The human AP-endonuclease 1 (APE1) is a DNA G-quadruplex structure binding protein and regulates KRAS expression in pancreatic ductal adenocarcinoma cells

Abstract Pancreatic ductal adenocarcinoma (PDAC), one of the most aggressive types of cancer, is characterized by aberrant activity of oncogenic KRAS. A nuclease-hypersensitive GC-rich region in KRAS promoter can fold into a four-stranded DNA secondary structure called G-quadruplex (G4), known to regulate KRAS expression. However, the factors that regulate stable G4 formation in the genome and KRAS expression in PDAC are largely unknown. Here, we show that APE1 (apurinic/apyrimidinic endonuclease 1), a multifunctional DNA repair enzyme, is a G4-binding protein, and loss of APE1 abrogates the formation of stable G4 structures in cells. Recombinant APE1 binds to KRAS promoter G4 structure with high affinity and promotes G4 folding in vitro. Knockdown of APE1 reduces MAZ transcription factor loading onto the KRAS promoter, thus reducing KRAS expression in PDAC cells. Moreover, downregulation of APE1 sensitizes PDAC cells to chemotherapeutic drugs in vitro and in vivo. We also demonstrate that PDAC patients’ tissue samples have elevated levels of both APE1 and G4 DNA. Our findings unravel a critical role of APE1 in regulating stable G4 formation and KRAS expression in PDAC and highlight G4 structures as genomic features with potential application as a novel prognostic marker and therapeutic target in PDAC.


INTRODUCTION
Pancreatic ductal adenocarcinoma (PDAC) is the third leading cause of cancer-related deaths in the United States, with a 5-year survival rate of ∼9% (1). Being one of the most aggressive and lethal solid malignancies, PDAC is highly resistant to chemotherapy and radiation and remains a major clinical challenge (2,3). The proto-oncogene, KRAS, is the most frequently mutated oncogene in PDAC (∼95% of cases) (4,5). Activating point mutations at glycine-12, glycine-13 or glutamine-61 impair the intrinsic GTPase activity of KRAS protein and lead to its constitutive activation and persistent stimulation of the downstream signaling pathways that drive oncogenesis (6)(7)(8)(9)(10). There is growing evidence that the maintenance of the malignant phenotype depends on the constitutive expression of KRAS, a phenomenon termed 'oncogene addiction' (7,(11)(12)(13)(14). Consequently, suppression of mutated KRAS impacts the viability of pancreatic cancer cells. Thus, the absolute importance of mutant KRAS expression in PDAC has driven multiple investigations to target it both at the transcriptional and translational levels (5,15,16).
Several studies have demonstrated that the promoter region of many cancer-driving oncogenes, including KRAS and c-MYC, have G-rich DNA sequences that form a noncanonical, four-stranded nucleic acid structure called Gquadruplex (G4) (17,18). The G4 structures are highly abundant in the gene regulatory elements such as gene promoters, 5 and 3 untranslated regions, replication initiation sites and telomeres (19,20). They have been implicated in modulation of gene expression including that of oncogenes, alteration in chromatin states, and in the maintenance of telomere and genome stability (21). Recent genome-wide analysis revealed that G4 DNA structures are hubs for transcription factor (TF) binding (22). Because of their important regulatory functions pertaining to oncogene expression, G4 structure has emerged as a potential therapeutic target in cancer (23)(24)(25)(26).
The core promoter region of the human KRAS, encompassed within the region from +50 to −510 bp with respect to the transcription start site (TSS), is highly G/C rich (75%) and can form three G4 structures (27)(28)(29). The proximal G4 that lies between −148 and −116 bp upstream of TSS overlaps with a nuclease-hypersensitive element and is recognized by several nuclear proteins. It was demonstrated that binding of TFs such as MAZ, hnRNP A1 and PARP-1 in this region activates KRAS expression (17,(30)(31)(32)(33). Furthermore, oxidative stress that induced guanine (G) oxidation in the regulatory G4 motif in KRAS was shown to upregulate KRAS expression (34). Despite these wellestablished roles of G4 in KRAS transcriptional regulation, little is known about the molecular machinery that regulates the formation and stability of these G4 structures in cells.
Recently, we demonstrated a genome-wide correlation between the occupancy of apurinic/apyrimidinic endonuclease 1 (APE1) and G4 structures in the cells (35). Originally discovered as a DNA repair enzyme, human APE1 plays a central role in the repair of endogenous oxidative and alkylating DNA damage through the well-conserved base excision repair (BER) pathway (36,37). Later, APE1 was independently identified as a multifunctional protein involved not only in DNA damage repair but also in regulating gene expression by its redox activity and is hence referred to as reduction-oxidation factor 1 (Ref-1) (38). APE1 reduces redox-sensitive cysteine residues in many oxidized TFs, including AP-1 (39), nuclear factor kappa B (40) and p53 (41), and increases their DNA binding by maintaining them in a reduced active state. Studies by us and others have also demonstrated that APE1 can act as a direct transcriptional co-activator or co-repressor of several different genes involved in cell growth, proliferation and chemotherapeutic drug resistance (42)(43)(44)(45)(46). APE1 is overexpressed in a variety of cancers, including pancreatic (47), prostate (48), cervical (49), gliomas (50), ovarian (51,52), lung (53) and colon (54). This increased expression has been associated with increased tumor growth, cell migration and drug resistance, as well as patients' poor prognosis (50,51,55,56).
In this study, we demonstrate that APE1 is a G4-binding protein that plays a critical role in the formation of stable G4 structures in the genome and regulates KRAS expression in PDAC cells. Our study shows that recombinant APE1 binds to KRAS promoter G4 with high affinity in vitro. Loss of APE1 abrogates the formation of stable G4 structures and alters KRAS expression in PDAC cell lines. Knockdown (KD) of APE1 sensitizes PDAC cells to chemotherapy both in vitro and in vivo. Furthermore, we provide evidence that increased APE1 levels are associated more often with elevated levels of G4 DNA in patients with aggressive PDAC.

Immunofluorescence
Cells were cultured on coverslips (Fisherbrand, Cat # 12-542-B) and fixed in 4% paraformaldehyde (PFA; Sigma) in phosphate-buffered saline (PBS) for 15 min at room temperature (RT). Cells were subsequently permeabilized and blocked for 1 h at RT using a blocking buffer containing 0.5% Triton X-100 (Sigma), 10% goat serum (Thermo Fisher, Cat # 50062Z), glycine (Fisher BioReagents, Cat # BP381-1) and sodium azide in PBS. After permeabilization and blocking, cells were incubated in blocking buffer with primary antibodies overnight at 4 • C and subsequently the corresponding secondary antibodies for 1 h at RT. Cells were then washed in PBS and mounted using mounting media with DAPI (Vector Laboratories, item # VV-93952-27). For G4 staining, an additional permeabilization step was performed after PFA fixation by incubating cells in PBS containing 0.5% Tween 20 (Fisher BioReagents, Cat # BP337-100) for 20 min at 37 • C. Following Tween 20 permeabilization, cells were treated with 0.04 g/l RNase A and subsequently blocked and processed as described ear-lier. For the DNase experiments, cells were fixed in 4% PFA in PBS for 30 min, permeabilized with 0.2% Triton X-100 in PBS for 1 min and washed in PBS at RT. Cells were then incubated for 2 h at 37 • C in DNase reaction buffer with or without 0.06 U/l of DNase I (RQ1 DNase, Promega). The working DNase reaction buffer contained 40 mM Tris-HCl (pH 8), 5 mM CaCl 2 , 2 mM MgCl 2 and 100 g/ml bovine serum albumin. Following DNase digestion, the cells were washed in PBS and stained as described earlier. All immunofluorescence images were captured by confocal microscopy or super-resolution (∼100 nm) structured illumination microscopy (SIM), where indicated. Confocal microscope images were captured using the Zeiss LSM 800 with Airyscan microscope in the Advanced Microscopy Core Facility (AMCF) at the University of Nebraska Medical Center (UNMC). Three-dimensional SIM images were collected with the Zeiss ELYRA PS.1 super-resolution microscope in the AMCF at UNMC. Pearson's correlation coefficients were calculated using the ImageJ software JACoP colocalization analysis module. Quantification of colocalization was determined by establishing a threshold using the JACoP threshold optimizer followed by calculation of correlation coefficients.

Proximity ligation assay
Cells were cultured on coverslips and fixed in 4% PFA in PBS for 15 min at RT. An additional permeabilization step was performed after PFA fixation by incubating the cells in PBS containing 0.5% Tween 20 for 20 min at 37 • C. Following Tween 20 permeabilization, cells were treated with 0.04 g/l RNase A and subsequently blocked using the Duolink blocking solution. The PLA was performed following the manufacturer's protocol for Duolink In Situ Red Starter Kit Mouse/Rabbit (Sigma-Aldrich, # DUO92101). The primary antibodies used for the PLA studies were rabbit polyclonal anti-APE1 (1:100) and anti-G4 clone 1H6 (1:50). Confocal images were captured using the Zeiss LSM 800 with Airyscan microscope in the AMCF at UNMC. For quantification, the number of foci per cell was counted for 50 cells under each experimental condition using the ImageJ software.

Western blot
Whole cell extracts were prepared, and western blot was performed as we have described in our previous study (59).

Fluorescence polarization assay
The fluorescence polarization (FP) measurements were obtained in a SpectraMax M5 Multimode Microplate Reader using SoftMax Pro software. 6-FAM-labeled G4-forming KRAS oligonucleotide (oligo) at a concentration of 10 nM was incubated in a buffer containing 50 mM Tris-HCl (pH 7.5), 1 mM MgCl 2 , 1 mM DTT, 0.1 mM EDTA and 50 mM KCl alone or with increasing doses of purified recombinant proteins (1.25-10 240 nM), G4 ligands, 1H6 antibody or unlabeled oligo in triplicates for 30 min at RT in the dark in a black opaque Perkin Elmer 384-well plate. The total volume of the samples in each well was 25 l, and each experiment with triplicate samples was repeated three times. The fluorescence (RFUs), FP and anisotropy values were recorded Nucleic Acids Research, 2022, Vol. 50, No. 6 3397 at 25 • C with the excitation filter set at 493 nm and emission filter set at 525 nm with a cutoff of 515 nm. The FP values expressed in the units of millipolarization versus the concentration of the proteins, ligands, antibodies or oligo were plotted using the nonlinear regression (curve fit) option in Prism 8.0, and dissociation constants (K d ) and EC 50 values were obtained.

Tryptophan fluorescence quenching
The fluorescence measurements were obtained using a Varian Cary Eclipse Fluorescence Spectrophotometer (Walnut Creek, CA). A working concentration (500 nM) of WT APE1 protein was prepared in a buffer containing 50 mM Tris-HCl (pH 7.5), 1 mM MgCl 2 , 1 mM DTT, 0.1 mM EDTA and 50 mM KCl. The fluorescence emission spectrum was obtained for the free protein from 300 to 450 nm, using an excitation maximum of 280 nm, with the emission and excitation slits set at 5 nm. This protein solution was titrated with increasing doses of G4-forming KRAS oligo (0 nM to 1 M) and for each concentration of the oligo, the emission spectrum was recorded. The spectra are reported as fluorescence intensity [arbitrary unit (a.u.)] versus wavelength (nm).

Circular dichroism spectroscopy
Circular dichroism (CD) spectra were obtained on a JASCO J-810 Spectropolarimeter, equipped with a Peltier thermoelectric type temperature control system and flow-through HPLC cell. The instrument was controlled by Jasco's Spectra Manager™ software. The experiments were carried out with 1 M oligonucleotide solutions in 50 mM Tris-HCl (pH 7.5), 2 mM MgCl 2 and 1 mM DTT alone or in combination with 1 M of the recombinant proteins. The spectra were recorded in a 10 mm quartz cuvette at 20 • C. Scans were performed over a wavelength range of 210-330 nm with a response time of 0.5 s, 1 nm pitch and 1 nm bandwidth. Blank spectra of samples containing buffer were subtracted from that of DNA samples. Each spectrum was recorded five times, smoothed and subtracted to the baseline. The spectra are reported as ellipticity (measured in units of millidegrees) versus wavelength (nm).

UV-Vis absorption titration experiment
Absorption spectra were measured on a thermoelectrically controlled Aviv 14 DS UV/Vis Spectrophotometer (Lakewood, NJ) in a 1 cm path length quartz cuvette. The reference solution contained 10 mM Tris-HCl (pH 7.5), 1 mM EDTA and 50 mM KCl. UV-Vis absorption titrations were carried out by the stepwise addition of unlabeled G4forming KRAS oligo (0 nM to 1.8 M) solution to the quartz cuvette containing 4 M solution of TMPyP4 in the above-mentioned buffer. Absorption spectra were recorded in the 350-500 nm wavelength range at RT. The titration was terminated when the wavelength and intensity of the absorption band for TMPyP4 did not change anymore upon three successive additions of the G4 oligo-containing solution. The spectra are reported as absorbance (a.u.) versus wavelength (nm).

Chromatin immunoprecipitation
Cells were plated in 150 mm cell culture dishes. The following day, cells were incubated with 1% formaldehyde in PBS for 15 min at RT to cross-link protein-DNA complexes and subsequently washed with PBS. Cells were collected in PBS containing 1× protease inhibitor cocktail (Roche cOm-plete™) with a plastic cell scraper and pelleted at 1000 rpm for 10 min at 4 • C. The pelleted cells were lysed and incubated on ice in SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl, pH 8) for 10 min and subsequently subjected to sonication (Misonix Sonicator 3000). Sonication was performed on ice with eight rounds of 15 s pulses with 15 s pause between each pulse. The sonicated lysate was centrifuged at 14 000 rpm for 15 min at 4 • C, and the supernatant containing the clear sheared chromatin lysate was collected. The collected lysate was diluted 1:10 to a total volume of 2 ml with ChIP dilution buffer containing 0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl (pH 8) and 167 mM NaCl. Immunoprecipitation was performed by incubating 5 g of the respective antibody with the diluted lysate overnight at 4 • C with constant nutating rocking. The next day, 25 l of Protein A/G Dyn-abeads™ (Invitrogen) was added to each reaction and incubated for 2 h at 4 • C with constant nutating rocking. The immunoprecipitates were washed sequentially with low-salt immune complex wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8, 150 mM NaCl), high-salt immune complex wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8, 500 mM NaCl), LiCl immune complex wash buffer (0.25 M LiCl, 1% NP-40, 1% sodium deoxycholate, 1 mM EDTA, 10 mM Tris-HCl, pH 8) and TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8). The protein-DNA complexes were eluted in ChIP elution buffer (1% SDS, 0.1 M NaHCO 3 ) and decross-linked in 200 mM NaCl overnight at 65 • C. ChIP DNA was purified by RNase treatment, proteinase K digestion, phenol-chloroform extraction and precipitation by 100% ethanol using a standard protocol. The ChIP-purified DNA was finally dissolved in ultrapure DNase/RNase-free water and was used for subsequent assays.

Colony formation assay
Five hundred cells were seeded in each well of six-well plates and incubated at ambient conditions in appropriate medium for 24 h to allow for their attachment. Following this incubation, cells were either left untreated (control group) or treated with increasing doses of oxaliplatin (0.5-8 M), 5-FU (2-10 M) or gemcitabine (0.005-0.5 M) for 24 h. After 24 h, the media containing drugs were discarded and fresh media were added, and the cells were allowed to grow for 12 days under ambient cell culture conditions. Following this, the media were removed and the colonies in the wells were stained with a 0.5% crystal violet solution (made in 25% methanol and stored at RT) for 10 min. Subsequently, the plates were carefully rinsed in double-distilled water. The plates were allowed to dry, and the number of colonies was counted and plotted as percent cell viability against the drug concentrations. For each condition, three independent experiments were done in triplicates.

Xenograft studies
All animal experiments were performed with the approval of the UNMC Institutional Animal Care and Use Committee. The experiments and reports adhered to the Animal Research: Reporting of In Vivo Experiments guidelines. MIA PaCa-2 APE1shRNA cells (2 × 10 6 in 100 l medium with Matrigel) were injected subcutaneously over the left and the right flanks in 6-week-old male athymic nude mice (Charles River). The average weight was 27 ± 3.6 g. Dox (2 mg/ml) was given in the drinking water with 1% sucrose to the Dox-treated group. All the water bottles were changed every 2 days. Subcutaneous tumors were allowed to grow for 1-2 weeks before treatments. The mice were divided into six treatment groups (n = 5 in each group) and received treatments twice per week for 4 weeks. The drugs 5-FU 30 mg/kg and oxaliplatin 2 mg/kg were injected intraperitoneally. Normal saline (100 l) was given to the vehicle control group. Body weights and tumor volumes were measured before each treatment. The mice were euthanized in a gas canister with gradual fill carbon dioxide at the end of the treatment cycles. Xenograft tumors were fixed in formalin, and paraffin-embedded tissue sections were used to perform immunohistochemical (IHC) staining with anti-APE1 antibody (1:600).

Patient tissue sample studies and immunohistochemical analysis
PDAC patient tissue microarrays (TMAs) were obtained from the Rapid Autopsy Program for Pancreas at UNMC. Tissues were collected in accordance with the institution's review board approval and informed consent waiver. The deparaffinized sections were stained as per standard IHC protocol. The antibodies used were anti-APE1 (1:600) and anti-G4 clone 1H6 (1:200). IHC staining scores for APE1 and G4 were calculated by multiplying scores for extent of staining (percentage of stained cells) and intensity of staining. The criteria used to score the staining of the tissue samples were based on the percentage of cells stained: 0% stained cells, score 0; 1-10% stained cells, score 1; 11-50% stained cells, score 2; 51-80% stained cells, score 3; and 81-100% stained cells, score 4. The criteria used to score the tissue sample staining based on intensity were as follows: no staining, score 0; weak staining intensity, score 1; moderate staining intensity, score 2; and high staining intensity, score 3. The information regarding the cancer stage, response to treatment and survival days for each patient was obtained from the Rapid Autopsy Program for Pancreas database.

APE1 plays a crucial role in the formation of stable G4 structures in PDAC cell lines
We examined the formation of G4 structures in PDAC cell lines MIA PaCa-2 and PANC-1 using the G4 structurespecific antibody, clone 1H6. This antibody has been extensively used for analyzing the G4 structures in cells in several earlier studies (60,61). Moreover, the FP assay revealed that 1H6 antibody has high affinity for binding Figure S1A). Using this antibody and SIM, we could detect G4 foci primarily in the nuclei of MIA PaCa-2 cells ( Figure 1A). This immunostaining of G4 DNA in cells was sensitive to DNase but not to RNase A treatment, confirming the specificity of the antibody toward the DNA G4 structures ( Supplementary Figure S1B). Using confocal microscopy, similar results for G4 staining were obtained in PANC-1 cell line (Supplementary Figure S1C). We observed colocalization between G4 and APE1 staining in these cells ( Figure 1A and Supplementary Figure S1C). Treatment with G4-stabilizing ligand PDS (62,63) increased the number of G4 foci and colocalization between G4 foci and APE1 staining ( Figure 1A-C). On the other hand, downregulation of APE1 by stable expression of APE1shRNA under a Dox-inducible promoter ( Figure 1D and Supplementary Figure S1D) led to a significant reduction in the number of detectable G4 foci in MIA PaCa-2 ( Figure 1E) and PANC-1 (Supplementary Figure  S1E) cell lines. Furthermore, treatment with H 2 O 2 and PDS showed no effect on the G4 staining in APE1 downregulated cells ( Figure 1E and Supplementary Figure S1E). In contrast, Dox treatment in cells stably expressing NTC-shRNA had no effect on the G4 staining ( Figure 1F). We also examined the effect of APE1 KD on G4 staining in the WT KRAS expressing BxPC-3 cell line. We observed an appreciable number of G4 foci and APE1 staining in BxPC-3 cells expressing control empty vector ( Figure 1H, upper panel). KD of APE1 in BxPC-3 cells using stable expression of APE1shRNA ( Figure 1G) abrogated the G4 staining in these cells ( Figure 1H, lower panel).
To eliminate the possibility that APE1 downregulation can affect the staining of other chromatin-bound proteins in general, we immunostained the APE1 downregulated MIA PaCa-2 and PANC-1 cells with histone H3K27Ac antibody. We observed no change in the staining of H3K27Ac under APE1 KD conditions, indicating that downregulation of APE1 does not affect staining of other chromatinbound protein, but it specifically reduced G4 staining in the PDAC cell lines (Figure 2A and Supplementary Figure S2A). Re-introduction of WT APE1 expressing plasmid but not empty vector plasmid in APE1 KD cells was able to rescue G4 foci formation, confirming the requirement of APE1 for stable G4 formation in the genome (Figure 2B and C). The association of APE1 with G4 structures in cells was also confirmed by PLA using control and APE1 KD cells ( Figure 2D). While a significant number of APE1:G4 PLA foci were seen in control cells, APE1 KD significantly reduced the number of PLA foci ( Figure 2D and E). As controls, we used only APE1 antibody, only G4 antibody and no primary antibody, all of which gave us almost no foci, emphasizing the specificity of the assay (Supplementary Figure S2B and Figure 2E). Together, our results suggest that APE1 colocalizes with G4 in PDAC cells and plays a critical role in the formation of G4 structures in the genome.

APE1 binds to KRAS promoter G4 structure with high affinity
The observations that APE1 colocalizes with G4 structures and is required for the formation of G4 structures in PDAC cells led us to test whether APE1 binds to G4 structures in cells. We first examined the ability of recombinant WT APE1 to bind directly to a preformed KRAS promoter G4 structure using in vitro biophysical assays. The KRAS promoter contains three G4 motifs, among which the proximal G4 motif (−148 and −116 bp), otherwise known as 32R, has been well characterized to form G4 in vitro and shown to be functionally involved in regulating KRAS expression in cells (30). We synthesized a 6-FAM-labeled 32-mer oligonucleotide corresponding to the KRAS G4-proximal motif (which we named as the G4-forming KRAS oligo) (Figure 3A), and induced G4 folding by incubating the oligo in the presence of 50 mM KCl. We confirmed the formation of a parallel G4 structure by a CD spectroscopy assay that demonstrated a strong positive ellipticity at around 265 nm and a weak negative ellipticity at around 240 nm, which are the established CD signatures for a parallel G4 structure ( Figure 3B). As a negative control, we used a KRAS oligo in which several G residues were replaced with T so that it cannot form G4 structure (non-G4-forming KRAS oligo) ( Figure 3A). We confirmed the inability of the non-G4-forming KRAS oligo to form parallel G4 structure by CD spectroscopy where the characteristic CD signature for parallel G4 structure was absent ( Figure 3B). The FP assay was performed using the G4-forming KRAS oligo and increasing doses of purified recombinant WT APE1 protein.
We found that the WT APE1 protein has a very strong affinity for binding to the KRAS G4 structure, with a dissociation constant (K d ) of 31 ± 3 nM, indicating that APE1 can bind with high affinity to a G4 structure without any AP site damage ( Figure 3C). In contrast, WT APE1 did not show any measurable binding affinity toward the non-G4-forming KRAS oligo that cannot form G4 structure, thus confirming that APE1 has a high affinity toward a G4 structure ( Figure 3D). A competition assay with unlabeled G4-forming KRAS oligo revealed an EC 50 value of 2.3 ± 1.5 M, further emphasizing the high affinity binding between APE1 and KRAS G4 structure ( Figure 3E). We also independently confirmed APE1's binding to the G4forming KRAS oligo by a tryptophan (Trp) fluorescence quenching assay. APE1 has seven Trp residues whose intrinsic fluorescence is highly sensitive to changes in the local environment upon addition of a binding partner (64). We determined the binding of WT APE1 to KRAS G4 structure by measuring the quenching of Trp fluorescence upon addition of increasing doses of preformed KRAS G4 to a solution of WT APE1 protein ( Figure 3F). We calculated the concentration of the bound protein (C b ), the concentration of the free protein (C f ) and r, which is the ratio of C b to the concentration of G4-forming KRAS oligo in the solution. A Scatchard plot revealed a K d of 12 nM, indicating a very strong binding affinity of APE1 to G4 ( Figure  3G). Taken together, data from both our FP assay and the Trp fluorescence quenching studies suggest that WT APE1 protein has very strong binding affinity to KRAS promoter G4 structure, with a dissociation constant in the nanomolar range.
To investigate which domain of APE1 is involved in its strong binding to KRAS promoter G4 structure, we performed FP assays with APE1 catalytically inactive mutant H309A, N-terminal 42-amino acid deletion mutant N 42 and preformed KRAS G4 structure. His309 residue, in the active site pocket of APE1, is necessary for the catalytic activity (AP-endonuclease) of APE1 (65). The H309A mutant showed ∼5-fold less affinity (K d = 152 ± 25 nM) for KRAS promoter G4 compared to WT APE1 protein ( Figure 3C). Previous studies have shown that the highly positively charged and unstructured N-terminal domain of APE1 participates in its interactions with DNA or RNA (66,67). To determine whether the N-terminus of APE1 is required for its strong binding to the KRAS G4 structure, we performed the FP assay with an N-terminal deletion mutant N 42. The FP assay revealed a K d of 313 ± 12 nM, which is ∼10-fold higher than that for WT APE1 protein, indicating that the N-terminus contributes to the binding with the KRAS G4 structure ( Figure 3C). To determine whether the N-terminus 1-42 amino acids of APE1 are sufficient to bind to the KRAS G4 structure, we purified GST-tagged APE1 1-42-amino acid peptide and examined its binding to the KRAS G4 structure using the FP assay. We found that this peptide by itself possesses no measurable binding affinity to the KRAS G4 structure (Supplementary Figure S3). These data together suggest that even though the positively charged N-terminal domain contributes to G4 binding in the context of the full-length APE1, the other domains of the protein also play an important role in its strong binding to the G4 structure in vitro.
Several earlier studies demonstrated that KRAS proximal G4 oligonucleotide is able to form a parallel G4 structure in the presence of 50 mM KCl (68). Consistent with this, we also observed formation of KRAS parallel G4 when the oligo was incubated in the presence of KCl in vitro ( Figure 3B). Interestingly, addition of recombinant WT APE1 to the G4-forming KRAS oligo resulted in an increase in positive ellipticity at around 265 nm, indicating that WT APE1 has the ability to bind and promote the folding/stacking of G4 structure ( Figure 3H). As a negative control, we used non-G4-forming KRAS oligo, to which addition of recombinant WT APE1 had no effect as indicated by no change in the CD spectra signature ( Figure 3H). We also performed CD spectroscopy with the N 42 mutant protein and observed almost no effect on the ellipticity at 265 nm with the G4-forming KRAS oligo indicating the absence of the stacking effect ( Figure 3H). This suggests that the positively charged N-terminal domain of APE1 may facilitate the stacking of the G-tetrads to promote G4 folding in vitro.

APE1 is associated with KRAS promoter G4 region in cells and modulates KRAS expression
The strong binding affinity of WT APE1 to KRAS G4 structure in vitro and their colocalization and interaction in cells led us to examine the occupancy of APE1 on the KRAS promoter G4 region. We tested for the co-occupancy of APE1 and G4 in the KRAS proximal promoter region by ChIP using APE1 and G4-specific antibody followed by a promoterdirected quantitative PCR (ChIP-qPCR) assay. Our results demonstrate significant enrichment of APE1 and G4 structures in the KRAS promoter G4 region as opposed to the control non-G4 sequence-containing region (Ctr-2) ( Figure  4A and B). To test whether APE1 is required for stable G4 formation on KRAS promoter in cells, we used control and  APE1 KD MIA PaCa-2 cells. Promoter-directed G4-ChIP-qPCR analysis revealed a significant reduction in G4 enrichment on KRAS promoter upon APE1 KD ( Figure 4B). Several earlier studies have shown that the recruitment of TFs MAZ and PARP1 to the KRAS promoter G4 region activates KRAS expression (31,69). Our promoter-directed ChIP-qPCR assay revealed that APE1 KD significantly reduced the enrichment of PARP1 and MAZ on the KRAS promoter G4 region compared to the control region ( Figure  4B). This suggests that APE1 plays a crucial role in PARP1 and MAZ TF loading onto the KRAS G4 promoter region.
Next, we examined the effect of APE1 KD on KRAS expression. Using qRT-PCR analysis, we demonstrated that downregulation of APE1 by stable expression of APE1shRNA in MIA PaCa-2, PANC-1 and BxPC-3 cell lines significantly decreased KRAS expression ( Figure 4C). We also examined the effect of G4-stabilizing ligands, PDS and TMPyP4, on KRAS expression in control and APE1 KD PDAC cell lines. First, we confirmed the robust binding of these G4 ligands to KRAS G4 by the FP assay (Supplementary Figure S4A and B). Since these ligands have an extended aromatic core that can change the FP by binding to the fluorophore (6-FAM) tag of the oligo, but not by directly binding to the G4 structure, we performed a UV-Vis titration experiment with TMPyP4 and unlabeled G4-forming KRAS oligo (Supplementary Figure S4C). We observed a dissociation constant (K d ) of 11 nM between TMPyP4 and KRAS G4, thus confirming that the ligand indeed has a very strong affinity to bind to the G4 structure (Supplementary Figure S4D). Upon treatment with PDS or TMPyP4, we observed a significantly increased KRAS expression in MIA PaCa-2 cells ( Figure 4D). In contrast, the same treatment in the APE1 KD cells failed to significantly increase KRAS expression ( Figure 4D), again highlighting the critical role of APE1 in modulating the expression of KRAS through regulation of the formation of stable G4 structure. Similar changes in KRAS expression were observed upon treating the PANC-1 cell line with PDS (Supplementary Figure  S4E). Taken together, these data demonstrate that APE1 plays a crucial role in G4 structure formation in the KRAS promoter, thereby regulating KRAS expression.

Redox function of APE1 is not involved in the regulation of G4 structure formation and the G4-mediated KRAS gene expression
So far, our data suggest that APE1 binds to KRAS G4 structure and promotes the binding of MAZ and PARP1 to the promoter region to activate gene KRAS expression. Since APE1 has been shown to stimulate the DNA-binding activity of a number of TFs through its redox function, thereby regulating gene expression (38,44), we investigated whether the redox function of APE1 is involved in regulation of G4 formation and G4-mediated KRAS expression. Using immunofluorescence studies, we observed that the loss of G4 staining in APE1 KD MIA PaCa-2 cells can be rescued by re-introduction of the APE1 redox mutant C65S/C99S (70,71) in these cells, indicating that the redox function of APE1 is not involved in the regulation of the G4 structure formation ( Figure 5A and B). Furthermore, our G4:APE1 PLA experiment with the C65S/C99S APE1 revealed appreciable number of PLA puncta indicating that the APE1 redox mutant has the ability to interact with G4 structures in the cells (Figure 5C and D). To further confirm that the redox function of APE1 is not involved in G4 regulation, we used APE1 redox inhibitor, E3330, and performed G4 and APE1 immunofluorescence studies. Small molecule E3330 has been extensively characterized as a direct and selective inhibitor of the redox function of APE1 (72)(73)(74). We observed no change in the G4 staining upon treating the cells with increasing doses of E3330, thus confirming that the redox function of APE1 is not involved in the regulation of G4 formation in cells ( Figure 5E). We also examined the effect of redox mutant C65S/C99S APE1 on the KRAS expression using the qRT-PCR assay. Our analysis indicated no significant change in the KRAS expression in C65S/C99S APE1 transfected cells when compared to the control cells, indicating that the redox function is not involved in regulating KRAS expression ( Figure 5F).

Downregulation of APE1 sensitizes PDAC cells to chemotherapy in vitro and suppresses xenograft tumor growth in vivo
Being a multifunctional protein with DNA damage repair, redox function and transcriptional regulatory roles, APE1 has been shown to promote cell survival and resistance to many chemotherapeutic drugs (75)(76)(77). In vitro colony formation assays revealed that APE1 KD in MIA PaCa-2 cells significantly reduced the number of colonies compared to the control MIA PaCa-2 cells ( Figure 6A). Next, we challenged the APE1 KD cells with increasing doses of chemotherapeutic drugs 5-FU, oxaliplatin or gemcitabine that are routinely used for the treatment of PDAC. Several studies have shown that these drugs can exert their cytotoxicity by multiple mechanisms, including generation of oxidative base damage lesions and abasic sites in the DNA that are repaired by the APE1-mediated BER pathway (78)(79)(80). Further, expression of WT or mutant KRAS was also found to be associated with promoting resistance to these drugs in colon cancer cells and patients (81,82). We found KD of APE1 resulted in significant increases in sensitivity of MIA PaCa-2 cells to oxaliplatin ( Figure 6B), gemcitabine (Figure 6C) and 5-FU (Supplementary Figure S5A). Similar increases in sensitivity to these chemotherapeutic agents were also observed upon APE1 KD in PANC-1 (Supplementary Figure S5B and C) and BxPC-3 cells (Supplementary Figure S5D-F) when compared to the control cells. These data together suggest that targeting APE1 represents an attractive approach to sensitize PDAC cells to chemotherapy.
To examine whether KD of APE1 can also enhance sensitivity to 5-FU and oxaliplatin and suppress PDAC tumor growth in vivo, we utilized a tumor xenograft model with MIA PaCa-2 APE1shRNA cells. When palpable tumors were visible, mice were fed drinking water without Dox or with Dox (2 mg/ml) to downregulate APE1 levels. Then, the effect of 5-FU and oxaliplatin on tumor growth of MIA PaCa-2 APE1shRNA cells was evaluated. While treatment with either drug alone showed moderate effect on the tu-mor growth in control (without Dox) mice as indicated by the tumor volume, the same treatment significantly reduced the tumor growth of MIA PaCa-2 APE1shRNA cells in mice fed with Dox, demonstrating that KD of APE1 sensitizes PDAC cell growth in vivo ( Figure 6D and E). Further, IHC analysis confirmed the KD of APE1 level in the tumor tissues of mice fed with Dox (Supplementary Figure S5G). Collectively, our data demonstrate that KD of APE1 sensitizes PDAC cells to routinely used chemotherapy drugs both in vitro and in vivo.

Elevated levels of APE1 and G4 DNA are present in PDAC patient's tissue samples
Given that APE1 is often overexpressed in multiple types of cancers including PDAC (47), and our current observation that APE1 regulates G4 DNA formation and promotes KRAS expression, we determined the levels of APE1 and G4 DNA in 23 PDAC patients' tissue samples and 2 normal control pancreas tissues using IHC analysis with APE1 and G4 structure-specific antibodies. PDAC patient TMAs were obtained from the Rapid Autopsy Program for Pancreas at UNMC. Among these samples, 18 patients were at stage IV, 1 at stage III and 4 at stage II. IHC scoring based on percentage of positively staining cells and staining intensity revealed that APE1 and G4 DNA levels were significantly higher in PDAC tumor tissues compared to normal controls ( Figure 7A). Further analysis revealed a strong positive correlation between APE1 and G4 staining in these tissue samples (Spearman's r = 0.8413, P < 0.0001), specifically in stage IV tumors ( Figure 7B). This indicates that increased APE1 levels are associated with elevated levels of G4 DNA in highly aggressive PDAC tumors.
To determine the clinical significance of elevated levels of G4 and APE1 in PDAC, we extended our analyses by correlating APE1 and G4 DNA levels with the survival of the patients who were administered chemotherapies including gemcitabine, capecitabine, leucovorin-modulated 5-FU, oxaliplatin, etc. Our survival analysis showed that among the 23 PDAC patients, 7 patients with the highest IHC staining score for APE1 and G4 survived for fewer days compared to the rest with lower staining scores ( Figure 7C). However, due to the limitation in the number of patient's tissue samples, we were unable to find a statistically significant difference between the two groups. Nonetheless, these data provide preliminary evidence that the expression levels of APE1 and G4 DNA negatively correlate with PDAC patients' prognosis.

DISCUSSION
Over the past decade, G4 structures have gained recognition as genomic features involved in KRAS transcriptional regulation (30,83). However, little is known about the factors that regulate the formation of stable G4 structures in the KRAS promoter in PDAC cells. In this study, we have demonstrated that APE1 plays a crucial role in the formation of stable G4 structures and regulating KRAS expression in PDAC cells. Loss of APE1 abrogated the formation of G4 structures and the loading of TFs such as MAZ and PARP1 on the KRAS promoter, resulting in reduced KRAS  were treated with or without Dox and the effect of APE1 KD on cell survival was evaluated using a colony formation assay. An unpaired Student's t-test comparing the number of colonies in control versus APE1 KD cells was used to determine the P-value (****P < 0.0001). MIA PaCa-2 APE1shRNA and MIA PaCa-2 NTC cells treated without or with Dox for 4 days and then exposed to indicated doses of oxaliplatin (B) or gemcitabine (C). The graphs show the percentage cell viability obtained from colony formation assays, performed three times in triplicates. (D) MIA PaCa-2 APE1shRNA cells were subcutaneously implanted in nude mice. When palpable tumors were visible, mice were divided into six groups (n = 5 in each group). Three groups were fed with Dox (2 mg/ml) and other three groups without Dox containing drinking water following which 5-FU (30 mg/kg) or oxaliplatin (2 mg/kg) were injected intraperitonially three times a week for 4 weeks and tumor volume was recorded using a caliper. Resected tumors after completion of treatment are shown. (E) Tumor volume was measured at indicated days and tumor growth curve was plotted. The P-values were determined using an unpaired Student's t-test (****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05). Error bars denote ±SD. expression in multiple PDAC cell lines. Further, by demonstrating that APE1 can strongly bind to KRAS promoter G4 structure and stabilize the folding in vitro, we identified APE1 as a G4-binding protein. Supporting this idea, we have provided evidence that APE1 colocalizes and interacts with G4 structures in cells. Importantly, using PDAC patients' tissue samples we have demonstrated higher APE1 and G4 DNA levels in the tumor tissues compared to control pancreatic tissues. Therefore, our study has identified and characterized APE1 as a key regulator of the formation of stable G4 structures in the genome and advanced our understanding of the mechanism underlying G4 regulation and G4-mediated transcriptional control of KRAS in PDAC cells. Given that APE1 is overexpressed in PDAC (47) and G4s are known to regulate cancer driver genes (23), our findings also implicate that G4 can be considered as a novel prognostic marker and promising therapeutic target in PDAC.
APE1 is a multifunctional protein with DNA damage repair and transcriptional regulatory Ref-1 functions (84). Its primary function in the repair of spontaneously generated AP sites in the genome via the BER pathway and its role in regulating gene expression via redox function (APE1 reduces the oxidized cysteine in many TFs) have been well established (37,38). In this study, we have discovered a novel and noncanonical function of APE1 in reg- ulating gene expression through regulating the formation of DNA secondary structures, G4s. Recent ChIP sequencing using a G4 structure-specific antibody has revealed that endogenous G4s are enriched in nucleosome-depleted promoter/enhancer regions upstream of TSSs (21). Furthermore, most of the endogenous G4s are hubs for binding of TFs (22). Thus, it is possible that elevated gene expression results from increased TF loading onto the promoter G4s and recruitment of RNA Pol II. A recent study has demonstrated that promoter G4 folding leads to the retention of RNA Pol II, suggesting that G4s act as a site for the recruitment of key components of the transcriptional machinery as well as TFs (85). This is supported by the observation that several TFs, including SP1 (86), CNBP (87), PARP1 (88), MAZ (89) and LARK (90), display high-affinity binding for G4s in vitro. Consistent with this, multiple previous studies from Xodo's group have established that binding of MAZ and PARP1 to KRAS promoter G4 plays a key role in activation of KRAS expression (69,89). Here, we add to the current knowledge by providing compelling evidence that APE1 regulates KRAS expression through a mechanism that involves facilitating the formation of promoter G4 structure and loading of the TFs such as MAZ and PARP1 to the promoter region. Several lines of evidence were pro-vided to support this: (i) APE1 colocalizes and interacts with G4 structures in cells, and recombinant APE1 can directly bind KRAS promoter G4 structure in vitro; (ii) downregulation of APE1 reduces stable G4 formation, loading of MAZ and PARP1 on KRAS promoter, and KRAS expression; and (iii) G4-stabilizing ligands fail to activate KRAS expression in APE1 downregulated cells, suggesting APE1 is required for G4-mediated KRAS expression.
The exact mechanism by which APE1 regulates the formation of stable G4 in cells is not fully understood. Active transcription-induced torsional stress and negative superhelicity had been proposed to stimulate G4 formation at promoters (91,92). However, a recent study has demonstrated that folding of G4s in promoters does not necessarily require active transcription but can be favored by an accessible open chromatin state (85). This raises the possibility that potential quadruplex sequences (PQSs) in open chromatin promoter regions are in dynamic equilibrium between double-stranded and transient G4 folding forms. We propose that binding of APE1 to transiently formed G4 may shift the equilibrium toward a stable G4 structure. Consistent with this, our recent genome-wide binding analysis revealed that APE1 is highly enriched in promoter enhancer regions that overlap with G4s (35). Although low-resolution mapping (300 bp) demonstrated significant overlaps between AP sites and G4s (35), our current study demonstrates that recombinant APE1 has high affinity for binding to a G4 structure without any AP site, suggesting that an AP site is not a prerequisite for APE1 to bind to G4. Therefore, we postulate that the PQS in open chromatin region transiently generates G4 and binding of APE1 promotes the formation of a stable G4 structure. The stable G4 structure acts as a loading platform for the TFs MAZ and PARP1 that subsequently drive KRAS expression (Figure 8). Binding of APE1 to noncanonical (non-B-form) DNA secondary structures is not unprecedented. APE1 was shown to recognize many non-B-form DNA structures such as cruciform DNA that is formed by a palindromic core sequence (TGAGACAGGGTCTCA) known as negative calcium response element sequence (nCaRE), present in many gene promoters (93,94). Indeed, previous studies from our lab and others have shown APE1-dependent regulation of nCaRE-B containing human PTH, renin and SIRT1 gene promoters (43,(93)(94)(95). Thus, it appears that APE1 binds and promotes the formation of stable G4 structures in promoter/enhancer regions of many genes that provide a loading platform for the recruitment of TFs to activate gene expression. Further studies examining how CRISPR/Cas9mediated deletions or mutations of G4-forming sequences on the promoter of KRAS gene affect the recruitment of TFs and KRAS expression are warranted.
Many earlier studies have shown that treatment of cells with G4-stabilizing ligands reduces KRAS expression and thus portrayed G4s as antagonists to KRAS gene expression (17,30). In contrast, our results have demonstrated that treatment with G4-stabilizing small molecules PDS and TMPyP4 for a shorter duration (2 h) activates KRAS expression in PDAC cells, suggesting G4 as a positive regulator for KRAS expression. The apparent discrepancy between these results can be explained by the fact that the previously published studies have relied on the effect of prolonged treatment with these ligands (12 h or more), which induced DNA double-strand breaks that negatively affect transcription (96). It was shown that treatment with the potent G4 ligand PDS caused DNA damage at G4 sites in a transcription-and replication-dependent fashion (97). Similarly, Hurley and coworkers reported that prolonged treatment with G4 ligands could lead to MYC downregulation by an indirect consequence of the ligands eliciting global G4 stabilization (98). Our study showed that APE1-mediated formation of stable G4 structure activated KRAS expression. In support of this, emerging evidence suggests that G4s can act as positive regulators of transcription by multiple mechanisms (99). This includes the role of G4s in mediating the placement of histone marks and in interacting with chromatin remodeling proteins, thus facilitating TF and RNA Pol II loading (85). Additionally, G4s have a positive effect on transcription by stabilizing R-loop formation (100). G4s can influence long-distance gene regulation by promoting distal interactions between enhancers and promoters (101).
APE1 is overexpressed in PDAC, and elevated levels of APE1 and its redox function in PDAC have been shown to promote tumor cell survival, migration and angiogenesis, and confer resistance to chemotherapeutic drugs (47,56,102). We have shown that downregulation of APE1 inhibited tumor growth and sensitized these cells to conventional chemotherapy drugs gemcitabine, 5-FU and oxaliplatin both in vitro and in a in vivo xenograft model. Even though oxaliplatin exerts its cytotoxicity primarily by inducing DNA intra-or interstrand cross-links that are repaired by the nucleotide excision repair pathway (103), studies have shown that treatment with platinum-based drugs cisplatin and oxaliplatin produces reactive oxygen species that damage DNA by oxidizing DNA bases in cells (78,79). APE1 plays a key role in the repair of these damaged bases via the BER pathway. Similarly, 5-FU or its metabolites that primarily inhibit thymidylate synthase (104), the key enzyme of de novo pyrimidine biosynthesis, can also directly incorporate into genomic DNA. A previous study has shown that uracil DNA glycosylase SMUG1 efficiently removes 5-FU from DNA in vitro and in cells to generate AP sites that are repaired by APE1-mediated BER (80,105). Furthermore, APE1 redox function was also shown to con-fer 5-FU resistance in colon cancer because treatment with APE1 redox inhibitor E3330 enhanced tumor response to 5-FU in a colon cancer xenograft model (54). Our previous studies and others showed that APE1 downregulation sensitizes colon and breast cancer cells to 5-FU and cisplatin/oxaliplatin in vitro (106,107). Since the expression of WT or mutant KRAS has also been shown to be associated with resistance to these drugs in cancer (81,82), our study implicates that APE1-mediated KRAS expression may contribute to promoting resistance to these drugs in PDAC. Therefore, we propose that elevated levels of APE1 promote proliferation and drug resistance in PDAC by facilitating repair of damaged bases, activating gene expression through its redox function and controlling G4-mediated KRAS expression.
Using rapid autopsy tissue samples from PDAC patients, we demonstrated for the first time a correlation between APE1 levels and G4 DNA in human PDAC tissue specimens. This observation has important implication for regulatory functions of APE1 in tumor progression, as cancer cells have increased levels of G4 DNA and APE1, both of which regulate expression of cancer driver genes. Elevated G4 staining in tumor tissues of stomach and liver cancer patients suggested that cancer cells have increased levels of G4 DNA (108). Notably, recent genome-wide mapping of G4 DNA forming regions in 22 breast cancer patientderived tumor xenografts revealed that G4s are significantly enriched in promoters of highly expressed genes when compared to those with medium or low expression (109). Furthermore, increased G4 DNA is associated more often with the signature cancer-driving genes in aggressive triple negative breast cancer, as compared to other subtypes. We propose that APE1 plays a crucial role in PDAC oncogenesis by binding and promoting the formation of stable G4 structures in promoter/enhancer regions, which in turn facilitate the loading of TFs onto the promoters to regulate G4-mediated gene expression. Thus, identification of small molecules that bind to APE1 and affect its interaction with G4 could be a promising strategy for PDAC therapy.
In conclusion, our study has identified APE1 as a G4 DNA-binding protein and key regulator of stable G4 formation, and unraveled G4-APE1 functional complex as a major determinant for KRAS oncogene regulation in PDAC, underscoring a novel therapeutic opportunity.