Switchable DNA‐Based Peroxidases Controlled by a Chaotropic Ion

Abstract Here we demonstrate a switchable DNA electron‐transfer catalyst, enabled by selective destabilization of secondary structure by the denaturant, perchlorate. The system is comprised of two strands, one of which can be selectively switched between a G‐quadruplex and duplex or single‐stranded conformations. In the G‐quadruplex state, it binds hemin, enabling peroxidase activity. This switching ability arises from our finding that perchlorate, a chaotropic Hofmeister ion, selectively destabilizes duplex over G‐quadruplex DNA. By varying perchlorate concentration, we show that the DNA structure can be switched between states that do and do not catalyze electron‐transfer catalysis. State switching can be achieved in three ways: thermally, by dilution, or by concentration.


Introduction
Biomolecules are highly attractive as potential programmable electron transfer catalysts and bioelectronics. A range of techniques have been employed, including microbial systems, [1] protein-directed assembly of metal clusters, [2] metallated base pairs, [3] biopolymer-directed assembly of nanoparticles, [4,5] and minimal peptides. [6] With 0.34 nm spacing between nucleobases and exquisite, atomic-precision self-assembly directed by nucleobase hydrogen bonding, nucleic acids are an especially attractive means by which to develop programmable electron transfer catalysts. One means by which DNA can perform multiple turnover electron transfer is by employing a prosthetic group, [7] as many enzymes that perform electron transfer in nature do. [8] Hemin, one such prosthetic group, binds selectively to a noncanonical structure formed by G-rich sequences of DNA, termed a G-quadruplex, which activates it to perform electron transfer. [7,9,10] By binding hemin, the same prosthetic group employed by natural redox enzymes, such as peroxidases and cytochrome P450s, DNA can perform electron transfer in a biomimetic fashion while mitigating the oxidative damage issues associated with attempting to use it as a classical conductor. [11] Sequences of DNA and RNA that can exhibit either Gquadruplex or non-G-quadruplex (e. g., unpaired or Watson-Crick base paired) structures depending on context are a ubiquitous feature of life. These sequences are actively remodeled in living organisms. [12][13][14] A bioinspired chemical system based on these phenomena that enabled reversible, programmable structure switching would afford a powerful tool for dynamic electron transfer behavior in DNA nanostructures. We thus sought to develop a chemical system to switch the same DNA between secondary structure states, reasoning that this would combine the speed and repeated reversibility of pHswitchable DNA nanomotors with the compatibility observed in static pH strand-exchange based systems. [15][16][17][18][19][20] Base pairs have ca. one half the solvent-buried hydrophobic surface area of G-quartets, and ions exert Hofmeister effects by interactions with the surface of biopolymers. [21] Similarly, Gquartets coordinate dehydrated ions, and high-salt solutions influence biopolymer folding by osmotic effects. [21,22] We thus reasoned that a DNA duplex and G-quadruplex would exhibit differential destabilization by chaotropes. Here, we demonstrate that perchlorate is a selective denaturant for duplex vs. quadruplex DNA. We have exploited this phenomenon to develop a minimal electron transfer catalyst made of DNA that can be switched between three states: a duplex, a Gquadruplex, and a single-stranded state. We show that this switching can be performed thermally, by dilution, or by concentration. We show that the DNA structure can be switched over 100 times without degradation, and that it can perform multiple-turnover electron transfer catalysis by binding hemin and catalyzing electron transfer in the G-quadruplex state ( Figure 1).

Results
We designed two FRET reporter systems, G4-Dark and Duplex-Dark ( Figure 2a and Table 1), to allow readout of their folding state. Both reporter systems were of the same length (24 nt) and sequence but differed in fluorophore and quencher placement. G4-Dark was comprised of equimolar amounts of Fluorescein-G4-Quencher, a DNA sequence that could form a G-quadruplex, and G4Comp, its Watson-Crick complement. Fluorescein-G4-Quencher was 5'-labeled with a fluorescein tag and 3'-labeled with a quencher. Thus, this system would exhibit fluorescence when Fluorescein-G4-Quencher was either unfolded or hybridized to G4Comp. When Fluorescein-G4-Quencher was folded into a G-quadruplex, fluorophore and quencher would be brought into spatial proximity and quenched.
Duplex-Dark was comprised of equimolar amounts of strands of the same sequences as we employed in G4-Dark with rearranged reporters. Fluorescein-G4, a 5'-fluorophorelabeled sequence that could form a G-quadruplex, and G4Comp-Quencher, a 3' quencher-labeled sequence. In this system, the fluorescein reporter exhibits fluorescence when Fluorescein-G4 is either folded into a G-quadruplex or is unfolded. Upon hybridization of the two strands to form a double helix, this system would quench.
These strands followed the Hofmeister trend of stability (sulfate > chloride > nitrate > perchlorate). (Supporting Information Figure S1). Both G-quadruplex and duplex structures were increasingly destabilized as chaotropicity of the salts increased; however, duplex was much more susceptible than the Gquadruplex, particularly in sodium perchlorate.
We next examined the thermal response of the Duplex-Dark system in LiClO 4 solution (conditions in which only duplex and unfolded states are possible, due to the stability trend in Gquadruplexes (K + > Na + ⋙Li + ) [22] (Table 2, Figure 2b, Supporting Information Figures S2-7). [22] At 100 mM LiClO 4 , Duplex-Dark exhibited fluorescence response consistent with a doublestranded to unfolded structural transition at 79.8°C (Duplex-Dark possesses spatially adjacent reporters in the duplex state, resulting in a systematic slight elevation of T M relative to other sequence-related systems with differing tags). As the salt concentration was increased, the thermal midpoint initially increased to a maximum of 84.7°C in 0.5 M LiClO 4 , above which it decreased to a minimum of 50.1°C in 4 M LiClO 4 . This is consistent with electrostatic stabilization effects being predominant at lower salt concentrations and Hofmeister ion effects at higher concentration, as has been previously observed. [23]  ; and at high temperatures, it is comprised of two single strands (right). At high perchlorate concentrations, it exists in only two states -the G-quadruplex/ single stranded complement state (center) and a two-single-strand (right) state. The catalyst can also be switched between two states by varying concentration (panel b). In high-perchlorate solution, the device exists as a G-quadruplex, while in low-perchlorate solution, it forms a duplex. Thus, by diluting a high-salt solution of DNA, the DNA can be switched from a Gquadruplex to a duplex; and by removal of water from a low-salt solution of DNA, the DNA structure can be switched from a duplex to a G-quadruplex. In each case, the G-quadruplex is competent to catalyze multiple-turnover electron transfer in the presence of hemin (panel c and Figure 4).   Figures S8-18). In all NaClO 4 -containing solutions, Fluorescein-G4-Quencher formed a G-quadruplex, giving a 65.1°C thermal midpoint in 100 mM NaClO 4 ; this temperature increased with increasing NaClO 4 , consistent with the electrostatic screening afforded by Na + as well as its binding to the central channel within the G-quadruplex. Between 1-4 M NaClO 4 , the G-quadruplex was so stable it was not fully denatured, even at 95°C (representative melting curve of 4 M NaClO 4 in Figure 2c). In 5 M NaClO 4 and above, thermal midpoints were measurable and decreased with increasing As a result of this differential stability, G4-Dark (panel e) and Duplex-Dark (panel f) can be switched thermally between duplex, G-quadruplex, and single-stranded states, and the temperatures at which these transitions occur can be tuned by varying the concentration of NaClO 4 . In low perchlorate (0.1 M), only the duplex-to-single stranded transition is observed (red circles). In intermediate perchlorate (4 M), the DNA structure transitions between duplex (at low temperature), G-quadruplex (at intermediate temperature), and single-stranded (at high temperature) states (blue squares). In high perchlorate (saturated, ca. 9.5 M), the DNA structure exists only in the G-quadruplex and single-stranded states (black hourglasses). 73.7 duplex n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a G4-Dark (Fluorescein-G4-Quencher and G4Comp)/ NaClO 4

G4
none none high [ n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a duplex 75. perchlorate but remained high. Even in saturated NaClO 4 (ca. 9.5 M and containing less than three water molecules per ion), Fluorescein-G4-Quencher exhibited a higher thermal midpoint (73.7°C) than in 100 mM NaClO 4 (65.1°C). The destabilization of the DNA duplex in G4-Dark and Duplex-Dark was marked (Figure 2d) and exhibited a nearlinear response in thermal midpoint of À 7°C/M NaClO 4 between 1 and 9 M (r 2 = 0.97); these duplexes exhibited thermal midpoints of ca. 75°C at 100 mM NaClO 4 and ca. 27-30°C at 7-8 M NaClO 4 . In contrast, perchlorate-induced destabilization of the G-quadruplex formed by the G-rich strands was less pronounced and/or more than compensated by the presence of additional sodium (Figure 2d). Fluorescein-G4-Quencher exhibited a thermal midpoint of 65.1°C at 100 mM NaClO 4 and 73.7°C in saturated sodium perchlorate (ca. 9.5 M). That is, in a polymorphic sequence, one possible secondary structure (dsDNA) that is more stable in low perchlorate than an alternative fold (a G-quadruplex) becomes less stable than the alternative fold in high perchlorate.
Because of this, we speculated that G4-Dark could be thermally switched between all three states (duplex, G-quadruplex, and single-stranded) at intermediate perchlorate concentrations. (Figure 2e, Supporting Information Figures S19-29). Consistent with this, we observed a single transition at low (0.1 M) NaClO 4 , which corresponded to the duplex-to-singlestrand transition. At high NaClO 4 (saturated/ca. 9.5 M), we also observed a single transition, which corresponded to the Gquadruplex to single-strand transition (Figure 2e), confirmed by the lack of thermal dequenching under these conditions with Duplex-Dark (Figure 2f).
We next sought to characterize the reversibility of cycling through the DNA structural states. Reversibility without degradation is essential to a switchable catalyst and a potential concern given some structure-switching nanodevices' propensity to exhibit degradation with repeated switching due to buildup of waste products. [16,18] To do so, we thermally cycled a sample of G4-Dark in 4 M NaClO 4 (conditions in which all three states are thermally accessible) 100 times while monitoring fluorescence. The sample did not exhibit degradation during this experiment (Figure 3a-d).
Given the results from Figure 2, we reasoned that it would be possible to switch the state of our catalyst from Gquadruplex to duplex (by diluting it with aqueous buffer, lowering the concentration of perchlorate) or from duplex to Gquadruplex (by removing water under vacuum, increasing the concentration of perchlorate). To do so, we performed dilutions of high-salt solutions of G4-Dark (which would initially exist in its dark state and transition to its light state) and Duplex-Dark (which would initially exist in its light state and transition to its dark state). G4-Dark recovered fluorescence upon dilution from 8 M to 0.8 M, and Duplex-Dark lost fluorescence following the same dilution (Figure 3e). Conversely, we sought to switch the system by removal of solvent. To do so, we took samples with an initial volume of 200 μL and initial concentration of 0.8 M NaClO 4 and placed them in a vacuum chamber. We monitored these samples by fluorescence imaging with a custom-built device (Supporting Information Figures S30-37). The samples, initially in low NaClO 4 , behaved as expected for the duplex state, with G4-Dark exhibiting fluorescence and Duplex-Dark in a dark state. Duplex-Dark remained dark during concentration while the fluorescence of G4-Dark gradually increased as the solution became more concentrated. Finally, both solutions reached a critical concentration of NaClO 4 at which they transitioned to the G-quadruplex state: Duplex-Dark became fluorescent and G4-Dark became nonfluorescent (Figure 3f-i, Supporting Information Videos S1 and 2, Supporting Information Code S1-3). Increases in strand concentration, as would be expected to occur concomitantly with concentration of salt, impacted the bimolecular duplex stability only slightly (Supporting Information Table S1).
G4-Dark and Duplex-Dark demonstrate the perchloratebased switchability of our system. To demonstrate the ability of G-quadruplex/duplex equilibria to enable switchable electron transfer, we constructed G4-SwitchR (switchable redox; Table 1), which was comprised of G4Redox (a sequence previously shown to form a highly active peroxidase in complex with hemin), [24] G4Redox-Comp, and the cofactor hemin. This Gquadruplex was shorter (18 nt) than the FRET systems used, but it behaved similarly in perchlorate solutions. We employed Amplex Red, a nonfluorescent dye that is oxidized to the fluorescent, red-colored pigment resorufin catalytically by the peroxidase-mimicking DNAzyme formed between a G-quadruplex and hemin (Figure 4a-c). [25] We compared the ratio of the rate of electron transfer performed by the G-quadruplex/hemin complex to that performed by hemin alone (Figure 4d Table S2, Supporting Information Figures S38). In saturated NaClO 4 , electron transfer is suppressed due to lack of hemin-Gquadruplex interactions.
In 0.1 M NaClO 4 , G4-SwitchR exists entirely as a duplex, hemin binding is thus abrogated (Supporting Information Figure S39); resorufin was produced at < 30 nM/min when G4-SwitchR was present or when it was absent (Figure 4f). In 5 M NaClO 4 , G4-SwitchR dissociates into G4Redox folded into a Gquadruplex, which is competent to bind hemin (Supporting Information Figure S40), and G4Redox-Comp exists as a single strand. The resulting hemin-G4Redox complex performs electron transfer (as measured by the reporter dye Amplex Red's conversion to resorufin) 35-fold more rapidly than hemin alone (Supporting Information Figure S38): the hemin-G4Redox complex produced resorufin at 300 nM/min vs. unbound hemin, which produced it at 9 nM/min. At still higher NaClO 4 , hemin-Gquadruplex binding decreases (as indicated by a decreased Soret peak, Supporting Information Figure S39), and in saturated (ca. 9.5 M) NaClO 4 , resorufin was produced at a much lower rate of 46 nM/min. Hemin remained in solution at this concentration of NaClO 4 (Supporting Information Figure S41). Consistent with the propensity of hemin to aggregate, [26] partial loss of hemin was observed on extended centrifugation. However, this occurred in both low-and high-perchlorate solutions, indicating this is not a perchlorate-induced phenom-enon. G-quadruplex-dependent Soret absorbance for hemin was observed (Supporting Information Figure S39), demonstrating G-quadruplex/hemin interactions. Additionally, heme dimers exhibit extremely low electron transfer catalysis behavior, consistent with G-quadruplex hemin being the active species. [27] Thus, perchlorate concentration can be varied to switch the state of the active G-quadruplex/hemin form of the catalyst, as well as to modulate the reaction rate.
In order to demonstrate that thermally switching DNA conformation could also switch electron transfer catalysis, we measured Amplex Red oxidation using G4-SwitchR in 4 M NaClO 4 (Figure 4g). At 20°C, electron transfer was near-zero. As the sample was heated and switched from duplex to Gquadruplex, the rate of electron transfer exhibited a significant increase, reaching a maximum at 60°C and again decreasing as the G-quadruplex unfolded. This is consistent with the thermal denaturation curve generated by monitoring A 295 (a diagnostic wavelength for G-quadruplex formation) of G4-SwitchR (Supporting Information Table S2, Supporting Information Figures S43-44). Notably, hydrogen peroxide was required. Perchlorate ion, despite its exceptionally high oxidation potential, did not suffice as an electron acceptor in this system.

Discussion
Here, we have demonstrated that duplex DNA is preferentially destabilized by perchlorate salts relative to G-quadruplex DNA, and that this enables solution conditions-selective formation of DNA-hemin complexes that can act as electron transfer catalysts. This phenomenon can be used to switch electron transfer behavior in three ways: 1) direct addition of perchlorate . Switchable electron transfer amplification behavior of G4-SwitchR-hemin complex. When folded into the G-quadruplex state, a solvent-exposed Gquartet (panel a) is present which is known to provide a tight binding pocket for the cofactor hemin (panel b); the complex formed by a G-quadruplex and hemin can catalyze electron transfer. We visualize this here using the colorimetric electron transfer probe Amplex Red (panel c, top), which is initially colorless and nonfluorescent and, when oxidized, converts to the red fluorescent pigment resorufin (panel c, bottom). Hemin itself has a minimal background rate of electron transfer that is suppressed by increasing NaClO 4 concentration (panel d). When G4-Redox is operated as the only oligonucleotide constituent (i. e., under conditions in which the strand will form only a G-quadruplex) in the presence of hemin, robust multiple-turnover electron transfer occurs in low (0.1 M, red circles) NaClO 4 solution, it is enhanced in moderate (5 M, blue squares) NaClO 4 solution, and it is suppressed in saturated NaClO 4 solution (black triangles) due to abrogation of hemin binding (panel e). When G4-SwitchR is operated in these same conditions, electron transfer is switched off in low (0.1 M) NaClO 4 solution due to duplex formation, switched on in moderate (5 M) NaClO 4 solution due to G-quadruplex formation with concomitant hemin binding, and switched off in saturated NaClO 4 solution due to abrogation of hemin binding (panel f). G4-SwitchR can also be switched thermally, exhibiting varying rate of reaction depending on its folding state. When ramped from 25 to 95°C in 5 M NaClO 4 solution, excess electron transfer (as measured by rate of resorufin production in excess of that observed in hemin alone) starts at near-zero as the DNA structure exists in duplex form, increases substantially between 50 and 70°C as the duplex melts and the G-quadruplex forms, and it decreases again by ca. 90 % between 70 and 90°C as the G-quadruplex unfolds and the singlestranded form of the DNA system forms (panel g). These results are consistent with UV-visible monitored thermal measurements of secondary structures formed (Supporting Information Figures S43-44). Error bar shading = S.E.M., n = 3 technical replicates. salts, 2) thermal switching, and 3) increasing perchlorate salt concentration by vacuum concentration.
While the results here relate to switching of secondary structure in high-salt solution, examples of secondary-structure remodeling are observed in life as well. Nucleic acids exhibit considerable polymorphism in biological systems, particularly in sequences that can form G-quadruplexes. Such systems are diverse and widespread in biology and extend to both DNA and RNA, including proto-oncogene promoters, [33] the expansion segments of rRNA of higher organisms, [34][35][36] the rDNA corresponding to those expansion segments, [37] and eukaryotic messenger RNAs. [12,13] Several proteins can remodel G-quadruplexes, and both energy-dependent (i. e., helicases) and energy-independent systems that can do so have been reported. [38,39] ATP-dependent helicases are known to unwind Gquadruplex structures, and the RNA-binding protein Lin28 has been shown to unfold G-quadruplexes without the requirement for ATP. [38] Given that such remodeling processes are also operative in nature and that hemin-G-quadruplex promoted electron transfer has been suggested as being physiologically relevant, we speculate that conditional hemin-G-quadruplex complexes are a means by which cells could conditionally enable electron transfer. [40] For example, ribosomes from the neurons of Alzheimer's Disease patients have been shown to contain more iron than those from healthy patients, and these ribosomes possess peroxidase activity. [41] G-quadruplexes may enable this phenomenon in vivo. Human ribosomes are known to be polymorphic, particularly in their G-rich expansion segments, some of which have been observed to lack electron density in EM maps, consistent with an equilibrium between multiple states. [34,42] Rarely, individual ribosomes clearly exhibit extended conformations consistent with Watson-Crick base pairing, but these same sequences possess exceptionally high G-quadruplex forming potential and form G-quadruplexes in cell-free in vitro experiments, which is consistent with polymorphism with one state capable of catalyzing electron transfer. [36,37] Beyond the high-salt solutions that are the focus of the present study, we suggest that conditionally folded Gquadruplexes in cells could collaborate with hemin, producing a means by which cells can perform conditional electron transfer that is analogous to the phenomenon we have employed in this work. In fact, recent work suggests that hemin-G-quadruplex associations occur in human cells. [40] Such a phenomenon could be exploited both by extant life or in synthetic biological systems, and we speculate this could have enabled conditionally active forms of prebiotic electron transfer catalysts, which have attracted intense interest in recent years. [6,43,44] Experimental Section Nucleic acids: Oligonucleotides were obtained from Integrated DNA Technologies (Coralville, IA)and used as received. Labeled strands were obtained with HPLC purification and unlabeled strands were obtained with standard desalting. DNA sample preparation: Oligonucleotides were annealed in a T100 thermal cycler (Bio-Rad) by incubating at 95°C for 2 minutes, then decreasing the temperature by 10°C steps and annealing at each step for 1 minute to a final temperature of 25°C. Except where otherwise specified, experiments were performed in 50 mM Li-HEPES, pH 7.4, with 1 μM of the duplexed or single stranded oligonucleotides.

Fluorescence-monitored experiments:
Fluorescence measurements were performed using a Gemini XS plate reader (Molecular Devices) or a Cary Eclipse (Varian Technologies) fluorometer equipped with a thermostated peltier holder (Agilent Technologies) when bidirectional temperature control was required. Excitation was performed at 495 nm and emission was monitored at 520 nm.
For thermal denaturation studies, the sample was ramped from 20°C to 95°C (two heat/cool cycles) at 5°C/min. Selected measurements were also performed with a slower 0.5°C/min ramp rate to mitigate hysteresis. Thermal midpoints are reported as a sigmoid fit of the transition(s) in the second heating trace of two temperature ramps.
Data from fluorescence melts were normalized by division, setting the highest RFU value to 1 and the lowest value to 0 for each state in a given system (quenched or dequenched/partially dequenched).

UV-Vis monitored experiments:
Thermal denaturation experiments were performed using a qCHANGER 6/Cary60 (Quantum Northwest) interfaced to a Cary 60 UV-Visible spectrophotometer (Agilent Technologies) using a custom ADL script developed by Quantum Northwest to collect full spectra at each temperature as described previously. Heat sinking for the Peltier device was provided by an EXT-440CU ambient liquid cooling system (Koolance). Duplex melting transitions were monitored using the 260 nm trace from these datasets and G-quadruplex melting temperatures were monitored using the 295 nm trace.
Amplex Red experiments (described further below) were monitored on a Cary 60 UV-vis (Agilent Technologies) or a SpectraMAX 340PC plate reader (Molecular Devices) with the PathCheck functionality enabled.
Fluorescence imaging under vacuum: Samples were pipetted into PCR tubes with the lids open, which were inserted into a 3Dprinted jig that was placed inside a black box. The jig was fabricated in PLA on a Prusa 3D printer. The box and its contents were placed in a vacuum desiccator (Supporting Information Figures S29-36) and the chamber was continuously evacuated with a diaphragm pump (Welch 2014B-01). The samples were excited with LEDs with emission centered at 462-465 nm (Item B01GDO9UNY, Amazon). The samples were imaged with a camera (Raspberry Pi Module V2, Amazon) fixed at 90°relative to the light sources. A 12.5 mm longpass filter (Schott OG 550, Edmund Optics) was affixed with polyvinyl acetate adhesive (Elmer's Glue-All, Costco Wholesale) directly in front of the camera lens to block excitation light and pass emitted light. The camera was interfaced to a Raspberry Pi B + (Amazon). A Python script controlled the Raspberry Pi's GPIO pins to illuminate samples at 1-minute intervals, collect images, and measured and plotted fluorescence intensity vs. time. Time-lapse movies were generated from captured images using FFmpeg. Electron transfer assay: Samples contained G4-SwitchR (i. e., G4Redox and G4Redox-Comp), G4Redox alone, or no oligonucleotide. G4Redox, when present, was at 5 μM and G4Comp, when present, was at 10 μM to ensure full duplex state at low perchlorate/temperature and that electron transfer catalysis observed was due to the G-quadruplex state of G4-SwitchR and not residual unduplexed G4Redox. The buffer used here was 5 mM sodium phosphate, pH 7.4 with varying sodium perchlorate concentrations. To initiate the reaction, hemin was added to a final concentration of 1 μM, hydrogen peroxide was added to a final concentration of 300 μM, and Amplex Red was added to a final concentration of 200 μM. Reactions were monitored by measuring the absorbance of resorufin at 570 nm. For the Cary 60 thermal activation-Amplex red assay, Amplex Red was added to a final concentration of 2 mM. All reactions contained 2 % DMSO, which came from the hemin and Amplex Red stocks.
Precipitation assay: Solutions consisted of 5 μM G4Redox and 1 μM hemin in 5 mM phosphate buffer pH 7.4 and 2 % DMSO with varying sodium perchlorate. UV-vis spectra were obtained from 230-500 nm at 0 minutes and 60 minutes. After 60 minutes, the samples were centrifuged at 13, 000 rpm for 10 minutes and 200 μL of supernatant was used to collect a spectrum.

Code Availability Statement
All computer code used during the current study is included in this published article (and its Supporting Information files).