide under Zn²⁺ mediated As³⁺ displacement experimental conditions (Fig 4A-B). Interestingly, there was no difference in intrinsic fluorescence (and hence in folding), compared to Zn²⁺-bound peptides, when As³⁺ was added to the Zn²⁺-bound peptides for the As³⁺ mediated Zn²⁺ displacement reactions (Figure 4A-B). These data suggest that interaction with As³⁺ alters the tertiary structure of either apo-peptide to a lesser extent than interaction with Zn²⁺ whether in the unfolded state (as compared to the unfolded apo-peptide for the As³⁺ binding experiment) or the folded state (as compared to the Zn²⁺ bound peptide for the As³⁺ mediated Zn²⁺ displacement experiment). Since the order of addition of As³⁺ or Zn²⁺ to the apo-peptides did not make a difference to the intrinsic fluorescence/folding (ZNF1+Zn+As vs. ZNF1+As+Zn and ZNF2+Zn+As vs. ZNF2+As+Zn), these two groups were merged (ZNF1 Zn²⁺+As³⁺ Group: ZNF1+Zn+As combined with ZNF1+As+Zn; ZNF2 Zn²⁺+As³⁺ Group: ZNF2+Zn+As combined with ZNF2+As+Zn) and a two way ANOVA analysis with interaction was carried out (No metal; Zn²⁺; As³⁺; Zn²⁺+As³⁺) to examine the effect of each metal individually as well as their potential interaction in terms of ZRANB2 apo-peptide folding (Table 2, Supplementary Fig 4). p Values indicate that addition of either As³⁺ or Zn²⁺ individually to each apo-peptide induces tangible structural changes compared to the native apo-peptide. A significant interaction between the two metals was found for ZNF1, but not for ZNF2 (Table 2; Supplementary Figure 4).

Figure 4.

Zn²⁺/As³⁺ binding and mutual displacement induces changes in intrinsic fluorescence of ZRANB2 apo-peptides. A, Intrinsic fluorescence of the native ZNF1 apo-peptide changes significantly with both As³⁺ and Zn²⁺ binding. Intrinsic fluorescence of Zn²⁺-bound ZNF1 does not change after displacement of Zn²⁺ by As³⁺; whereas displacement of As³⁺ by Zn²⁺ significantly alters the intrinsic fluorescence of As³⁺-bound ZNF1. However, there is no difference in intrinsic fluorescence when both As³⁺ and Zn²⁺ are added to ZNF1, irrespective of the order of addition (ZNF1+Zn+As vs. ZNF1+As+Zn). ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05 for pairwise comparison by Fisher’s least significant difference test, following one-way ANOVA with five levels. B, Intrinsic fluorescence of the native ZNF2 apo-peptide changes significantly with both As³⁺ and Zn²⁺ binding. Intrinsic fluorescence of Zn²⁺-bound ZNF2 does not change after displacement of Zn²⁺ by As³⁺; whereas displacement of As³⁺ by Zn²⁺ significantly changes the intrinsic fluorescence of As³⁺-bound ZNF2. However, there is no difference in intrinsic fluorescence when both As³⁺ and Zn²⁺ are added to ZNF2, irrespective of the order of addition (ZNF2+Zn+As vs. ZNF2+As+Zn). ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05 for pairwise comparison by Fisher’s least significant difference test, following one-way ANOVA with five levels. Please see Table 2 for two-way ANOVA analyses of the data.

Table 2:

Two-way ANOVA p values for As³⁺/Zn²⁺ interaction with ZRANB2 zfm apo-peptides

Parameter	ZNF1	ZNF2
pZn	<.0001	<.0001
pAs	0.0058	0.0004
pInteraction	0.0006	0.3418

Zn²⁺ Mediated As³⁺ Displacement from ZRANB2 Zfm Apo-peptides is Transient.

The MS data suggested that Zn²⁺ was not able to bind to or displace As³⁺ from either peptide, in contrast to the results obtained using spectrophotometry or colorimetric free Zn²⁺ analysis. While the apparent lack of Zn²⁺ binding could be explained by possible Zn²⁺ extraction from the apo-peptides during LC-ESI-MS analysis ⁵⁴, we expected to see the mass signature of the native apo-peptides rather than the As³⁺-bound peptides in the As³⁺ displacement experiments. We were intrigued by the appearance of the mass signature of As³⁺ bound peptides in these experiments. Given the samples were stored overnight at −20°C before being analyzed by mass spectrometry, we addressed the hypothesis that As³⁺ could re-displace Zn²⁺ in the intervening time period. Consequently, we performed several experiments using different biophysical techniques to address this question. First, we wanted to examine if Zn²⁺ binding to the apo-peptides was stable. There was no difference in the amount of peptide bound Zn²⁺ determined by measuring remaining free Zn²⁺, whether the readings were taken immediately or following storage at −20°C for 24 h (Figure 5A). Having demonstrated that Zn²⁺ does not come off from the apo-peptides because of storage, we performed experiments to quantify if the Zn²⁺ was displaced by As³⁺ during storage. We demonstrate that most of the apo-peptide bound Zn²⁺ was lost from each apo-peptide after storage for 24 h at −20 °C in the presence of As³⁺ (Fig 5B). Next, we wanted to explore whether Zn²⁺ was simply being displaced from the apo-peptides, or As³⁺ was re-binding to the apo-peptides. To test this, we performed UV-Vis analysis of Zn²⁺ mediated As³⁺ displacement experiments immediately upon the completion of the displacement reaction and after 24 h storage at −20 °C. We found that there was no change in ΔA₂₈₅ for As³⁺ bound apo-peptide compared to native apo-peptide between the two time points, indicating that As³⁺ was bound to each of the apo-peptides similarly at both time points (Figures 5B-C). Interestingly, for the Zn²⁺ mediated As³⁺ displacement condition, there was a significant decrease in ΔA₂₈₅ compared to As³⁺ bound apo-peptide, immediately after the completion of the displacement reaction (Figure 5C-D), indicating a decrease in As-S bond formation owing to As³⁺ displacement. However, the ΔA₂₈₅ values were similar to As³⁺ bound apo-peptide following 24 h storage at −20°C. These results suggest, during 24 h storage, the As-S bond has reformed, replacing the Zn²⁺ that was bound to the peptide, even at −20°C. Taken together, the data demonstrate that Zn²⁺ mediated As³⁺ displacement from ZRANB2 zfm apo-peptides was transient in nature.

Figure 5.

As³⁺ displacement from ZRANB2 apo-peptides by Zn²⁺ is transient. A, Zn²⁺ binding to ZRANB2 apo-peptides is not affected by storage at −20°C for 24 h (colorimetric quantification of free Zn²⁺). B, Zn²⁺ can displace As³⁺ from As³⁺-bound ZRANB2 apo-peptides when readings are taken immediately following incubation (ZNF1 is represented by open circles, ZNF2 by open squares). Please note that these two graphs (ZNF1-Immediate and ZNF2-Immediate) represent the same dataset as in Figure 2C; added for clarity). However, most of that peptide-bound Zn²⁺ is lost if the same samples are stored for 24 h (legend: overnight) at −20°C before the readings are taken (ZNF1 is represented by closed circles, ZNF2 by closed squares; colorimetric quantification of free Zn²⁺). C, UV-Vis spectrophotometric evidence that As³⁺ displaces Zn²⁺ from the ZNF1 apo-peptide if the samples are stored for 24 h (legend: overnight) at −20°C before the reading is taken (legend: overnight), but not if the readings are taken immediately following incubation (legend: immediate). When the readings are taken immediately, addition of Zn²⁺ to the As³⁺-bound apo-peptide causes a significant decrease in absorbance at 285 nm (absorption maximum for ZNF1). However, if the readings are taken after the same samples have been stored for 24 h at −20°C, there is no difference in absorbance of the As³⁺ bound peptides with and without Zn²⁺ addition. ***p<0.001 by Tukey’s multiple comparisons post-hoc test following one-way ANOVA. D, UV-Vis spectrophotometric evidence that As³⁺ displaces Zn²⁺ from the ZNF2 apo-peptide if the samples are stored for 24 h (legend: overnight) at −20°C before reading is taken (legend: overnight), but not if the readings are taken immediately following incubation (legend: immediate). When the readings are taken immediately, addition of Zn²⁺ to the As³⁺-bound apo-peptide causes a significant decrease in absorbance at 285 nm (absorption maximum for ZNF2). However, if the readings are taken after the same samples have been stored for 24 h at −20°C, there is no difference in absorbance of the As³⁺ bound peptides with and without Zn²⁺ addition. ***p<0.001 by Tukey’s multiple comparisons post-hoc test following one-way ANOVA.

Acute As³⁺ Treatment Induced ZRANB2 Protein Expression and Altered Splicing Profile in ZRANB2 Target.

Given that ZRANB2 has two C4 zfms, we hypothesized that binding of As³⁺ to these sites will lead to a decreased function of ZRANB2, leading to an increase in its expression at post-exposure time points, to compensate for the reduced activity. In order to determine a suitable acute As³⁺ treatment dosage, we carried out cell viability assay by alamar blue. The data demonstrated that 5 μM was the highest dosage that did not induced any toxicity at 72 h time point (Supplementary Figure 5) and hence was chosen for these proof of principle experiments. RT-qPCR data showed that the steady state levels of ZRANB2 mRNA between control and As³⁺ exposed cells were similar except at the 48 h time point (Figure 6A, Table 3). Immunoblot experiments, however, clearly demonstrate a significant increase in the expression of ZRANB2 protein starting at 3 h post-exposure and continuing until 24 h following which, it returns to basal level (Fig 6B- C, Table 3).

Figure 6.

Time course of ZRANB2 expression and splicing function in HaCaT cells exposed to 5 μM As³⁺ for 0 – 72 h. A, Quantification of ZRANB2 mRNA n HaCaT cells by RT-qPCR at different time points. The data are represented as fold change compared to unexposed control at each time point. **p<0.01 by Bonferroni post-hoc test following two-way ANOVA. B, ZRANB2 immunoblot from HaCaT cell lysates at different times. C, Densitometric analysis of ZRANB2 protein expression in B. Data are represented as % unexposed control expression at each time point. ***p<0.001, *p<0.05 by Bonferroni post-hoc test following two-way ANOVA. For two-way ANOVA analyses results, please see Table 4. D, Schematic representation of all the TRA2B exons in the region bound by the primers (represented by closed triangles) and the composition of ß1 and ß3 splice isoforms. Please note that the size of the exons/introns is not to scale. E, RT-PCR analysis of ß1 and ß3 splice isoforms at different time points. ß1 is the predominant isoform represented by the 502 bp amplification product, while the ZRANB2-dependent ß3 isoform is represented by the 368 bp product. F, Densitometric analysis of % ZRANB2 dependent ß3 isoform at different times. **p<0.01, *p<0.05 by Bonferroni post-hoc test following two-way ANOVA. For two-way ANOVA analyses results, please see Table 3.

Table 3:

Two-way ANOVA p values for ZRANB2 expression and splice function upon acute As³⁺ exposure

Parameter	ZRANB2 mRNA Expression	ZRANB2 Protein Expression	ZRANB2 Splice Function
PAs	0.08	<0.0001	<0.001
pTime	0.03	<0.0001	<0.001
pInteraction	0.03	<0.0001	0.03

Zn²⁺ binding is critical for maintaining both properly folded tertiary structure as well as biological function of zfm proteins ^31,46. Having demonstrated that As³⁺ can bind to each ZRANB2 zfm by displacing Zn²⁺ under cell free conditions, we wanted to examine if arsenic exposure could modulate the function of ZRANB2 protein under cellular conditions. For this purpose, we looked at the proficiency of ZRANB2 to alternatively splice its well characterized target TRA2B mRNA (Figure 6D). ZRANB2 has been shown to be responsible for the exclusion of exon 3 of TRA2B leading to the generation of β3 isoform (consisting of exons 1 and 4) at the expense of β1 isoform (consisting of exons 1,3 & 4). RT-PCR data unequivocally demonstrate significant reduction of ZRANB2-dependent β3 isoform at 3 h post exposure and continuing until 24 h compared to unexposed control (Figure 6D-E, Table 3). This result indicates that ZRANB2 splicing function is compromised upon acute arsenic exposure.

Environmentally Relevant As³⁺Treatment Induced ZRANB2 Protein Expression and Altered Splicing Profile in ZRANB2 Target.

Having demonstrated that acute As³⁺ exposure at therapeutic levels of 5 μM alters both the expression as well as splicing function of ZRANB2, we wanted to examine if lower (and environmentally/toxicologically relevant) exposure to As³⁺ is capable of eliciting similar effects. The data demonstrate that all the As³⁺ exposures used in this study (0.1, 1 and 5 μM) significantly induced ZRANB2 protein expression at 6 h (Figure 7A-B), although, there was no significant change in the level of expression within the different exposures. TRA2B splicing data shows a clear exposure response, with the percentage of ZRANB2 dependent β3 isoform significantly reduced at each successive increasing As³⁺ exposure (Figure 7C-D).

Figure 7.

ZRANB2 expression and splicing function As³⁺ concentration-response. A, ZRANB2 immunoblot from lysates of HaCaT cells exposed to 0-5 μM As³⁺ for 6 h. B, Densitometric analysis of ZRANB2 protein expression in panel A. Data are represented as % unexposed control expression at each time point. **p<0.01, *p<0.05 by Tukey’s multiple comparisons post-hoc test following one-way ANOVA. C, RT-PCR analysis of ß1 and ß3 splice isoforms in RNA isolated from HaCaT cells exposed to 0-5 μM As³⁺ for 6 h. ß1 is the predominant isoform represented by the 502 bp amplification product, while the ZRANB2-dependent ß3 isoform is represented by the 368 bp product. D, Densitometric analysis of ZRANB2 mRNA isoform RT-qPCR in panel C expressed as % ZRANB2 dependent ß3 isoform. ***p<0.001, **p<0.01, *p<0.05 by Tukey’s multiple comparisons post-hoc test following one-way ANOVA.

DISCUSSION

Molecular etiology of arsenic-induced skin cancer is complex and probably consists of multiple cellular mechanisms taking place simultaneously as well as longitudinally. This complexity makes it difficult to gain a clear insight into the global mechanisms that might be operative in the process. The absence of a suitable in vivo model system further hampers understanding. It has been impossible to induce skin cancer in any animal model with arsenic exposure alone, irrespective of the dosage or time of exposure used (reviewed in ⁵⁵). While several different mechanisms have been suggested for arsenic carcinogenicity, significant gaps still exist in our understanding of how these mechanisms act together to bring about cancer. Furthermore, “interaction with protein sulfhydryls” has long been suggested as a possible mode of arsenic carcinogenicity ^56-58; however, an understanding of how this interaction could lead to cancerous outcomes was lacking until recently^{31, 46, 59}.

Zfm proteins are widely distributed in prokaryotes and eukaryotes, including 3% of all human gene products ⁶⁰. Zfm proteins include a diverse group of members capable of interacting with DNA, RNA and other proteins ⁶¹. They can carry out a spectacular range of biological operations including, but not limited to DNA repair, signal transduction, splicing and transcriptional regulation ⁶¹, and are widely implicated in cancers ^{62, 63}. Zfms, which are central to the functional flexibility of this class of proteins, are targets for arsenic toxicity. As³⁺ can bind to C3H1 and C4 zfms, ^{31, 46} both in vitro and in vivo. Such binding results in the abrogation of the structure as well as function of the concerned protein ^{33, 44, 64}. An overwhelming majority of the studies on As³⁺-mediated disruption of zfm proteins pertains to DNA repair proteins ³⁴. This is surprising considering that zfm proteins play key roles in almost every major cellular process ⁶¹.

Here, we demonstrate for the first time that ZRANB2, a key protein involved in alternative splicing, is targeted by As³⁺. Biophysical data from our experiments unequivocally demonstrate that As³⁺ binds to each ZRANB2 zfm stably. We showed As³⁺ binding directly through mass spectrometry-based mass shift signatures. Furthermore, mass-spectrometry data are corroborated by a dose-dependent increase in As-S bond formation as evidenced by UV-Vis studies. This is the first report of arsenic binding to a zfm protein involved in alternative splicing, although, it is not surprising, given the propensity of arsenic to interact with other zfm proteins ^{30, 31, 34, 46, 65}. Interestingly, the data strongly suggest that the characteristics of As³⁺ binding and As³⁺ mediated Zn²⁺ displacement for two similar C4 zfms of the same protein vary considerably. Such differences could mean that As³⁺ could be modulating each of the zfms differently, resulting in differential structural and functional outcomes in the physiological milieu. While the Kd values for As³⁺ binding is quite dissimilar for the two zfms, we would like to point out that UV-Vis spectrophotometry is not the most sensitive technique when it comes to quantitative characterization of binding affinity, especially given the small differences in absorbance between the different As³⁺ concentrations used. This outcome could be a consequence of the different amino acid residues surrounding the two zfms (Table 1). While structures of each of the zfms of ZRANB2 have been resolved ^26,37, it still remains to be seen how the binding of As³⁺ affects the tertiary structures of these motifs. Clarification of the structures of both the zfms in the As³⁺ bound state could potentially shed light on why there are differences in As³⁺-zfm interaction parameters, and what it means in terms of RNA binding/splicing capabilities.

In addition, the data indicate that As³⁺ is capable of displacing Zn²⁺ from the Zn²⁺-bound apo-peptides. This effect is particularly important, as ZRANB2 zfms are Zn²⁺-bound under physiological conditions. This Zn²⁺ binding is of utmost importance in maintaining the properly folded state. This folding, in turn is critical to bind RNA molecules and mediate splicing functions ^26,37,38. Previous studies, using in vitro assay systems, showed that such As³⁺ mediated Zn²⁺ displacement from zfm DNA repair proteins leads to altered DNA binding capacity ^32,65. It is thus reasonable to hypothesize that similar Zn²⁺ displacement in ZRANB2 would lead to structural changes and sub-optimal/faulty RNA binding, adversely affecting its splicing capabilities, or even, altering its usual splicing profile.

Interestingly, we demonstrate that Zn²⁺ is also capable of displacing As³⁺ from ZRANB2 zfm peptides using a combination of biophysical techniques. One study showed that 5 μM Zn²⁺ treatment can restore the Zn²⁺ content of XPA and PARP-1 to basal levels or higher in HEKn cells treated with 2 μM As^{3+ 64}. However, the study did not examine if comparable changes in Zn²⁺ content of the proteins could occur in Zn²⁺ treated cells in the absence of As³⁺ treatment. The study thus does not refute the possibility that As³⁺ did not bind all the possible Zn²⁺ binding sites in the first place. It is possible that increased Zn²⁺ content upon Zn²⁺ treatment is an outcome of increased occupancy of the free Zn²⁺ binding sites, as opposed to occupancy of As³⁺-bound Zn²⁺ binding sites. Our data clearly demonstrate a dose-dependent reduction in As-S bond formation upon incubation of As³⁺-bound apo-peptide with increasing concentrations of Zn²⁺. This evidence is in agreement with the only previous study available to our knowledge that has studied Zn²⁺ mediated As³⁺ displacement using similar As-S bond formation directly ⁶⁴.

Maintenance of properly folded tertiary structure is important for each of the two ZRANB2 zfms for RNA binding ^26,37. Intrinsic fluorescence data show that addition of Zn²⁺ to the native apo-peptide promotes characteristic folding for each of the two zfms. As³⁺ addition affects the folding of the two apo-peptides in opposing fashion, promoting folding for ZNF1, while promoting unfolding for ZNF2. This might partially explain the differences in As³⁺ binding parameters between them. It is interesting to note that As³⁺ addition has relatively smaller effects on the tertiary structure of each apo-peptide, compared to Zn²⁺ addition. Moreover, addition of As³⁺ to the Zn²⁺-bound apo-peptides does not bring about any significant changes in the tertiary structure, compared to the Zn²⁺-bound folded conformation (Figure 4). This result can be explained by the fact that Zn²⁺ coordinates with all 4 cysteine residues, and the resultant Zn²⁺-bound motif is expected to have a rigid structure. As³⁺ on the other hand, binds to only three of the four cysteine residues. It is possible that As³⁺ binds alternately to any three of the four cysteines available, making the conformation less rigid than the Zn²⁺-bound structure. However, from the two-way ANOVA data (Table 2 and Supplementary Figure 4), it is evident that ZNF1 and ZNF2 interact slightly differently when both As³⁺ and Zn²⁺ are present in the system. This result indicates that the two zfms could be modulated in somewhat different manner, consistent with the data from UV-Vis spectrophotometric experiments.

Furthermore, we elucidate that Zn²⁺ mediated As³⁺ displacement is transient in nature. This result is a novel finding of the current study. The same reaction mixture that showed Zn²⁺ mediated As³⁺ displacement lost most of the bound Zn²⁺ when stored for 24 h at −20 °C. The observation that Zn²⁺ binding to apo-peptides in the absence of As³⁺ is not altered by storage conditions (Figure 5A) argues against the possibility that the Zn²⁺ loss seen in the presence of As³⁺ (Figure 5B) is the result of Zn²⁺ chelation by TCEP, a weak chelator of Zn^{2+ 71}, and a component of the buffer used in these studies. However, it is still possible that weak chelation of Zn²⁺ may be sufficient to shift the equilibrium in favor of As³⁺ binding when both are present for an extended period of time. These results also provide one possible explanation as to how we could be seeing the mass signature of As³⁺-bound ZRANB2 zfm apo-peptides upon Zn²⁺ mediated As³⁺ displacement.

An important question to address was if As³⁺ mediated Zn²⁺ displacement from ZRANB2 results in any physiologically relevant outcome, given most of our metal interaction studies were performed with synthetic apo-peptides in cell-free systems. We chose to treat HaCaT cells initially with 5 μM of As³⁺ for up to 72 h, as this was the highest exposure that did not induce any significant cell mortality in our cell line at the longest time point (Supplementary Figure 5). While the concentration (5 μM) and the time of exposure (up to 72 h) do not emulate a chronic exposure scenario, it does correspond to the plasma As³⁺ level in subjects following a therapeutic regimen ⁷⁴. We also chose this treatment regimen as a cost and time effective proof of concept experiment to investigate if it will be worthwhile to pursue future experiments using long term, low dose exposures that accurately simulate in vivo chronic exposures. We show that acute exposure of HaCaT cells to As³⁺ induces the expression of ZRANB2 protein between 3-24 h post-exposure, although, there is no increase in the steady state mRNA levels of ZRANB2 at these (or earlier) time points. This result could mean that the increase in ZRANB2 expression is perhaps due to stabilization of the ZRANB2 protein, rather than an increase in its steady state mRNA levels. As³⁺ exposure has previously been demonstrated to inhibit protein degradation by inhibiting autophagosome-lysosome fusion ⁷⁵. Furthermore, the induction of ZRANB2 protein expression decreased by 48 h. This transient protein induction could signify that most of the arsenic has been metabolized by methylation, glutathionylation, thiolation and possibly effluxed from the cell by 48 h ⁷³, and is consequently, no longer available to bind to ZRANB2. Interestingly, we do see a significant induction in ZRANB2 steady state mRNA levels in the As³⁺ exposed cells at 48 h post-exposure, when the ZRANB2 protein expression is down to basal level. We are not entirely sure what causes this spike. Furthermore, we demonstrate that acute As³⁺ exposure in the HaCaT cells results in significant alteration of the functional profile of ZRANB2 leading to aberrant splicing of its downstream target TRA2B. ZRANB2 is known to promote the exclusion of exon 3 of TRA2B, resulting in higher expression of the β3 isoform ⁴⁰. In our experimental system, the expression of ZRANB2 dependent β3 isoform is significantly suppressed between 3-24 h post-exposure. Interestingly, the ZRANB2 expression is significantly induced at exactly those time points. We speculate that As³⁺ mediated Zn²⁺ displacement from ZRANB2 is leading to reduced splicing capability, thereby activating feedback systems to express more and more ZRANB2 to compensate for the observed functional impairment (Figure 6). The cycle of altered function and heightened expression continues possibly till most of the free As³⁺ capable of binding to ZRANB2 has been removed from the system, possibly by a combination of metabolism, binding and efflux.

Having demonstrated that acute As³⁺ exposure can alter both expression and function of ZRANB2, we asked if such effects can also occur at much lower environmentally and toxicologically relevant exposures. Our data demonstrate this is true (Figure 7). While both ZRANB2 protein expression and TRA2B splicing function were altered at each As³⁺ concentration tested, it is noteworthy, that the lowest concentration used (0.1 μM) corresponds to the blood serum level of As³⁺ in chronically exposed human populations ⁷⁶. These data suggest that chronic As³⁺ exposure can alter the splicing profile in chronically exposed populations. This result is particularly interesting, given that the ß1 isoform corresponds to the canonical TRAB2B-205 isoform (288 amino acids) of this gene, while the ß3 corresponds to the TRA2B-203 isoform (188 amino acids) and lacks the first 100 amino acid residues. Although, the functions of the β3 isoform have not been studied, it is interesting to note that this isoform misses a big portion of the RA1 domain which resides within the residues 31-113. Thus, it is fair to speculate that β3 and β1 isoforms might have different RNA binding capacity and possibly different splice targets. As³⁺ exposure at exposures relevant to both chronic low dose exposures and acute therapeutic doses seems to favor an increase of canonical TRA2B-203 isoform.

TRA2B mRNA is one of the most well-known splice targets of ZRANB2 ⁴⁰. TRA2B mRNA codes for the protein Tra2β which is closely related to the SR family of splice regulator proteins and is a central player in splicing regulation itself. Tra2β is known to be upregulated in several cancers ⁴². Furthermore, several Tra2β target exons have been experimentally demonstrated to have important pro-oncogenic roles, prime examples being CD44, HIPK3 and NASP ⁴². TRA2B knockdown leads to widespread altered splicing events in the genome, including many novel and unannotated isoforms ⁷⁷. It is possible that by altering the expression and function of ZRANB2, As³⁺ exposure could give rise to genome-wide aberrant alternative splicing and set cells on the path of carcinogenesis via altering the splice patterns of a few target genes like TRA2B.

Although the present work examines the interaction of one alternative splicing zfm protein with one metalloid, these results open up new lines of research in several directions. Several metal containing chemotherapeutic agents (platinum, gold, ruthenium, cobalt, antimony and selenium), as well as toxic and/or carcinogenic metals/metalloids (arsenic, cadmium, chromium, cobalt, nickel, lead) are capable of binding to and displacing Zn²⁺ from zfm proteins⁸⁵. Surprisingly, little or no work has been done on zinc displacement from RNA binding proteins including splice regulator proteins, which form a major class of zfm proteins. We speculate that differential alternative splicing as a result of structural and functional abrogation of splice regulating zfm proteins by heavy metals/toxicants/pollutants/therapeutic molecules could be playing an important role both in health and disease.