Cysteine oxidation to sulfenic acid in APE1 aids G-quadruplex binding while compromising DNA repair

Abstract

Apurinic/apyrimidinic endonuclease-1 (APE1) is a base excision repair (BER) enzyme that is also engaged in transcriptional regulation. Previous work demonstrated the enzymatic stalling of APE1 on a promoter G-quadruplex (G4) recruits transcription factors during oxidative stress for gene regulation. Also during oxidative stress, cysteine oxidation is a posttranslational modification (PTM) that can change a protein’s function. The current study provides a quantitative survey of cysteine oxidation to sulfenic acid in APE1 and how this PTM at specific cysteine residues affects the function of APE1 toward the NEIL3 gene promoter G4 bearing an abasic site. Of the seven cysteine residues in APE1, five (C65, C93, C208, C296 and C310) were prone to carbonate radical anion oxidation to yield sulfenic acids that were identified and quantified by mass spectrometry. Accordingly, five Cys to Ser mutants of APE1 were prepared and found to have attenuated levels of endonuclease activity, depending on position, while K_D values generally decreased for G4 binding indicating greater affinity. These data support the concept that cysteine oxidation to sulfenic acid can result in modified APE1 that enhances G4 binding at the expense of endonuclease activity during oxidative stress. Cysteine oxidation to sulfenic acid residues should be considered as one of the factors that may trigger a switch from base excision repair activity to transcriptional modulation by APE1.

Graphical Abstract

Organisms living under aerobic conditions are exposed to oxidative stress via reactive oxygen species (ROS) resulting from cellular respiration or inflammation.¹ These ROS have the ability to damage various cellular macromolecules such as DNA, RNA, proteins, or lipids. In DNA, the guanine (G) base is the most electron rich and thus, the most susceptible base for oxidation leading to several oxidation products, among which 8-oxo-7,8-dihydroguanine (OG) is a major one (Figure 1A).^2,3 The human genome contains certain G-rich sequences that can adopt G-quadruplex (G4) folds (Figure 1B).^4,5 These potential G-quadruplex forming sequences (PQSs) consist of at least four closely spaced G-runs with at least three Gs per run that form G-tetrads via Hoogsteen bonds; the G-tetrads stack on one another to form a secondary structure stabilized by coordination to K⁺ ions in the core of the fold (Figure 1C).^6,7 Some gene promoter PQSs including NEIL3 contain a fifth G-track that is recruited to function as a “spare tire” when one of the four tracks of a G4 becomes damaged due to oxidative stress, and the undamaged fifth track is swapped in to replace the oxidized section in the G4 (Figure 1D).^8,9 A G oxidation event can occur at either a G in a core position or in a loop position of the G4 structure. Genome-wide sequence analyses reveal the enriched presence of PQSs in human gene promoters that also manifest formation of OGs suggesting a possible role for these PQS in modulating gene transcription of many human genes under oxidative stress.^4,10-12 In the human genome, the DNA repair gene promoters show nearly twofold greater PQSs density than average.¹² NEIL3 is a gene encoding a DNA repair glycosylase and is one member of this family.

Figure 1: G tracks have both a high susceptibility to oxidation and a propensity to fold to G4s.

(A) Two-electron oxidation of G yields OG, which is a substrate for OGG1. (B) The consensus sequence for a PQS. (C) The structure of a G-tetrad and a 3-layer parallel-stranded G4. (D) Oxidative stress on a promoter PQS (NEIL3) yields OG that destabilizes the G4 structure when in a core position. This results in the recruitment of the fifth track to act as a “spare tire” extruding the damaged G run into a loop.

Formation of OG under oxidative stress, if left unrepaired, gives rise to G0T transversion mutations that are observed in cancers.¹³ When OG is base paired with C, the base excision repair (BER) glycosylase 8-oxoguanine glycosylase 1 (OGG1) removes the oxidized base resulting in an abasic site (AP) that is passed on to the next BER protein, apurinic/apyrimidinic endonuclease-1 (APE1).^14-16 The response of DNA repair proteins to oxidized lesions in gene promoters positioned in secondary structures such as G4s can modulate gene expression.^17,18 For example, it was suggested that APE1 plays a role in transcription regulation of VEGF and NEIL3 genes after the BER pathway is initiated in the gene promoter PQSs (Figure 2A).^19-21

Figure 2: The dual functionality of APE1 may arise from post-translational modification (PTM) of cysteine residues in the protein.

(A) Proposed mechanism for the dual function of APE1 when a G nucleotide in a gene promoter PQS is oxidized to OG and then converted to an AP site. In a duplex substrate context, APE1 acts as a BER protein, whereas in a G4 substrate context APE1 acts as a transcription modulator. (B) Human APE1 has seven cysteine residues some of which are close to the catalytic residues of the enzyme.

Numerous laboratories have carried out experiments to demonstrate that G oxidation in a gene promoter uses the BER pathway to recruit both DNA repair proteins and transcription factors to drive gene transcription.^22-27 Several groups have further explored the impact of cysteine oxidation, a PTM, in BER proteins and how it alters repair and gene transcription; notably, cysteine oxidation to sulfenic acid in OGG1 helped to switch the glycosylase from a repair protein to a transcriptional activator protein.²⁴ Other work studied the impact of cysteine oxidation in OGG1 by making Cys to Ser mutants and observing their impact on OGG1 function.²⁸ These studies support mechanisms proposed by several groups on how OGG1 induces transcriptional activation and demonstrate that oxidation of specific cysteine residues in OGG1 can be involved in gene expression.^24,28 Moreover, oxidation of cysteine residues to sulfenic acids is a reversible modification that can be reduced after the redox balance in a cell is restored, which makes this an ideal on/off switch in response to ROS to modulate gene expression.²⁹

Like OGG1, APE1 is a BER protein involved in transcription modulation and containing seven cysteine residues that suggest further investigation (Figure 2B). Also known as redox effector factor 1 (Ref-1) due to its ability to regulate gene transcription during redox changes, APE1 was first discovered as a nuclear protein that co-purified with mammalian transcription factor activator protein-1 (AP-1).³⁰ APE1 has the ability to modulate the reduction of an inactivating disulfide in the DNA binding domain of Jun to yield a protein capable of DNA binding as part of the AP-1 heterodimer.^30-33 Since the initial studies, APE1 has been found to reduce a number of other important transcription factors such as NF-κB, HIF-1α, and p53 for their activation.^34-36

Apart from being multifunctional, another unique characteristic of APE1 is that despite the thiol-based redox activity, APE1 lacks the conserved CXXC motif that is present in other redox regulating proteins such as thioredoxin and glutaredoxin.^37-39 In fact, none of the seven cysteine residues present in APE1 are well positioned to favor the formation of an intramolecular disulfide bond as seen in several crystal structures of APE1.^40,41 Thus, the redox mechanism of APE1 has been deduced to be more complicated than a traditional thiol-exchange mechanism.⁴²

Numerous studies were conducted to elucidate which cysteine residues are important in the mechanism of APE1 redox activating transcription factors that subsequently modulate eukaryotic gene expression. Among the seven cysteines in human APE1, C65 stands out as being indispensable for the redox activity of the protein.⁴³ Several laboratories have observed that C93 is also required for APE1 redox activity and is important in the redox mechanism by possibly forming a disulfide bond with C65.^30,39 Additionally, C99 and C310 were found to be potential players in triggering the dual function of APE1.^44-46 Prior studies treated APE1 with H₂O₂ and found intramolecular disulfide bonds and intermolecular disulfide bonds with glutathione as terminal oxidation products.³⁶⁷ Altogether, these studies demonstrate cysteine residues in APE1 are susceptible to oxidation that impacts function.

In our previous work, we proposed that the interaction of APE1 with a promoter G4 is a key step in the mechanism of oxidative stress-mediated gene induction.^20,47 Questions regarding cysteine oxidation as a PTM in the G-quadruplex-affiliated mechanism have not been addressed by our laboratory, and sulfenic acid formation in APE1 has not been studied. We were particularly interested in cysteine oxidation induced by carbonate radical anion, CO₃^•−, an important ROS during inflammatory oxidative stress. Our current study focuses on a quantitative survey via mass spectrometric analysis of APE1 oxidation by CO₃^•− to sulfenic acids as a possible PTM. Individual Cys to Ser and Asp mutants were prepared as models of key sites of sulfenic acid residues in APE1. Then, endonuclease activity and binding properties with the NEIL3 promoter duplex vs. G4 DNA bearing an AP site were studied. Among the five Cys to Ser APE1 variants, C65S and C310S were intriguing due to their attenuated endonuclease activity and extremely tight binding with G4 substrates. The data from our study support the concept in which APE1 binds to an AP site in a promoter G4 in such a way that endonuclease activity is greatly diminished although the binding interaction is strong; the stalling of APE1 on the G4 DNA could then enable the recruitment of transcription factors for gene regulation. Cysteine oxidation in APE1 facilitates this change from repair enzyme to transcription modulator by decreasing the affinity of APE1 for duplex substrates and increasing its binding to G4s, a context in which cleavage at AP sites is very inefficient.

Results

Oxidation of APE1 decreases endonuclease activity while increasing G4 binding

When a PQS is positioned in a gene promoter region and the cells are undergoing oxidative stress, APE1 has the ability to act as a base excision repair protein or a gene transcription modulator. One factor enabling the interchange between the dual functions of APE1 could be the PTMs occurring in APE1 during oxidative stress. In order to study the effects of oxidized WT APE1 on the endonuclease activity and the binding affinity with duplex and G4 substrates, APE1 was oxidized with CO₃^•−, and APE1_OX was used in both activity and binding assays.

To study APE1_OX cleavage of an F residue, a tetrahydrofuran mimic of an AP site in DNA, the NEIL3 promoter PQS folded either as a G4 or in the duplex DNA context was used (denaturing PAGE shown in Figure S1). The 4-track and 5-track G4s with the F site positioned in the core of the G4 structural fold were explored, and the corresponding duplex substrate with a central F was used as a positive control for the enzyme activity (Table 1). When the DNA substrates were in the duplex context, the incision activity of APE1_OX oxidized for either 15 or 60 min decreased only moderately compared to those observed with 0 min oxidation. The cleavage yields for duplex DNA decreased from 91% to 80% over the course of 60 min exposure to oxidation (Figure 3A). When the DNA substrates were in a G4 context, the cleavage yields were much lower than observed with duplex substrates. As shown in Figure 3A, when the F was in the core position of the NEIL3 4-track G4, APE1_OX produced cleavage yields of 30% after no oxidative stress and this decreased to 7% after 60 min exposure to carbonate radical anion. Similarly, in the NEIL3 5-track G4, APE1_OX yielded 10% cleavage of the AP site with no oxidative stress and only 3% cleavage yield after 60 min of oxidation. Thus, the impact of APE1 oxidation was moderately detrimental to endonuclease activity while the major factor was the structural context—duplex vs. quadruplex.

Table 1:

Sequences used in the study

NEIL3 4-track WT G4	N3-4 G4	5‘TT GGG C GGGG CCT GGG C GGGG CC 3‘
NEIL3 4-track F-core G4	N3-4 core	5‘TT GGG C GGGG CCT FGG C GGGG CC 3‘
NEIL3 4-track F-loop G4	N3-4 loop	5‘TT GGG C FGGG CCT GGG C GGGG CC 3‘
NEIL3 5-track WT G4	N3-5 G4	5‘TA GGG TGCTGTTT GGG C GGGG CCT GGG C GGGG CC 3‘
NEIL3 5-track F-core G4	N3-5 core	5‘TA GGG TGCTGTTT GGG C GGGG CCT FGG C GGGG CC 3‘
NEIL3 5-track F-loop G4	N3-5 loop	5‘TA GGG TGCTGTTT GGG C FGGG CCT GGG C GGGG CC 3‘
NEIL3 F duplex	N3 duplex	5‘TT GGG C GGGG CCT FGG C GGGG CC 3‘ 3‘A CCC G CCCC GGA CCC G CCCC GG 5‘
NEIL3 WT duplex	N3 WT duplex	5‘TT GGG C GGGG CCT GGG C GGGG CC 3‘ 3‘AA CCC G CCCC GGA CCC G CCCC GG 5‘

Figure 3: Oxidation of WT APE1 decreases endonuclease activity while increasing G4 binding affinity in NEIL3 DNA substrates.

Data for (A) endonuclease activity assays, and (B) binding assays conducted with WT APE1 oxidized with CO₃^•− for 0 min,15 min or 60 min. Significance values for each comparison were calculated by a Student’s t test (* P > 0.05 or ** P > 0.01 is indicated).

Binding affinities with APE1_OX were measured with the NEIL3 promoter PQS annealed either as a duplex DNA substrate or as 4-track or 5-track G4s with the F site in a core position; this results in a poorly folded G4 for the 4-track sequence but a stably folded G4 for the 5-track one. The results indicate that APE1_OX binds to duplex DNA substrates with K_D values dependent on oxidation time; namely, when APE1 is not oxidized, K_D is 29 nM, but it increases sharply at 15 and 60 min exposure to oxidant with K_D = 75 and 164 nM, respectively (Figure 3B). In contrast, the K_D values of APE1_OX binding NEIL3 G4s showed the opposite trends with K_D values decreasing as a function of protein oxidation, indicated tighter binding to G4s as a function of oxidation (Fig. 3B). For example, the K_D value for the better folded G4 structure (NEIL3 5-track core F) binding to APE1_ox was nearly as low (37 nM) as the duplex DNA substrate for wild-type APE1 (29 nM). Two Mg²⁺ ions serve different roles in APE1; one is involved in the endonuclease activity while the other is involved in substrate binding.^48-50 Thus, to prevent strand breaking, 10 mM Sm³⁺ was used as a replacement for Mg²⁺ in these studies because it is not catalytically active but enables binding to occur.⁵¹

Overall, the APE1_OX protein with duplex DNA substrates displayed similar cleavage yields as native APE1 and weaker binding affinities with increased oxidation time of the protein. In contrast, with F-core 4-track G4, and the 5-track G4, APE1_OX showed somewhat lower cleavage yields and markedly stronger binding reflected by lower K_D values with increasing oxidation time. With the aim of locating which cysteines present in APE1 are susceptible to oxidation and responsible for these impacts, we next quantified cysteine oxidation in APE1 via mass spectrometry.

Oxidation of APE1 has a greater effect on certain cysteine residues

Human APE1 has seven cysteines that could be oxidized. As shown in Figure 4A, the thiol sidechain in cysteine has the ability to undergo several oxidative post transcriptional modifications including sulfenic acid, sulfinic acid, sulfonic acid, or disulfides.⁵² Due to the inherently reactive and transient nature of sulfenic acids, dimedone is used as a trapping agent to provide the corresponding thioether derivative, which is stable.^53,54 In order to identify which cysteines are more susceptible to oxidation resulting in new products, APE1 was oxidized with either CO₃^•− or H₂O₂ in the presence or absence of dimedone. Quantification was carried out by the addition of ¹⁵N-labeled WT APE1 post-reaction as an internal standard for the samples in a 1:1 molar ratio. The protein mixture was digested by trypsin and analyzed by LC/MS/MS following a standard protocol (Figure S2).

Figure 4: Oxidation of APE1 cysteine thiols yields different oxidation products.

(A) Cysteine thiols can be oxidized to sulfenic, sulfinic, sulfonic acids, or disulfides. Following APE1 oxidation with CO₃^•−, cysteine thiol oxidation yield for different products were quantified using LC-MS/MS. The data obtained (B) in the presence of DMD, and (C) in the absence of DMD.

When WT APE1 was oxidized with CO₃^•− in the presence of dimedone, the major oxidation products observed were sulfenic acids at five cysteine residues and a disulfide bond between C65 and C93 (Figure 4B). The oxidation yields of sulfenic acids ranged from 11% for C93 to 49% for C65, and the C65-93 disulfide yield was 27% under these conditions. Sulfenic acid formation for cysteines 99 and 138 could not be detected, and neither sulfinic acids nor sulfonic acids could be detected under the current experimental conditions. However, in the absence of dimedone as a sulfenic acid trapping agent, the over-oxidation products sulfinic acids and sulfonic acids were observed (Figure 4C) with yields ranging from 12-37% for sulfinic acids and from 19-33% for sulfonic acids. Overall, C65 displayed the highest level of total oxidation. Similar sites of oxidation, products, and yields were observed when WT APE1 was oxidized with H₂O₂ in the presence or absence of dimedone (Figure S3). Negative controls were conducted in the absence of an oxidant, in which all cysteine residues were thiols verifying the products resulted from oxidation and were not formed during protein preparation (Figure S4).

These in vitro data are consistent with a mouse model study in which the researchers discovered the equivalent murine cysteines to those found herein exist as sulfenic acids when the mice experience oxidative stress.⁵⁵ To further study the effects of cysteine thiol oxidation into sulfenic or sulfinic acids in APE1, amino acid mimics of sulfenic or sulfinic acids were designed, expressed, and purified. Cysteine (Cys) to serine (Ser) variants of WT APE1 were generated individually at C65, C93, C208, C296, and C310 to study sulfenic acids (Figure S5). Mutants to study sulfinic acids at positions C65, C208, and C310 were generated by the replacement of Cys with aspartate (Asp).

Impact of certain cysteines on the thermal stability of APE1

In order to test the impact of Cys to Ser variation in APE1, the thermal stabilities of the proteins were determined by measuring the change in UV absorbance at 280 nm as a function of the temperature (Figure S6). The WT APE1 has a thermal melting (T_m) value of 45 °C (Figure S7), which is consistent with the literature.⁵⁶ Among the cysteines found to be oxidized the most into sulfenic acids, the variants were found to have T_m values of C65S = 46 °C, C93S = 43 °C, C296S = 50 °C, and C310S = 45 °C that were similar to the WT APE1. The most significant loss of thermal stability was observed with the C208S mutant leading to a T_m value of 38 °C. Further studies using circular dichroism showed the Cys variants did not result in possible changes on the protein folding (Figure S8).

APE1 cysteine mutants affect APE1 endonuclease activity

To study APE1 Cys to Ser variants for cleavage of an F residue within the context of DNA substrates, the NEIL3 promoter PQS folded either as a G4 or in the duplex DNA context was studied. The 4-track and 5-track G4s with two different positions of the F site in the G4 structural fold were explored, and the duplex substrate with a central F was used as a positive control for the enzyme activity (Table 1). In the APE1 activity assays, the ability of each APE1 Cys to Ser variant to hydrolyze the phosphodiester bond at the 5‘ side of the AP site to yield a strand break was explored using 10 nM DNA substrates with 3 nM enzyme concentration. The cleavage yields were calculated by quantifying the cleavage bands resolved on a denaturing PAGE as shown in Figure 5A (PAGE gels for the other sequences are shown in Figure S9).

Figure 5: Monitoring APE1-mediated endonuclease activity of an AP site in the NEIL3 sequence using PAGE analysis to identify position- and context-dependent yields with different C to S mutants.

(A) Example of denaturing PAGE analysis of APE1-catalyzed cleavage of AP site analogue F in the NEIL3 PQS in ds DNA with WT APE1 and the cysteine mutants of APE1. Yield (%) was plotted for cleavages with different C to S APE1 mutants at 60 min in the (B) ds DNA context with F at position 14 (N3 ds), as well as four- and five-track G4 contexts with F at (C) a core position or (D) a loop position. The reactions were conducted by pre-incubating 10 nM DNA in 20 mM Tris at pH 7.5, 50 mM KOAc, 10 mM Mg(OAc)₂, and 1 mM DTT present at 37 °C for 30 min. After the pre-incubation, APE1 or the respective cysteine mutant was added to a 3 nM final concentration, and the reactions were quenched at 60 min by adding a stop solution and heating at 65 °C for 20 min before PAGE analysis. (n.r. = no reaction) Significance values for each comparison were calculated by a Student’s t test (* P > 0.05 or ** P > 0.01 is indicated).

When the DNA substrates were in the duplex context, all Cys to Ser variants except C310S cleaved the DNA with yields similar to those observed with WT APE1, about 90%. In contrast, C310S showed a drastic decrease in the AP site cleavage to show a yield of only 18% (Figure 5B).

When the DNA substrates were in a G4 context, the cleavage yields were much lower than observed with duplex substrates. As shown in Figure 6C, when the F was in a core position of the NEIL3 4-track G4 such that the G4 was poorly folded, WT APE1 produced a cleavage yield of 30%; similarly, C65S APE1 yielded 23% cleaved product. All other Cys to Ser mutants led to lower yields of cleavage at the F site, and the cleavage yield with C310S was below the detectable level. Cleavage yields with F at a core position in the NEIL3 5-track G4, where one expects the fifth track to replace a damaged track to provide a more stable G4, were lower than the yields for the NEIL3 4-track G4 (Figure 5C). WT APE1 gave a cleavage yield of 10%, while C65S resulted in 5% yield, and C208S resulted in 10% yield. All other cleavage yields were lower than the detectable level under the current experimental conditions. Overall, the 5-track F-core G4s show more than twofold lower cleavage than 4-track F-core with all APE1 variants, as well as WT APE1, suggesting that the more stably folded G4s with the 5^th track engaged are poorer substrates for cleavage.

Figure 6: Fluorescence anisotropy binding assays for the interaction between APE1 and NEIL3 PQS to identify position- and context-dependent effects with different C to S APE1 mutants.

(A) Binding plots with log[APE1] vs. anisotropy (r) for the analysis of binding affinities with different APE1 Cys to Ser variants. The K_D values were obtained from fitting the sigmoidal binding curves with the Hill equation for binding assays. The NEIL3 duplex DNA without F (red) was studied to demonstrate the lack of APE1 binding in the absence of an AP in a duplex substrate. (B) The K_D values for the duplex DNA context with F at position 14 (N3 duplex), as well as four- and five-track G4 contexts with F at (C) a core position or (D) a loop position. The reactions were conducted by pre-incubating APE1 WT or mutants in 1 mM EDTA for 30 min at 22 °C. After the pre-incubation, APE1 or the respective cysteine mutant was added in a series of 1 to 5000 nM to 50 nM FAM-labeled DNA in 20 mM Tris at pH 7.5, 50 mM KOAc, 10 mM Sm(OAc)₃ and 1 mM DTT present at 22 °C for 30 min. Significance values for each comparison were calculated by a Student’s t test (* P > 0.05 or ** P > 0.01 is indicated).

The APE1 cleavage yields for NEIL3 4-track G4 with F in a loop were similar between the WT and Cys to Ser variants of APE1 except for C310S (Figure 5D). The C310S APE1 variant did not result in a detectable cleavage yield. Similarly, NEIL3 5-track G4 F-loop also showed similar cleavage yields between WT APE1 and the Cys to Ser variants except for C310S which did not result in detectable cleavage yields. Within error, the cleavage of F in a loop position by the APE1 enzymes were nearly identical except C310S.

APE1 cysteine mutants bind to AP DNA substrates with variable affinities

In order to understand the role of cysteine oxidation to sulfenic acids in APE1 to influence DNA binding, fluorescence anisotropy binding assays were conducted following a method previously reported.⁵⁷ The binding studies were conducted using the sequences shown in Table 1.

Initially, all binding assays were conducted without Mg²⁺ and in the presence of 1 mM EDTA to prevent any cleavage activity because this divalent metal is required for strand scission.⁵⁷ To observe the effect of the multivalent cation presence in APE1 binding, the assays were repeated in the presence of 10 mM Sm³⁺ as a substitute for Mg²⁺ with the purpose of preventing any catalytic activity on the substrates but not to impact effects on the binding.⁵¹ Our prior work found the binding between catalytically inactivated APE1 and a G4 was Mg²⁺ dependent.⁵⁷ Therefore, the discussion here will focus on the Sm³⁺ studies, but for completeness of the study, binding without Mg²⁺ was analyzed and is provided in Figure S10. First, an endonuclease assay control study was conducted to show that no cleavage occurs with 10 mM Sm³⁺ present (Figure S11). All binding assays mentioned in the main text were carried out with 10 mM Sm³⁺ present. The dissociation constants (K_D) were calculated by fitting binding profiles obtained from fluorescence anisotropy analyses (Figure 6A). Anisotropy curves for the other sequences and R² values for the curves are shown in Figure S12.

The results indicate that WT APE1 binds tightly to duplex DNA substrate with a K_D value of 32 nM which is similar to dissociation constant values previously reported (24-25 nM).^57-59 Among the APE1 mutants, the C65S variant showed the strongest binding affinity with the lowest K_D value of 44 nM. Binding affinities for other APE1 mutants were C93S = 98 nM, C208S = 104 nM, and C296S = 59 nM. The cysteine to serine mutant with the weakest binding affinity and the highest K_D value for duplex DNA was C310S at 190 nM. In duplex substrates, all APE1 cysteine to serine variants showed weaker binding affinities than WT APE1 with 10 mM Sm³⁺ present (Figure 6B).

The K_D value of WT APE1 binding NEIL3 4-track G4 F core was 131 nM, which shows considerably weaker binding affinity than WT APE1 bound to the duplex substrate (K_D = 32 nM). In contrast, all Cys to Ser mutants showed stronger binding affinities with F-core G4s compared to WT APE1. The Cys variants displayed K_D values ranging from 57 nM (C65S) to 71 nM (C208S) in 10 mM Sm³⁺. The binding affinity of WT APE1 to the 5-track core G4 was 76 nM, which is somewhat weaker than the binding affinity of WT APE1 to the 4-track core G4. Similar to the 4-track core G4, the 5-track G4 also showed stronger binding reflected by lower K_D values with all the APE1 cysteine mutants compared to WT APE1; the 5-track G4 K_D values were lowest for C65S (31 nM) and slight higher for all other Ser variants. As shown in Figure 6C, G4s with the F site in the core of the G4 structure showed tighter binding to all Cys to Ser variants compared to WT APE1. In the 4-track core G4, all variants showed tighter binding than WT APE1 and the binding affinities among the cysteine to serine sulfenic mimics were quite similar. In contrast, in the 5-track core G4, C65S and C310S showed stronger binding affinities compared to the other cysteine variants. Also, in NEIL3 5-track G4 the K_D values with cysteine mutants C65S and C310S decreased at least twofold compared to NEIL3 4-track G4. The 5-track G4 shows stronger binding affinities with C65S and C310S mutants compared to 4-track G4 when the F site is in the core position. Interestingly, APE1 Cys to Ser variant binding assays conducted on NEIL3 4-track and 5-track sequences with no F site present also showed tighter binding with C65S and C310S (Figure S13) following the same pattern as F-core G4s.

When the F site is positioned in the loop of the G4 structure, binding affinities between the Cys to Ser variants and WT APE1 did not show a considerable change (Figure 6D). In the 4-track loop G4 WT APE1 showed a K_D value of 57 nM whereas the cysteine mutants showed K_D values ranging from 45 nM to 78 nM. A similar pattern was followed by the 5-track loop G4 with K_D values of WT APE1. An overall comparison of K_D values between the 4-track and 5-track loop G4s with APE1 cysteine mutants showed similar binding affinities with no major changes.

Discussion

Oxidative damage at potential G4 sequences in some gene promoters is known to modulate transcription initiation via the BER pathway.^17,20 APE1, which is a crucial protein in this repair mechanism, can participate in the gene regulation process when a promoter G4 with an AP site is bound but not cleaved by APE1 allowing transcription factor recruitment. Similar to cysteine oxidation to sulfenic acids in OGG1 as a mediator of gene regulation,²⁸ APE1 is proposed here as another BER repair protein that modulates gene transcription via a similar PTM. In this study, we found that oxidation of APE1 had an effect on the catalytic function and binding of APE1 to the NEIL3 promoter G4 bearing an AP mimic F. APE1 was oxidized with CO₃^•− or H₂O₂, and APE1_OX was used in activity assays and binding assays with duplex and G4 substrates. Carbonate radical anion was used to oxidize APE1 as it is the predominant intracellular reactive oxidative species focusing damage to G-rich regions such as promoter PQSs.⁶⁰ APE1_OX was observed to give slightly diminished cleavage yields and weaker binding affinities with duplex DNA substrates as the oxidation time of the protein increased. In contrast, with F-core 4-track G4 and 5-track G4, APE1_OX showed lower cleavage yields and stronger binding with increasing oxidation time. These data suggest that oxidation of APE1 as a PTM has an effect on APE1 function on G4 substrates. In order to study the role of cysteine sulfenic acids in the functioning of APE1, we mapped and quantified sites of CO₃^•−-induced oxidation of cysteine residues by LC-MS/MS. APE1 exposed to CO₃^•− led to sulfenic acid formation at five cysteine residues with position dependency in the yields.

Cysteine sulfenic acid is a posttranslational modification in proteins that plays an important role in enzyme catalysis and regulating protein activity.^52,61 However, sulfenic acid is unstable and very reactive toward further oxidation in solution, and its detection or quantification requires the presence of a trapping agent such as dimedone.⁵² The oxidation products of APE1 in the presence and absence of dimedone revealed the cysteine thiol oxidation pathways followed when APE1 is oxidized. In the presence of dimedone, five sulfenic acid-forming cysteines (C65, C93, C208, C296 and C310) out of the seven APE1 cysteines were discovered to be oxidized, and the yields were quantified. In a mouse model study, the equivalent murine cysteines to the five human cysteines we identified here in our mass spectrometry study were found to form sulfenic acid under chronic and acute oxidative stress.⁵⁵

When APE1 was oxidized with CO₃^•−, a disulfide bond between C65 and C93 was manifested. In the absence of dimedone under our experimental conditions with CO₃^•−, C65, C93, C208, C296 and C310 resulted in sulfinic and sulfonic acids at these sites. No disulfide bonds were observed in the absence of dimedone. A prior study has investigated the redox activity and disulfide bond formation in APE1 in an attempt to elucidate the redox mechanism of the protein.³⁸ The formation of disulfide bonds via oxidation of APE1 by H₂O₂ resulted in intramolecular disulfide bonds between the APE1 cysteines and an intermolecular disulfide bond between C99 and thioredoxin. The numerous cysteine residues in APE1 that formed intramolecular disulfide bonds as a time-dependent cascade reaction further support that APE1 is a redox factor with unique properties. As sulfenic acid is another form of reversible thiol oxidation, our study focuses on cysteine oxidation to sulfenic acid in which APE1 variants were made by using serine as the sulfenic acid mimic. Recognizing that serine may not be the most suitable mimic for sulfenic acid we also made cysteine to aspartic acid APE1 mutants. However, the cleavage yields and binding affinities with Cys to Asp APE1 variants were very similar to the results obtained with Cys to Ser APE1 variants (Figure S14).

In order to study the impact of these Cys to Ser variants on APE1 function and binding, synthetic DNA strands of the NEIL3 promoter PQS sequence with a tetrahydrofuran analog (F) of an AP site positioned in a core or a loop of the G4 were used. For further comparison, both 4-track and 5-track G4s were studied to address whether the 5^th track, or “spare tire,” was important for these features of APE1. Supporting prior studies, WT APE1 cleavage of NEIL3 promoter G4 substrates is lower than for duplex substrates.²⁰ The Cys to Ser APE1 variants show lower cleavage in the F-core G4 substrates than the WT APE1; on the other hand, APE1 mutant cleavage yields in F-loop G4 are similar to the WT APE1. The Cys to Ser variants of APE1 can be generalized into three main groups based on endonuclease activity. The first group consists of APE1 variants C65S and C208S that were the same as WT APE1, i.e. no effect on the cleavage yields, which is consistent with the position of the cysteine residue in the protein structure being distant from the active site.^40,41 The second group is made of C310S whose activity was compromised on all substrates, which can be explained by the position of C310 adjacent to the enzyme active site.⁴⁰ The third group comprised of C93S and C296S showed activity dependent on the substrate. Their activity was either the same as WT APE1 or they displayed compromised activity. The structure of APE1 does not lead to an explanation for the substrate structure dependency in the activities of these two variants. We note that a prior study found the C99S mutant of APE1 demonstrated reduced binding and cleavage activity on duplex DNA substrates;⁴⁵ however, in our hands, C99 was not a readily oxidized site so the importance of a PTM at this residue is difficult to rationalize.

To determine whether the activity differences measured were a result of loss in substrate binding, DNA-protein binding was studied by fluorescence anisotropy. WT APE1 shows stronger binding to duplex substrates than to G4 substrates when Sm³⁺ was present. All cysteine mutants, especially C310S, displayed weaker binding with duplex DNA substrates compared to WT APE1. To our surprise, the Cys to Ser variants showed tighter binding to all G4 DNA substrates. These data support the concept that APE1 can tightly bind AP-containing gene promoter G4s without cleaving them. Even more intriguing are the C65S and C310S mutants that show extremely tight binding with F-core G4 compared to WT APE1 binding, which is even more pronounced in the 5-track G4s compared to 4-track G4s. As expected, when F was positioned in the loop both 4-track and 5-track G4s show similar binding affinities with all cysteine to serine variants compared to WT APE1. These data support the hypothesis that cysteine oxidation to sulfenic acid in APE1 could be a trigger for APE1 to engage in gene regulation during oxidative stress by increasing G4 binding while decreasing enzyme cleavage activity. After the stress is lowered, the sulfenic acid is readily reduced back to a cysteine thiol allowing APE1 to function as an endonuclease in BER (Figure 7).

Figure 7: Proposed mechanism for cysteine oxidation to sulfenic acid triggering the reciprocal function of APE1 in a NEIL3 promoter PQS in ds vs. G4 DNA contexts.

Among the APE1 Cys to Ser variants, C65S and C310S are particularly noteworthy. They both display especially tighter binding to F-core 5-track G4 which is the most relevant to the complete promoter sequence. About half of promoter G4 sequences possess a 5^th track, and this number is higher among the 191 genes coding for proteins involved in DNA repair; thus, the mechanism could extend well beyond the example of NEIL3 demonstrated here.¹² Among the seven cysteine residues in APE1, C65 and C310 have been implicated to play a role in the protein’s redox activity.^30,38,45,46 This work shows that the Cys to Ser variant of C65 is an excellent G4 binder with reduced nuclease activity. Also, C310S is another Cys to Ser variant with very strong G4 binding and no endonuclease activity. These two sites of cysteine to sulfenic acid oxidation have been found in mouse models, speaking to their biological relevance.⁵⁵

PTMs in APE1 have the ability to drive the protein down either the DNA repair or the gene regulation pathway. Acetylation of lysine residues in APE1 is also an example of PTMs being involved in the transcriptional regulation function of the protein. After binding the AP-containing DNA substrate, APE1 is acetylated at multiple N-terminal Lys residues and acetylated APE1 apparently stabilizes the G4 structure in cells while also increasing the residence time of APE1 on the G4 structure to coordinate transcription factors and to act therefore as a gene expression modulator.²¹ Nitrosylation at C93 and C310 is yet another PTM that controls nuclear export of APE1 in cells and is suggested to regulate DNA repair by APE1.^62,63 Phosphorylation of serine and threonine residues is also an important PTM that was found to completely abolish APE1 repair activity suggesting a possible role to switch the dual function of APE1 from repair to gene regulation.^64,65 Cysteine oxidation to sulfenic acid as described in the present studies is yet another PTM that had not been thoroughly explored; we provide evidence in support of the hypothesis that this modification could drive APE1 down the functional pathway of gene regulation.

Conclusion

When certain gene promoters containing potential G-quadruplex sequences are oxidized, the BER pathway apparently guides APE1 to this region as an enhancer of transcription initiation. Previous studies, including cell-based assays, have proposed that an AP-containing duplex that is a PQS can refold into a G4 to present an AP site to APE1 for binding but not cleavage.^23,60 Some aspects of this binding and regulation by APE1 were previously known,¹⁷ but more details remained to be revealed. The present work investigated how cysteine oxidation to sulfenic acid in five different sites in the protein may play a role in stalling APE1 on G4 substrates via tight binding but without endonuclease cleavage.

Five out of the seven cysteines in APE1 were found to be more prone to oxidation forming sulfenic acid. The activity and binding assays with the cysteine APE1 mutants mimicking the sulfenic acid modification conducted with DNA sequences from the NEIL3 promoter PQS showed lower cleavage yields but stronger binding affinities with G4 substrates. The APE1 C65S and C310S variants were intriguing due to their low to undetectable cleavage yet tighter binding with G4 substrates. However, C310S showed lower cleavage and lower binding with duplex substrates whereas C65S displayed similar cleavage and binding affinities as WT APE1. Thus, C310S may mimic the sulfenic acid oxidation state of APE1 in which the enzyme will selectively bind G4 substrates tightly via a separate domain with no endonuclease activity. Such a structure has recently been proposed in which the N-terminal disordered domain of APE1 is a possible G4 binding region.⁵⁷ These data provide a survey on how different cysteines are oxidized under oxidative stress and which sulfenic acids of APE1 may have an impact on G4 substrates to facilitate stronger binding with greatly reduced endonuclease activity in order to regulate gene expression.

Materials and methods

Expression and purification of recombinant APE1

Human WT APE1 was expressed from the pET28HIS-hAPE1 plasmid obtained from Addgene (Plasmid #70757).⁶⁶ Site-directed mutagenesis with the Q5 Site-Directed Mutagenesis Kit (NEB) was used to create the following seven APE1 variants: C65S, C93S, C99S, C138S, C208S, C296S, and C310S. The isotopically labeled WT ¹⁵N-APE1 protein was expressed in media containing ¹⁵NH₄Cl following a literature protocol.⁶⁷ The proteins were expressed in BL21(DE3) competent E. coli cells (NEB), grown at 37 °C until they were at an OD_600nm of 0.6 and then induced with 100 μM IPTG, and grown overnight at 37 °C. After harvesting the cells by centrifugation, they were lysed by sonication in a lysis buffer comprised of 20 mM sodium phosphate (pH 7.4), 300 mM sodium chloride, 1 mM PMSF, and 5 mM BME. The lysate was pelleted at 18 000 × g for 30 min. The resulting supernatant was passed through HisPur^™ Ni-NTA resin (Thermo Fisher Scientific) equilibrated with the lysis buffer. The protein was eluted from the Ni-NTA column with a linear gradient of aqueous imidazole from 10 mM up to 250 mM. The protein eluted at high imidazole concentration was buffer-exchanged into 10 mM sodium phosphate (pH 7.4), 50 mM NaCl, 1 mM DTT, and 50% glycerol. The resulting protein was stored at −80 °C. Final concentrations were determined by the Bradford assay. The purity of the proteins was determined by denaturing SDS-PAGE.

APE1 oxidation and protein digestion for LC-MS/MS

WT APE1 (1 mg) in the storage buffer was initially reduced with 10 mM DTT for 1 h at 4 ° C. After the reaction, the DTT was removed with a NAP-5 column following the manufacturer’s protocol (GE Healthcare Life Sciences). Next, the reduced WT APE1 (25 μM) was oxidized and simultaneously treated with 10 mM dimedone (DMD) by constant shaking for 2 h at 37 °C to trap sulfenic acids as they are formed. The oxidation was conducted by treating the protein with 30 mM NaHCO₃ and 10 mM SIN-1 to generate CO₃^•− for 2 h at 37 °C, or with 50 μM H₂O₂ for 1 h at 37 °C. The protein was purified by methanol/chloroform precipitation and then lyophilized down to a 200-μL volume. The protein was unfolded by adding a 100 μL of 6 M urea in 100 mM Tris buffer (pH 7.8) followed by vortexing the sample to homogeneity and then sonication for 2 min. Protein samples were then treated with 20 μL of 200 mM iodoacetamide in 100 mM Tris buffer (pH 7.8), vortexed and incubated for 1 h at RT. The samples were diluted and mixed with 775 μL of deionized water. Protein samples were digested using trypsin protease (Thermo Scientific^™ Pierce^™, MS grade) to a ratio of 50:1 at 37 °C overnight following the manufacturer’s protocol.

The digested samples were analyzed by an ACQUITY UPLC I-Class coupled with a Xevo G2-S QToF mass spectrometer (Waters, Milford, MA). The digested peptides were separated by an ACQUITY UPLC peptide CSH C18 column (2.1 mm x 100 mm, 1.7 μm particle size; Waters, Milford, MA). A linear gradient (5% – 40% B) over 56 min at a flow rate of 0.2 mL/min over 56 minutes while maintaining the column at 40 °C was used for the UPLC run. The mobile phase A was ddH₂O with 0.1% formic acid and mobile phase B was acetonitrile with 0.1% formic acid. The LC-MS/MS data were acquired by operating the mass spectrometer at MSE mode (acquisition mass range: 100 - 2000 m/z). A collision energy ramp was set between 15 V and 40 V to fragment peptides. The ESI source was maintained at 150 °C. The capillary voltage was 3 kV and the sampling cone voltage was 30 V. The cone gas flow was 10 L/h and the desolvation gas flow and temperature were 800 L/h at 500 °C. A mass signal (m/z: 556.277) from a continuously infused leucine enkephalin standard through the lockmass channel was used to provide the external mass calibration.

LC-MS/MS data analysis

The acquired LC-MS/MS data were processed using BiopharmaLynx software (Waters, Milford, MA). In the data processing method, trypsin was selected for the digestion reagent and the maximum number of missed cleavages was one. The MS mass match tolerance and MSE match tolerance were both 30 ppm. The DMD modification representing sulfenic acid (+ 138.068 Da), sulfonic acid oxidation (+ 47.985 Da), sulfinic acid oxidation (+ 31.990 Da), and carbamidomethylation of cysteine were set as variable modifications. The disulfide-containing peptide ions were further manually inspected.

Thermal melting analysis of APE1

The thermal melting (T_m) values were determined on samples of 15 μM protein in buffered solutions with 50 mM sodium phosphate buffer, pH 7.5 at 25 °C. The melting experiments were initiated by thermally equilibrating the samples at 20 °C for 10 min followed by heating at 0.5 °C/min and equilibrating at each 1 °C increment for 1 min. Readings at 280 nm were taken after each 1 °C change in the temperature from 20 to 100 °C. Plots of absorbance at 280 nm versus temperature were constructed, and the T_m values were determined by a two-point analysis protocol using the instrument’s software.

Oligomer preparation

All oligomers were synthesized and deprotected by the DNA/Peptide core facility of University of Utah following standard protocols. The site-specific introduction of an AP analog was conducted by incorporating a tetrahydrofuran (F) phosphoramidite at the desired site in the DNA sequence. The crude oligomers were purified using a semi-preparative, anion-exchange HPLC column running a mobile phase system consisting of A (1 M lithium chloride, 20 mM lithium acetate at pH 7 in 1:9 acetonitrile: distilled water) and B (1:9 acetonitrile: distilled water). The method was initiated at 20% B and increased to 100% B via a linear gradient over 30 min with a flow rate of 3 mL/min while monitoring absorbance at 260 nm. The purified samples were dialyzed against ddH₂O for 18 h to remove the purification salts and then lyophilized to dryness followed by resuspension in ddH₂O. The concentrations of the stock samples were determined by measuring the absorbance at 260 nm, and using the nearest neighbor approximation for the extinction coefficient. The extinction coefficients for these oligomers were estimated by omitting a nucleotide for F.

DNA substrate preparation for APE1 activity assays

Oligomers containing F were radiolabeled with ³²P-ATP at the 5‘ end by T4-polynucleotide kinase from NEB following literature methods.²⁰ The radiolabeled oligomers were purified using PD Spin-Trap^™ G-25 columns. Substrates contain a 2:3 molar ratio of hot oligomers to cold oligomers. In order to form G4 folds, the F-containing oligomers were annealed in APE1 activity buffer (20 mM Tris (pH 7.5 at 37 °C), 50 mM KOAc, 10 mM Mg(OAc)₂ and 1 mM DTT) by heating them to 90 °C for 5 min followed by gradual cooling to room temperature over 18 h. To make duplex DNA, a 1.5-fold excess of the complementary strand to the F-containing strand were annealed together in (1x) APE1 activity buffer.

APE1 activity assays

The activity assays with the oxidized WT APE1 (APE1_OX) protein were conducted by oxidizing WT APE1 with CO₃^•− generated from 30 mM NaHCO₃ plus 10 mM SIN-1 for 15 min or 60 min at 37 °C, followed by the addition of APE1_OX (3 nM) to 10 nM of pre-annealed substrate DNA in a 10 μL volume. The activity assays with Cys to Ser APE1 variants were conducted by adding APE1 (3 nM) to 10 nM of pre-annealed substrate DNA in a 10 μL volume. Enzyme and substrate were incubated at 37 °C for 60 min. The reactions were terminated by adding an equal volume (10 μL) of stop buffer (95% formamide, 10 mM EDTA, 10 mM NaOH, 0.1% bromophenol blue, 0.1% xylene cyanol, 5% glycerol) followed by heat denaturing the mixture at 65 °C for 20 min. Assay mixtures without enzymes were used as negative controls. The samples were then analyzed using a 20% urea-denaturing PAGE. The gels were exposed on phosphorimager screens for 18 h and the bands were scanned with a Typhoon^™ 9400 Variable Mode Imager (GE Amersham Biosciences) and quantified using ImageQuant^™ Image Analysis Software. The yield percentage was calculated by dividing the band intensity of the product (cleaved strand) by the summed band intensity of the product and the substrate (full-length strand). Error bars indicate standard deviation of three independent experiments.

DNA substrate preparation for fluorescence anisotropy binding assays

Modified oligomers containing F were 5‘ labeled with the fluorescein isomer 6-carboxyfluorescein (6-FAM) during the solid-phase synthesis of the strands. Labeled-DNA substrates folded into G4s were annealed in the APE1 binding buffer (20 mM Tris (pH 7.5 at 22 °C), 50 mM KOAc, and 1 mM DTT) and the concentration was determined by absorbance at 260 nm. To make duplex DNA, a 1.5-fold excess of the complementary strand to the F-containing strand was annealed together.

Fluorescence anisotropy binding assays

Fluorescence anisotropy measurements were used to quantify binding of WT APE1 and the seven cysteine variants to the FAM-labeled DNA substrate described above. The proteins were pre-incubated with 1 mM EDTA for 30 min at 22 °C. The binding assays with APE1_OX protein were conducted by oxidizing WT APE1 with CO₃^•− generated from 30 mM NaHCO₃ plus 10 mM SIN-1 for 15 min or 60 min at 37 °C, followed by the addition of APE1_OX to 5‘-FAM-labeled pre annealed substrate DNA. The binding assays were carried out with 50 nM 5‘-FAM-labeled DNA titrated with a concentration series of APE1 ranging from 0 to 5000 nM incubated 30 min at 22 °C in 20 mM Tris (pH 7.5 at RT), 50 mM KOAc, and 1 mM DTT buffer in 10 mM samarium(III) acetate.

Fluorescence anisotropy measurements were carried out on a BioTek Synergy2 Multi-Mode Microplate Reader. The excitation and emission wavelengths were 485 and 520 nm, respectively. The anisotropy values (r) were calculated with the equation mentioned below where I_par is the parallel emission intensity and I_per is the perpendicular emission intensity.⁶⁸

$$r=\frac{I_{par}-I_{per}}{I_{par}+2I_{per}}$$

The r values obtained were plotted against the log[APE1] to produce sigmoidal curves that were fit to the following Hill equation where bottom is the lowest r value and top is the highest value of the sigmoidal curve. The dissociation-binding constant is K_D and n is the Hill coefficient.⁶⁹

$$r=bottom+\frac{(top-bottom)}{1+(\frac{{10}^{loglog{K}_D}}{{10}^{log[APE1]}}{)}^n}$$

Error bars indicate standard deviation of three independent experiments.