8‑Oxo-7,8-dihydropurines as Building Blocks to Enhance the Selectivity of an RNA Aptamer for Aminoglycosides

Abstract

The use of nucleic acids as potential therapeutic tools, sensors, or biomaterials, among other applications, has dramatically increased. Among these, RNA aptamers are of interest due to an innate high specificity toward their cognate targets, which include small molecules, proteins, ions, or cells. In this work, we took advantage of the ability that 8-oxo-7,8-dihydroguanine (8-oxoG) has to participate in unique H-bonding interactions, and probed its use to increase/control the selectivity/affinity of aptamers of RNA and DNA. The chosen model is a 23-nt long RNA (Neo61/Neo1–5′-GGA CUG GGC GAG AAG UUU AGU CC) strand that folds into a pentaloop hairpin with a stretch of three G·U Wobble pairs within the stem, which is known to have affinity toward various aminoglycosides. 8-OxoG was incorporated at positions G6, G7, G10, G12, or G15, within aptamers composed of RNA, DNA, or 2’-OMe modified RNA. Their recognition was tested toward 9 small molecule targets with aminoglycoside (x8) or antibiotic (x1) backbones, and their affinities were measured via circular dichroism (CD). Isothermal titration calorimetry (ITC) was used to corroborate the use of CD as a reliable technique. It was determined that incorporation of 8-oxodG at position-12 within DNA (OG12-DNA) led to increased selectivity toward neomycin or ribostamycin (K _d ≈ 2.5 and 2.2 μM), displaying 1–2 orders of magnitude tighter binding compared to other targets. Furthermore, functionalization with 8-oxodG at position-6 (OG6-DNA) displayed increased selectivity toward neomycin or tobramycin, albeit with decreased affinities (K _d ≈ 46 and 53 μM). Interestingly, the canonical DNA aptamer also displayed 4–10 fold enhanced selectivity toward neomycin, ribostamycin, and gentamicin, compared to its RNA homologue. On the other hand, the corresponding RNA analogues containing 8-oxoG or other modifications, specifically 8-oxoinosine, inosine, 8-oxoadenosine, or uridine, resulted in a high level of promiscuity toward most aminoglycosides, with kanamycin and streptomycin generally exhibiting higher dissociation constants. The presence of 2’-OMe-modified ribose led to trends similar to those obtained with their corresponding canonical RNA constructs. From a structural perspective, all nucleobase modifications led to thermal destabilization of the aptamer (including the DNA analogues), while the presence of the 2’-OMe ribose modification resulted in increased thermal stability. Among the molecules tested, neomycin and ribostamycin induced significant structural changes (measured via CD) on aptamers of RNA or DNA. Changes in RNA included the formation of a new band with positive ellipticity (λ_max ∼ 285 nm), associated with glycosyl bond rotation along the G·U wobble pairs that presumably facilitates recognition. On the other hand, binding by canonical and OG12 DNA aptamers resulted in a B-to-A form transition, where the smaller major groove may serve to facilitate DNA-target interaction. Further structural data were obtained by carrying out structural probing assays in the presence of RNase A, T₁, or DNase I; which displayed varying degradation patterns and thus changes in secondary structure as a function of the position of 8-oxoG/8-oxodG and presence/absence of the small-molecule target. Overall, the results reported herein show that (1) the use of 8-oxodG within DNA increases aptamer selectivity toward neomycin and/or ribostamycin; (2) the presence of 8-oxoguanine can alter the function of RNA and DNA, which is of broad biological relevance; and (3) the introduction of 2’-OMe modifications does not affect the selectivity of the aptamers in this work. While it is early to predict how 8-oxoG will affect the selectivity of aptamers at large, this work provides a link between the structure and function of oxidized RNA.

Introduction

RNA and DNA have a wide array of functions and properties that make them attractive from a biological and artificial (biomolecules made in the test tube and with distinct functions/applications) standpoint. The number of approaches that use aptamers as tools to target molecules/biomolecules of interest has dramatically increased over the past decade, yet their full potential is still to be exploited. To increase chemical/enzymatic stability or enhance their affinity/selectivity, common strategies in aptamer development aim at modifying the nucleobase(s), the ribose, or the phosphate backbone. − Some interesting recent approaches in development and characterization include: (1) the use of spiegelmers (L-RNA or L-DNA aptamers , ), which retain selectivity while remaining unrecognizable to other biomolecules and evade biodegradation; (2) nucleobase functionalization with aromatic hydrophobic rings, which take advantage of protein shape recognition; , (3) the use of distinct/enhanced selection approaches; or (4) in the use of exonucleases for aptamer characterization; among many others. In this work, we decided to further establish the hypothesis that 8-oxo-7,8-dihydroguanine (8-oxoG) can be used to enhance or alter the selectivity of aptamers, as previously reported for the theophylline aptamer. The impact of 8-oxoG was also explored within RNA constructs containing 2’-OMe modifications as well as within their corresponding 2’-deoxy DNA analogues, which impart chemical/enzymatic stability. Functionalization of purine rings at the C8-position induces an anti → syn conformational change around the glycosidic bond that results in the exposure of a different H-bonding pattern (Figure A). This has implications for the overall structure of the RNA construct and can lead to distortion of its duplex content (observed on DNA) as well as affect the thermal stability of different structural motifs of RNA, among other possible implications. Changes that potentially induce differences in target recognition.

1.

(A) Structure and H-bonding pattern of 8-oxoG. (B) Conformational changes of the modified nucleobases and ribose rings that were used in this work: 8-oxoG, Inosine (I), 8-oxo-7,8-dihydroinosine (8-oxoI), uridine (U), 2’-OMe-8-oxoG, and 8-oxodG. (C) Structure of neomycin B and sequence of the chosen aptamer. (D) Interactions arising from recognition between the aptamer (positions colored in yellow and orange represent interactions along the G·U stretch, pink involves the loop, and blue represents the stem closing base pair) and the aminoglycoside target, as described through NMR contacts.

8-OxoG, as well as other C8-modified nucleosides, can participate in distinct H-bonding interactions that form supramolecular arrangements as possible materials. , In addition, its use as an advantageous/novel building block within aptamers of RNA is an approach that has shown promise to generate aptamers with distinct selectivity/affinity. , On a different note, understanding how this oxidatively generated modification affects the overall structure of RNA, and as a consequence its function and properties, is also of importance in a biological context, albeit a concept/topic that remains largely unexplored. To gain a better understanding of structural and functional changes arising from the incorporation of 8-oxoG, other chemical modifications were incorporated onto the aptamers (Figure B). To probe for H-bonding interactions, the model aptamer was modified with 8-oxo-7,8-dihydroinosine (8-oxoI), inosine, or 8-oxo-7,8-dihydroadenosine (8-oxoA), which serve to understand the impact of the exocyclic amine in both syn- and anti-conformations, as well as the carbonyl group at position-6. In addition, uridine was also used as a pyrimidine mimic of 8-oxoG, given its analogous H-bonding pattern. The impact of the ribose and duplex conformation was tested by using the analogous DNA-aptamers, including strands of DNA containing 8-oxodG. Lastly, as a step into generating aptamers with enhanced chemical and enzymatic stability, we explored the use of 2’-OMe chemical modifications to expand on the potential use of 8-oxoG within biocompatible aptamers. , Importantly, these efforts served to increase the chemical space of nucleotide derivatives, which is important to diversify the nature/scope of aptamer-target pairs that are currently available.

In this work, an aptamer that recognizes neomycin B (Neo61 previously identified in a selection process) was chosen as model (Figure C). , This represented a good construct to probe for increased selectivity/specificity given that (1) the binding motif with neomycin, established via NMR, involves three guanine nucleobases (G6, G10, and G15, Figure D); ,, (2) the aptamer is highly promiscuous and recognizes/interacts with various glycoside targets; and (3) its size makes it suitable for solid-phase synthesis. To put into context, neomycin aptamers of varying sequence have been reported (Figure S1) and include their presence within natural riboswitches, or with available crystal structures on models corresponding to the ribosomal aminoacyl-tRNA site (using various aminoglycosides as target molecules). , From an application standpoint, the (16S) and (18S) bacterial and human sections of the ribosome have been used in microarrays, and Neo61 has been conjugated with nanoparticles for the detection of neomycin and other aminoglycosides in milk.

Since this work describes the use of single-nucleotide mutations as a strategy to enhance/change the selectivity/function of the nucleic acid constructs, it is worth noting that canonical and modified nucleobases have been effectively employed to this end. Inosine has been reported to increase the selectivity of aptamers of DNA, specifically one that recognizes cocaine. Single point mutations (C → U or G → A) on a functional riboswitch can alter its selectivity toward different target molecules with significant affinity changes (>4 orders of magnitude). , Nucleotides that induce a conformational change around the glycosidic bond (A → 8-BrA) can be used to probe the reactivity of the twister ribozyme. Of particular importance to this report is the observation that mutating a neomycin aptamer can also lead to changes in recognition and has been used in the development of sensors. Nucleotide mutations can also alter the selectivity toward tobramycin, while preserving the overall aptamer structure. It is also important to note that the affinity of aptamers can depend on specific conditions, such as ionic content, including [NaCl].

Furthermore, the importance of studying aminoglycosides in different fields has been reviewed, and different chemistries have been developed to synthesize/derivatize the small molecules. Functionalization of the targets has also been carried out to tag RNA and track RNA binding. The structures of the target molecules that were probed in this work are shown in Figure , where penicillin and streptomycin were used as negative controls and were not expected to function as targets for any of the aptamers tested, given the stark structural differences compared to neomycin.

2.

Structure of target molecules that were used in this work. Highlighted regions correspond to structural differences with respect to neomycin B.

Methods

General

The procedures, along with full characterization, of all synthetic intermediates that yielded the phosphoramidites used in this work are included within the supporting materials (pp S4–S12, Figures S2–S6). All water used in the oligonucleotide work was RNase-free via treatment with diethyl pyrocarbonate.

RNA Synthesis

Oligonucleotides were synthesized on a 394 ABI DNA synthesizer using CPG supports and 2’-O-TBDMS- or 2’-OMe-phosphoramidites of U, A, C, G, as well as phosphoramidites of T, dA, dC, and dG (purchased from Glen Research or ChemGenes) (Table ). Universal supports (Glen Unysupport, Glen Research) were used for the syntheses of the 2’-OMe-modified aptamers. 0.25 M 5-Ethylthio-1H-tetrazole in acetonitrile was used as the coupling reagent; 3% trichloroacetic acid in dichloromethane was used for detritylation; a 2,6-lutidine/acetic anhydride solution was used for capping; and an iodine (0.02 M) in THF/pyridine/water solution was used in the oxidation step (also purchased from Glen Research). Coupling times of 10 min or 45 s were used in the syntheses of RNA or DNA, respectively. Oligonucleotides (ONs) were deacetylated/debenzoylated/deformylated and cleaved from the CPG support in the presence of 1:1 aq methylamine (40%) and aq ammonia (28–30%) with applied heat (60 °C, 1.5 h). A mixture of N-methylpyrrolidinone/triethylamine trihydrofluoride/triethylamine (3:2:1) was used for deprotection of the silyl groups (60 °C, 1.5 h), followed by purification via gel electrophoresis (20% denaturing PAGE). C18-Sep-Pak cartridges were used to desalt the oligonucleotide solutions using 5 mM NH₄OAc as the ion exchange buffer, followed by concentration under reduced pressure. Oligonucleotides were dissolved in H₂O and used as obtained for subsequent experiments. All oligonucleotide solutions were quantified via UV–vis and used without further purification.

1. Sequences Used in This Work .

^aUnderlined letters represent 2’OMe-modified nucleotides. 8-OxoG was used for RNA, while 8-oxodG was used for DNA.

RNA Characterization (MALDI-TOF)

Samples to obtain mass spectra for all the modified oligonucleotides were prepared, as described elsewhere (see Acknowledgments and Supporting Information, p. S13). Mass spectra of all ONs can be found within the supporting materials (pp S18–S186, Figures S10–S124).

UV–Vis Spectroscopy

Concentrations of all oligonucleotides were obtained by recording their UV–vis spectra at 60–90 °C on a PerkinElmer Lambda 365 UV–vis spectrophotometer using quartz cuvettes with a 1 cm path length.

Circular Dichroism (CD) Spectroscopy and Thermal Denaturation Transitions (T _m)

CD spectra were recorded at various temperatures (PTC-348W1 Peltier thermostat) by using Quartz cuvettes with a 1 cm path length. Spectra were averaged over three scans (350–200 nm, 0.5 nm intervals, 2 nm bandwidth, 1 s response time), and the background was corrected with the appropriate buffer. Solutions of RNA always contained sodium phosphate (10 mM, pH 7.2), NaCl (100 mM), and MgCl₂ (1 mM). All solutions were hybridized prior to recording spectra by heating to 90 °C, followed by slow cooling to RT (1–2 h). T _m values were recorded at 265 or 280 nm with a ramp of 1 °C/min and step size of 0.2 with temperature ranges from 4 to 95 °C. A thin layer of mineral oil was added on top of each solution to keep concentrations constant at higher temperatures. Origin 9.1 was used to determine all of the T _m values and to generate the corresponding figures. All experiments were carried out in triplicate.

Titrations with Small Molecules and Determination of Dissociation Constants (K _d)

Prepare two solutions as follows: (1) RNA (2–5 μM; 10 mM sodium phosphate-pH 7.2; 100 mM NaCl; 1 mM MgCl₂), and (2) buffer (10 mM sodium phosphate-pH 7.2; 100 mM NaCl; 1 mM MgCl₂). Apply heat (90 °C), followed by slow cooling to rt (1–2 h), and transfer a known volume (160 μL) into their corresponding cuvettes. Record the CD spectrum for each and carry out a buffer subtraction (Spectra Manager V2, JASCO). Add aliquots of small molecule (0.5–1 μL, concentration varied between 10 μM – 110 mM, dissolved in water), record the CD spectra, and carry out the corresponding buffer subtraction. Repeat this process (up to 10 times) without exceeding the addition of 7–10 μL in total volume. Subtract the spectrum without small-molecule from the spectrum after each addition; this allows for the generation of plots as ΔΦ (change in the dichroic signal) as a function of [small-molecule]. The selected wavelength for the recorded ΔΦ varied with each aptamer, since changes in CD spectra were distinct for each RNA/DNA-target pair; however, the majority displayed a recorded change at ∼210 nm. The dissociation constants were then obtained as previously reported, , fitting the obtained plots in the form of ΔΦ as a function of [small molecule] using the Hill analysis (also known as the Clark equation). , All experiments were carried out in duplicate or triplicate and are available within the SI (pp. S18–S186).

$$\begin{array}{ccc}\mathrm{Hill}-\mathrm{Langmuir}/\mathrm{Clark}\mathrm{equation}& {\mathrm{Fraction}}_{\mathrm{bound}}=\frac{[\mathbf{L}{]}^{\mathit{n}}}{{{\mathit{K}}_{\mathbf{d}}}^{\mathit{n}}+[\mathbf{L}{]}^{\mathit{n}}}& n=1\\ & \mathrm{L}=\mathrm{aminoglycoside}& \end{array}$$

The analysis was also corroborated by using a quadratic binding equation as an alternative to fit the same data as above, working under the assumption that [L]_free ≈ [L]_total. This analysis led to dissociation constants that yielded the same trends and comparable values as those obtained with the Hill analysis. All plots are also included within the SI.

$${\mathrm{Fraction}}_{\mathrm{bound}}=\frac{([\mathrm{L}]+[\mathrm{RNA}]+K_{\mathrm{d}})-\sqrt{([\mathrm{L}]+[\mathrm{RNA}]+K_{\mathrm{d}}{)}^2-4[\mathrm{L}][\mathrm{RNA}]}}{2[\mathrm{RNA}]}$$

To corroborate that equilibration times between the aptamer:target pairs were sufficient, binding affinities were obtained as a function of time. The data on a model system (OG12OMe:neomyicin) can be found within the SI (Figure S126, p. S187).

5′-³²P-Radiolabeling of Oligonucleotides of RNA

Oligonucleotides were radiolabeled by mixing polynucleotide kinase (PNK), PNK buffer, γ-³²P-ATP, RNA (5 μL, 10 μM), and water (final volume = 50 μL), according to the manufacturer’s procedure, followed by incubation at 37 °C for 45 min. Radiolabeled materials were passed through a G-25 Sephadex column followed by purification via electrophoresis (20% denaturing PAGE). The bands of interest (slowest) were extruded and eluted over a saline buffer solution (0.1 M NaCl and 0.1 M sodium phosphate, pH 7.2) for 12 h at 37 °C. The remaining solution was filtered and concentrated to dryness under reduced pressure, followed by precipitation over NaOAc (pH 5.5) and ethanol. Supernatant was removed, and the remaining oligonucleotide was concentrated under reduced pressure and dissolved in water. Activity was assessed using a Beckmann LSC 6500 scintillation counter.

Degradation with Nucleases

RNase A and RNase T₁ were obtained from Thermo Scientific and added to water (4 μL into 46 μL of H₂O) to obtain the first dilution. Subsequent 10-fold dilutions were made each time to obtain solutions with lower enzyme concentrations. Fresh solutions were prepared for each experiment each time. All RNA solutions were hybridized by heating to 90 °C with slow cooling to rt (2 h), prior to treatment with the corresponding ribonuclease. Experiments carried out in the presence of the small molecule were incubated for 15 min, prior to treatment with the ribonuclease of interest. The desired ribonuclease concentration (2 μL) was mixed with the RNA of interest (5 μL, 3000–5000 counts), followed by incubation for 45 min (rt), and diluted with loading buffer (6 M Urea) before transferring onto the gel.

DNase I was obtained from New England Biolabs and added to water (10 μL into 40 μL of H₂O) to obtain the first dilution. Subsequent 5-fold dilutions were made to obtain solutions with lower enzyme concentrations. Fresh solutions were prepared for each experiment. All DNA solutions were hybridized by heating to 90 °C with slow cooling to rt (∼2 h) prior to the start of the experiment. Experiments carried out in the presence of the small molecule were incubated for 15 min prior to treatment with DNase I. Then, aliquots of DNase I solutions (2 μL) were added to each tube containing the desired DNA (5 μL – 3000–5000 counts) and incubated at RT for 40 min. Each solution was then diluted with loading buffer (90% formamide) before loading onto the gel.

Hydrolysis Ladder

γ-³²P-5′-RNA of interest is diluted with water (7.5 μL), typically containing 3–4 times as many counts as the oligonucleotide solution for the experiment, followed by the addition of hydrolysis buffer (3.5 μL of 125 mM NaHCO₃, pH 8.9) and heated to 90 °C for 20 min. The mixture is briefly placed on a microcentrifuge, followed by the addition of loading buffer (6 M Urea) before loading onto the gel (20% dPAGE).

An adapted procedure was used to generate the hydroxyl radical ladder of the DNA aptamers. A solution of γ-³²P-5′-oligonucleotide is prepared (10 mM Tris-HCl]­pH 7.5, 10 mMNaCl), followed by the addition of a hydroxyl radical mix (10 μL of a fresh mixture of 400 μM ammonium ferrous sulfate, 2.5 μM EDTA, 10 mM ascorbic acid and either 0.25% H₂O₂, 0.15% H₂O₂, 0.10% H₂O₂, or 0.05% H₂O₂ in a 1:1:2:2 ratio). The reaction mixture is incubated at RT for 30 min before precipitating twice with ethanol. The remaining pellet is then dissolved in 20 μL of stop solution (10 mM EDTA, 1 mM NaOH, and 80% formamide before loading onto the gel (20% dPAGE).

Isothermal Titration Calorimetry

Measurements were acquired on a MicroCal iTC200 microcalorimeter (Microcal, Inc.) or an Affinity ITC (TA Instruments-Waters) at 25 °C with RNA concentrations of 10–15 μM and small molecule concentrations of 100–500 μM. ONs were dialyzed in the corresponding buffer with a 2 kDa dialysis cassette followed by heating to 90 °C with slow cooling to rt (2–10 h). Both the RNA aptamers and small molecule were dissolved in a buffer consisting of sodium phosphate (10 mM, pH 7.2), NaCl (100 mM), and MgCl₂ (1 mM). RNA concentrations were determined via UV–vis spectroscopy. The small molecule was titrated into the sample cell in 2 μL injections with a reference power of 10 μcal s^–1, an initial delay of 600 s, 180 s spacing, and a stirring speed of 750 rpm. Analysis of the data was performed using Origin 7.0 ITC software (Microcal Software, Inc.) via fitting to a single-site binding model.

Results

RNA Structure and Thermal Stability

The incorporation of all chemical modifications into oligonucleotides of RNA and DNA was carried out as previously reported, via solid-phase synthesis (Scheme S1). − A synthetic route for the synthesis of the 2’-OMe phosphoramidite analogue of 8-oxoG was developed and is reported herein (p. S5–S12, Scheme S2). The positions of focus were G6, G7, G10, G12, and G15, which are known to be involved in the recognition of the neomycin target or within the vicinity of the binding domain. Position-G8 was not studied, given its lack of participation within the binding pocket. It was hypothesized that the incorporation of 8-oxoG at these positions would render: (1) structural flexibility through additional base pairing interactions (for G6 and G7); (2) structural rigidity via 8-oxoG base pairing with A at the loop (for G10); (3) thermal destabilization (for G15) along with possible formation of a different structure, as the canonical G-C WC or G·U wobble base pair(s) would be disrupted; and (4) no structural impact/binding for G12, which is positioned at the center of the loop (Figure , left). Circular dichroism was used to confirm the formation of a secondary structure in all oligonucleotide constructs (ON1-ON16) as well as to establish their thermal stabilities. As expected, the incorporation of any of the chemical modifications within the stem resulted in decreased thermal stability (Figure , entries 2–7 and 10), presumably from disruption of the G:C/G·U with 8-oxoG:C/8-oxoG·U base pair interactions, consistent with reported trends on duplexes of RNA. On the other hand, modification at the loop also induced thermal destabilization, although its impact was less severe (positions G10/G12, entries 8, 9). This is a result that is more difficult to explain without the use of other techniques, such as NMR; however, it has been shown that positioning 8-oxoG at the loop of tetraloop hairpins can lead to thermal destabilization or stabilization as a function of position. It is possible that conformational changes around the glycosidic bond at these positions may contribute to the overall thermal destabilization of the hairpin. Furthermore, experiments carried out on the 2’-OMe-modified aptamers displayed increased thermal stability compared to their canonical analogues (entries 11–13 and entries 1, 2, and 9), due in part to contributions from hydrophobic interactions. Lastly, all DNA analogues also displayed substantial thermal destabilization (entries 14–16), in agreement with previous reports that compared the thermal stability of RNA and DNA hairpins. , Importantly, all aptamers displayed the expected features that were consistent with formation of a hairpin with A-form (bands displaying positive and negative ellipticity at ∼265 and 210 nm) or B-form (bands displaying positive and negative ellipticity at ∼280 and 240 nm) duplexes (Figure , right). More in-depth structural features are discussed later.

3.

Aptamer sequence highlighting the positions that were chosen for single point mutations, along with a legend describing the expected structural effect (left); table describing the thermal stability of each aptamer, used in this work, recorded via circular dichroism (center) and representative examples of three aptamers for RNA, 2’-OMe RNA, and DNA, folding into structures containing A- or B-form duplex regions (right). ΔT _m values represent the difference between the canonical RNA aptamer (Can) and each construct, with all measurements being carried out in triplicate. T _m values were recorded at 265 nm for RNA and 280 nm for DNA.

Aptamer Binding Selectivity and Affinity

We then used isothermal titration calorimetry (ITC) to establish the affinity and selectivity of all aptamers toward different aminoglycoside/small molecules. A report that explored the selectivity of the aptamer of interest (NEO1 in the ref) against 11 aminoglycoside derivatives, including neomycin, paromomycin, ribostamycin, tobramycin, kanamycin, Geneticin, and gentamicin, displayed affinity toward all except the last two derivatives. Thus, we decided to explore binding to the target molecules shown in Figure , which included the aminoglycosides described in that work, as well as streptomycin and penicillin (not an aminoglycoside, used as a negative control). Initial efforts focused on the use of isothermal titration calorimetry (ITC) to measure RNA-aminoclygoside binding constants. Experiments were carried out on a number of aptamers, unfortunately two important factors diverted our efforts from pursuing the use of ITC to establish binding affinities: (1) poor small-molecule solubility in the chosen buffer, which hindered reproducibility (particularly for cases that displayed larger K _d values); and (2) we considered that the required volume/concentration (aptamer: 300 μL, 10–15 μM; small-molecule: 150 μL, 0.1–10 mM) for each experiment, were a limiting factor. Despite these challenges, we obtained preliminary data with some aptamers (Table S2, p. S14) that led to the following conclusions: (1) Can aptamer displayed a high level of promiscuity, showing affinity toward almost all aminoglycosides; (2) OG6 displayed an apparent 2–5 fold increase in selectivity toward neomycin; (3) OG10, OG15, and I6 did not display an increase in selectivity for any target; and (4) introduction of a 2’-OMe modification (Can-OMe, OG6-OMe) displayed similar trends as the 2’–OH analogues.

We then decided to probe RNA-target binding using circular dichroism, keeping in consideration that small structural changes can be recorded in this manner with varying changes in magnitude, , and that smaller quantities of both aptamer and target are needed (aptamer: 150 μL, 2–5 μM; small-molecule: 0.5–10 μL, 0.05–100 mM, dissolved in water). Gratifyingly, titration of all aptamers with the corresponding aminoglycosides displayed clear spectroscopic changes as a function of [small-molecule]; the only exceptions were geneticin and penicillin, which are active upon irradiation with circularly polarized light, thus rendering them unfit for this technique. A representative example is shown in Figure A (left), where the CD spectra of aptamer OG6 display significant hypodichroic/hyperdichroic shifts at different wavelengths, upon small-molecule addition. The change in ellipticity was then plotted as a function of [ribostamycin] and fitted to the Hill equation (Figure A, center and right) to yield values that were in close agreement with those observed via ITC. Importantly, independent plots corresponding to changes at different wavelengths yielded comparable values and provided evidence about the synergistic spectroscopic variations that validate measurements at different regions of the CD spectrum. Thus, CD experiments were carried out on each aptamer-small molecule combination (Figure B–D) to yield reproducible data of dichroic signal changes as a function of [target]. Each data set was also plotted using a quadratic binding equation that led to the same trends and with comparable dissociation constants. Spectra, plots, and pertinent information with regard to each experiment can be found within the SI (Figures S10–S126, pp. S18–S188).

4.

(A) Representative example of the CD changes as a function of [ribostamycin] using the OG6 aptamer as model (left); along with the corresponding plot of the change in dichoric signal as a function of ribostamycin added, highlighting the major changes at 215, 245, and 265 nm (center); and its corresponding plots (ΔΦ vs [ribostamycin]) leading to K _d values derived at each wavelength (right). (B–D) Dissociation constants for all aptamer:small-molecule pairs, as obtained via CD, with all measurements carried out in duplicate or triplicate. Entries where a (−) is present represent no binding or binding with K _d above ∼10 mM.

Experiments were first carried out on the canonical model (Can), which displayed low selectivity toward the tested aminoglycosides. Given the known interaction between ring I in neomycin and position G6 of the aptamer (Figure B), we decided to modify this position with 8-oxoG, 8-oxoI, I, U, or 8-oxoA and probe for potential changes in target recognition. Unfortunately, CD experiments carried out on these aptamers displayed similar trends as those observed with the canonical aptamer, generally showing a mild increase in selectivity toward ribostamycin/neomycin and weakest affinity for kanamycin and streptomycin (Figure B). Although the 8-oxoI and inosine analogues showed a promising trend, with slightly improved affinity toward neomycin and ribostamycin, most values were below a 10-fold enhancement mark, which prompted us to explore other positions. To this end, 8-oxoG was incorporated at positions 7, 10, 12, and 15, also located within the binding domain or at the loop. However, CD experiments carried out on these modified aptamers displayed a comparable lack of selectivity to that obtained with RNA modified at position-6 (Figure C). Aiming to increase the chemical and enzymatic stability of the aptamers, while preserving their physicochemical properties, the 2’-OMe and 2’-deoxy (DNA) analogues were prepared and probed against all aminoglycosides. Since aptamers modified at positions-6 and −12 yielded a slight increase in selectivity toward neomycin/ribostamycin, we decided to further explore these positions. Measurements on the Can-OMe and OG6-OMe aptamers exhibited affinities that were in a comparable range to their canonical analogues, and with a slight increase in selectivity toward kanamycin and tobramycin, showing dissociation constants 5 to 20-fold higher. The poor selectivity and high affinities are perhaps expected results given that aminoglycosides have been reported to bind to 2’-OMe-modified oligonucleotides. On the other hand, experiments carried out with the OG12-OMe aptamer also displayed low selectivity, albeit with different patterns and higher K _d values for gentamicin and kanamycin. Focus was then shifted to three DNA analogues with modification at positions -6 or -12 (DNAcan, DNAOG6, DNAOG12). As illustrated in Figure D (highlighted entries), all DNA aptamers displayed an increase in selectivity toward different targets, where those modified with 8-oxodG arose as better candidates. While DNAcan recognized neomycin, ribostamycin, and gentamicin with comparable dissociation constants, DNAOG6 aptamer displayed a preference toward neomycin and tobramycin, albeit with weaker affinity (∼10-fold) compared to the unmodified analogue. Interestingly, modified DNA aptamer DNAOG12 displayed an increase in selectivity (1–2 orders of magnitude) toward neomycin and ribostamycin, while retaining comparable affinities as the other aptamers (K _d ∼ 2 μM). The increased selectivity arising from both 8-oxodG-modified 2’-deoxy aptamers was surprising, given that their RNA analogues did not show any important changes in aminoglycoside selectivity. However, it is known that the introduction of chemical modifications at the ribose or phosphate backbone can result in changes in selectivity and/or affinity. As with the RNA aptamers, ITC was carried out via titration of the canonical DNA and 2’-OMe RNA aptamers with neomycin, to yield dissociation constants that were in close agreement with those obtained via CD. These results confirmed that CD is a reliable technique to establish K_d values on the oligonucleotide-target pairs.

From this screening the following points are noteworthy: (1) that position-12 is important when using the DNA aptamer, with increased selectivity toward neomycin/ribostamycin; (2) that the 2’-OMe RNA modifications did not affect the overall selectivity in a significant manner; and (3) that the DNA aptamer is more amenable to further applications, given its increased chemical/enzymatic stability. Importantly, these results highlight the possibility that the use of 8-oxoG/8-oxodG represents a promising strategy to modify and potentially enhance the selectivity and affinity of aptamers broadly.

Circular Dichroism and Aptamer Structure

All aptamers displayed the expected features of either A- or B-form duplexes, and all binding aptamers exhibited structural changes that differed as a function of added small molecules as well as chemical modification within the aptamer. Representative examples of such changes are illustrated in Figure A–C, showing examples of RNA, 2’-OMe-modified RNA, and DNA, titrated with buffer, neomycin, or tobramycin. Titration of the aptamers with buffer (up to 10 μL; Figure A-left/C-left) induced minor spectroscopic shifts that allowed us to unequivocally assign changes associated with oligonucleotide binding to the small-molecule. Interestingly, the addition of neomycin or ribostamycin to the RNA aptamers induced significant structural alterations that were observable by their CD spectra. Experiments corresponding to the canonical (Can) or OG-6 aptamers displayed the appearance of two bands with positive ellipticity (λ_max ≈ 285 and 265 nm, Figure A-middle), which is consistent with features of a Z-form RNA and is also associated with glycosyl torsions that perhaps facilitate/enable target recognition, presumably through the stretch of G·U wobble base pairs. The other noticeable change can be observed in the band at 210 nm, which intensifies in the presence of the small molecule and can be associated with increased intrastrand interactions. On the other hand, titration of these aptamers with the other aminoglycoside derivatives did not exhibit such pronounced spectroscopic changes, which suggests that the modes of binding for each target might be different, where a representative example is illustrated with the OG12 aptamer and tobramycin (Figure A-right). This is also consistent with the obtained thermodynamic parameters using ITC (Figure S127). For example, binding to neomycin for both OG10 and OG15 displayed positive entropy, associated with conformational changes on these RNA constructs and/or disturbance of the solvent sphere between the aptamer and the small-molecule. , Binding to gentamicin displayed the same behavior for all aptamers, with increased entropy values; while titration with kanamycin or paromomycin behaved differently depending on the aptamer used, with aptamer OG-15 exhibiting unfavorable interactions. Furthermore, decreased ΔS values were obtained for tobramycin and ribostamycin, consistent with minor changes associated with CD spectroscopic features. It is noteworthy that neomycin and paromomycin are remarkably similar, while neomycin and ribostamycin differ in the presence of ring IV. Thus, it is uncertain at this point on the factors that give rise to these changes, likely a combination of the presence of different functional groups and electrostatic interactions between the small molecule and oligonucleotide. As with various aptamers, spectroscopic changes associated with binding were less pronounced for the 2’-OMe modified constructs. Examples of the three 2’-OMe modified aptamers binding to neomycin are displayed in Figure B, where the band at 210 nm generally showed hypodichroic shifts following addition of the target molecule, while the band at 260 nm underwent shifts of varying intensity.

5.

(A–C) CD measurements before and after the addition of buffer, neomycin, or tobramycin onto solutions containing an aptamer of interest. Insets within each panel represent the change in dichroic signal (ΔΦ) as a function of added small-molecule; and (D) T _m measurements of canonical, OG6, and OG12 modified aptamers of RNA, DNA, 2’-OMe-RNA (∼2 μM, 1 M NaCl, 100 mM Na₂PO₄, 1 mM MgCl₂), before and after treatment with 200 mol equiv of the corresponding small molecule, recorded at 265 nm for RNA and 280 nm for DNA.

With regard to the DNA aptamers, the expected B-form structure was observed with each construct. As with the RNA analogues, minor changes were observed upon addition of buffer (Figure C, left), thus validating the assignment of other spectroscopic variations to binding interactions. While the changes for the OG6-DNA aptamer were detected on the bands with ellipticity at 280 and 220 nm (Figure C, middle), titration of Can-DNA or OG12-DNA with neomycin led to changes that were consistent with a B → A form transition with the appearance of a band at 210 nm (Figure C, right). This result is in agreement with previous experiments on dG _n ·dC_n duplexes, which showed preferential binding by neomycin driven by affinity to the smaller major groove within A-form duplexes. This change was not observed for OG6-DNA, presumably because of the position of 8-oxoG within the stem, which may induce a larger energy barrier to overcome the corresponding B → A conformational transition.

Lastly, the thermal stabilities for the aptamers of interest were measured in the presence of neomycin or ribostamycin. Since DNA aptamers containing an 8-oxodG modification at positions -6 and -12 displayed increased selectivity, these were compared to their corresponding RNA and 2’-OMe RNA analogues (Figure D). Generally, the addition of neomycin increased the thermal stability of the aptamers by varying magnitudes. Unexpectedly, the largest stabilization was observed on the OG12RNA aptamer, which suggests that 8-oxoG at this position plays a major role in the recognition of the small molecule. The only RNA construct that exhibited thermal destabilization was the methylated aptamer OMeOG6, and although the nature of this difference is unclear, it indicates the importance of the G·U stretch in recognition. On the other hand, the impact of ribostamycin on the thermal stability of the aptamers had no effect or milder stabilization than that observed with neomycin. Titration of the other RNA aptamers with the different aminoglycosides presented in this work did not display a measurable difference in thermal stability (Figure S128).

Structural Probing with RNase T₁, RNase A, and DNase I

To gain a better understanding of the impact of 8-oxoG on aptamer structure and recognition, we carried out experiments in the presence of different endonucleases and three RNA constructs (Can, OG6, and OG12), as well as their corresponding DNA analogues. All oligonucleotides were radiolabeled at the 5′-end and treated with the corresponding nuclease, followed by electrophoretic analysis. Treatment with RNase T₁ leads to cleavage on single-stranded regions at G-sites while not recognizing 8-oxoG, thus, we expected positions G10 and G12 to be sensitive to nuclease-induced hydrolysis, given their location within the pentaloop. However, only a minor amount of cleavage at position G12 was observed for the Can and OG6 aptamers (Figure A, lanes b and c) in the absence of neomycin, consistent with an extent of loop availability at this position and the possibility that G10 is involved in intrastrand interactions. The lack of cleavage at the loop for the OG12 aptamer is consistent with the lack of recognition toward 8-oxoG by this ribonuclease (lane d). Unexpectedly, the addition of the small molecule led to increased degradation of all RNA constructs (lanes e–g), suggesting that binding to the target increases the availability of the loop region to solvent, with aptamer OG6 displaying the least amount of degradation. Position G10 displayed minor reactivity for the canonical aptamer Can, while aptamer OG12 showed hypersensitivity. The results suggest that the environment around the loop region changes following target binding and that both positions 10 and 12 become available to the solvent sphere, with aptamer OG6 displaying the least reactivity.

6.

Treatment of RNA or DNA aptamers with RNase T₁, RNase A, or DNaseI in the presence/absence of neomycin B. For clarity, gel images were cut to display the lanes of interest and full images are available within the SI (Figures S132 and S133).

Experiments were then carried out in the presence of RNase A (Figure B), which cleaves single-stranded RNA at pyrimidine sites and with recognition toward 8-oxoG depending on the sequence context. All aptamers displayed cleavage at positions U18, C9, U5, and C4, in the absence of neomycin, with no significant differences (lanes k-m). It is possible that the G·U base pair stretch destabilizes the stem region and makes these positions more prone to enzymatic cleavage by RNase A. On the other hand, reactions carried out in the presence of the small molecule led to different results as a function of the position of 8-oxoG. Ribonuclease activity was significantly diminished for the canonical RNA (Can, lane n), which supports a binding mechanism dictated by the stem region. Results with aptamer OG6 displayed a binding pattern similar to that observed in the absence of the small molecule (lanes l and o), indicating a different binding mode than the canonical analogue. Interestingly, results with aptamer OG12 led to the disappearance of the band corresponding to positions U18 and C4, along with the appearance of a band at G12, consistent with target binding through the G·U stretch while freeing position 12 at the loop. Further indication of the structured nature of the stem region was obtained upon treatment of the aptamers with exonuclease T, which catalyzes cleavage of single-stranded RNA, in the 3′ → 5′ direction with preference when a 3′-overhang is present. None of the aptamers displayed reactivity, while the control single-stranded RNA was processed effectively (Figure S130). Although neomycin has been shown to function as an inhibitor of some ribonucleases, our results displaying differences in degradation provided the necessary information and suggested that potential inhibition of RNase T₁/A activity is not a factor.

Lastly, experiments were carried out using DNase I with the DNA aptamers (Figure C), where this endonuclease binds to the minor groove of B-DNA and cleaves each strand independently. Interestingly, cleavage for all DNA aptamers occurred along the G·T Wobble base pair stretch in the presence/absence of a small molecule target (lanes s-x, confirmed with a OH radical ladder – Figure S131). The only difference was that modification at position-6 (OG6-DNA) displayed an increased level of cleavage, presumably from alterations along the duplex region within the stem. This is also consistent with depressed T _m values obtained for this aptamer, compared to its canonical and OG6-modified DNA analogues (Figure , entries 14–16). In addition, the lack of protection at the stem from neomycin suggests that the target molecule binds in a different manner to the aptamer, compared to its RNA analogous structures.

Discussion

The neomycin RNA aptamer (NEO 61) was used as a model to probe for the use of 8-oxoG as a building block to enhance its selectivity. The focus for the incorporation of single-point mutation was positions G6, G7, G10, G12, and G15, which are within the aminoglycoside binding pocket or in its vicinity (Figure ). In addition, the 2’-OMe and 2’-deoxy (DNA) analogues were probed against all small-molecule targets. CD was used to determine dissociation constants and establish changes in selectivity, which were confirmed via ITC. Lastly, structural probing experiments were used to establish structural differences arising from the incorporation of 8-oxoG/8-oxodG on the RNA/DNA constructs.

Importantly, dG12→8-oxodG and dG6→8-oxodG mutations within the DNA aptamers led to an increase in selectivity toward neomycin/ribostamycin or neomycin/tobramycin/paromomycin, respectively. This was a surprising result given that the RNA analogue OG12/OG6 did not display significant selectivity increases; however, changes along the major groove, such as those induced by the formation of a B-form duplex in DNA, are known to result in enhanced binding affinity to neomycin. In fact, the footprinting studies in the present work indicate that the modes of binding between the aptamer and the target molecules differ between RNA and DNA. While binding of RNA to neomycin protects the stem and makes the loop region more available, binding of DNA to neomycin does not seem to be interrelated or affected by the G·U stretch in the stem. We hypothesize that structure stability along the stem is a contributing factor, specifically for the OG6-DNA aptamer, which displayed the lowest T _m value, suggesting that the stem region is associated with the lack of recognition toward gentamicin, kanamycin, and ribostamycin. Thus, destabilization of the duplex region in DNA could be used as a strategy to increase aptamer selectivity, in this case arising from potential flexibility within the G·U/T wobble base pair stretch, while still within an A- or B-form duplex structure (observed via CD in this work and previous reports). Further comparison of the three DNA aptamers indicates that the loop plays a role in the recognition of gentamicin, as the unmodified DNA aptamer yielded the lowest dissociation constant. It is known that varying the conformation of an oligonucleotide by using DNA or RNA can lead to retention of selectivity/activity. On the other hand, homologues of DNA and RNA can share their recognition motif while leading to altered function. These are important observations, as the use of DNA in aptamers renders chemical and enzymatic stability compared to RNA. We are now expanding this study and exploring the use of these DNA aptamers as tools for the detection of these aminoglycosides.

To explore more alternatives that increase chemical/enzymatic stability, 2’-OMe modifications were introduced; however, no significant changes were observed compared to its canonical RNA analogues. This family of aptamers folded into hairpins with an A-form duplex region in agreement with its preference to adopt C3′-endo conformations, while increased favorable hydration interactions are presumably responsible for their increased thermal stability. It is possible that these structural similarities account for the comparable trends in selectivity with respect to their corresponding canonical RNA analogues. The results suggest that this modification may be used without inducing changes in the selectivity, although more examples are needed in order to generalize this trend.

While it is tempting to use known NMR data to explain the results in this work, it is likely that other binding modes are in play, given the changes in experimental conditions and the presence of different chemical modifications. Thus, the use of other biophysical techniques/tools such as X-ray, modeling, or NMR (mimicking the conditions of this work) will be needed to establish the exact binding modes/sites. It is reasonable to expect that varying electrostatic interactions, with the use of different chemical modifications, can play a role in the recognition of the cognate aminoglycoside targets. Lastly, although the use of 2’-deoxy or 2’-OMe chemical modifications enhances the aptamers’ enzymatic and chemical stability, it is likely that further functionalization will be needed for these aptamers to become functional, , e.g., the use of phosphorothioate or phosphoramidate groups. Besides the chemical and enzymatic stability of these modifications, both DNA and 2’-OMe modifications are attractive due to their ability to serve as efficient polymerase substrates for their evolution.

In addition, expanding the number of techniques that can be used to obtain binding affinities/selectivity in aptamer technologies is of value. To this end, we were inspired by previous work and employed CD to reliably derive dissociation constants. Advantages for the use of this technique include (1) lower substrate (oligonucleotide and small-molecule) amounts needed, compared to ITC; (2) no fluorophore/reporter needed, compared to microscale thermophoresis (MST); (3) no need for aptamer immobilization, such as those employed in surface plasmon resonance (SPR); and (4) that solubility challenges may be circumvented, since substrate does not necessitate the use of the same solvent system/buffer. On the other hand, limitations on the use of CD for this purpose include that (1) the substrate(s) amounts are still larger than those needed for MST or SPR; (2) although CD is highly sensitive to structural changes, their measurability is not granted; and (3) one must establish the lack of signal background from titrant and buffer.

With regard to changes in structure, it is interesting to note that binding of aptamer to neomycin or ribostamycin induced conformational changes on some of the RNA aptamers that led to the appearance of CD features consistent with a Z-form duplex content. Furthermore, these aminoclygosides induced B → A form transitions on two DNA aptamers. Neomycin has also been reported to alter the structure of DNA in a significant manner and induce the conversion of B- to A-form duplexes, with interactions involving GGG regions across the major groove. Furthermore, the use of other aptamers does not display this behavior, indicating that the structural changes occur in a sequence-/structure-dependent manner. Experiments to induce the formation of a Z-form duplex were carried out by recording CD spectra of aptamers Can & OG6 at increased [Na⁺] (up to 4 M NaCl), however, only decreased signal intensity was observed. Differences in CD and increased thermal stabilities have been observed in the presence of neomycin, using ψ-RNA. On the other hand, the addition of other aminoglycosides, e.g., paromomycin, increased the thermal stability of octamers of RNA:RNA, DNA:DNA, RNA:DNA hybrids, as well as triple helices. Thus, the thermal stabilization of the RNA or DNA secondary structure is not unprecedented.

It is important to note that the incorporation of certain modifications, e.g., inosine, can result in loss of recognition; thus, care must be taken in the design of aptamers. Furthermore, binding/recognition of different target molecules varies, e.g., tobramycin binds to poly­(rI)·poly­(rC) A-form duplex at the major groove over a span of four nucleobases, and in a pH-dependent manner. Overall, the results provided in this work are promising; however, it is worth emphasizing that they are limited to one pentaloop hairpin and a set of aminoglycoside small molecules. More structural information is necessary with the modifications used in this work to take a rational approach in the design of aptamers containing 8-oxoG as a building block. To this end, we are currently combining theoretical models with experimental results to establish the impact of 8-oxoG on common structural motifs of RNA, which will allow us to develop a predictive platform that will also serve to identify metabolite:oxidized RNA pairs. We are also taking advantage of the changes that occur on the RNA structure, following the introduction of an oxidatively generated lesion, to target these motifs as potential druggable sites. Another research area of interest relies on recent results that point to a connection between riboswitches and the recognition of 8-oxoguanine, associated with DNA damage; thus, we hypothesize that the incorporation of 8-oxoG within riboswitches may result in altered activity and, with potential biological implications, a concept that we will be exploring in the near future. Lastly, besides the importance of establishing structural patterns involving 8-oxoG in aptamer development, understanding the impact that this chemical modification has on RNA structure and function is of importance in a biological context. 8-OxoG is an oxidatively generated lesion that is formed under oxidative stress and is a common biomarker of many biological processes, also associated with the development/progression of some diseases.

Conclusions

In conclusion, we have shown that 8-oxoG can be used as a building block in combination with ribose modifications (2’-OMe and 2’-deoxy) for the design of aptamers of RNA, to increase their selectivity. Specifically, we showed that functionalization of position-12 and −6 with 8-oxodG results in increased selectivity and represents a promising probe to obtain aptamers of DNA with unique/enhanced recognition properties. Furthermore, the incorporation of 2’-OMe modified aptamers (including 8-oxoG) displayed similar selectivity, with slightly lower affinity, which indicates that the formation of chemically/enzymatically stable aptamers can be achieved in this manner. While it is premature to claim that this will be a general procedure, given the lack of a selection process that can be explored with RNA containing 8-oxoG, the incorporation of 8-oxopurine derivatives represents a promising approach to obtain aptamers of RNA/DNA with unique and interesting properties. The work herein showed its impact when present in different regions of the hairpin aptamer, and we believe that the study/use of oxidatively generated modifications within RNA is an area with high potential for discovery, broadly.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acschembio.5c00518.

• Experimental procedures and full characterization for the 2’-OMe phosphoramite of 8-oxoG; MALDI-TOF for all oligonucleotides used; pertinent CD spectra and T _m traces; isothermal titration calorimetry of pertinent experiments; and electrophoretic analysis for the reactions with RNase A, RNase T1, exonuclease T, and DNase I (PDF)

†.

T.-A.J. and S.W.S. contributed equally to this manuscript.

The authors declare no competing financial interest.