Cytosine-Rich DNA Aptamers Reveal Recognition Principles for Structurally Diverse Lipopolysaccharides

Abstract

Lipopolysaccharides (LPS) are structurally heterogeneous glycolipids whose variable composition and supramolecular organization challenge the design of analytical recognition elements. Here, independent magnetic bead-based SELEX against four LPS preparations yielded three cytosine-rich DNA aptamers that bound five structurally distinct LPS preparations, including both O-antigen-containing and O-antigen-deficient forms, with nanomolar apparent dissociation constants. Mutational analysis showed that disruption of the native C-rich sequence architecture strongly impaired binding, and the repeated emergence of such motifs across independent selections points to an underexplored sequence solution for LPS recognition. UV melting and native PAGE showed acidic pH-induced i-motif-like transitions, whereas binding was maximal at neutral pH, indicating that the acidic folded state is not the functional recognition conformation. Comparisons across immobilization formats, component competition assays with isolated lipid A and free Kdo, and biolayer interferometry with isolated lipid A indicate that recognition is not explained by isolated lipid A or free Kdo alone but is instead consistent with intact LPS presentation in a core-proximal context under the assay conditions. Together, these results reveal recognition principles that link cytosine-rich sequence motifs, pH-dependent folding, target presentation, and core-proximal LPS features in the analytical molecular recognition of heterogeneous LPS.

Introduction

Lipopolysaccharides (LPS) are major components of the outer membrane of Gram-negative bacteria that exhibit pronounced structural heterogeneity across species. − LPS consists of three structural domains: lipid A, the core oligosaccharide region containing 3-deoxy-d-manno-oct-2-ulosonic acid (Kdo), and the highly variable O-antigen. , Although lipid A is often considered a conserved structural motif, variations in acylation patterns, phosphorylation states, and supramolecular organization are frequently observed among bacterial species. , Additionally, diversity in O-antigen composition, together with differences in aggregation behavior and amphiphilic assembly, results in highly heterogeneous LPS architectures. , Consequently, intact LPS presents not a single well-defined epitope but a structurally heterogeneous, multivalent surface with variable charge distribution and conformational flexibility. This complexity poses a fundamental challenge for developing molecular recognition elements capable of robust, consistent cross-species binding. As LPS constitute the molecular basis of bacterial endotoxin activity and must be tightly controlled in pharmaceutical and bioprocessing contexts, achieving robust, species-insensitive recognition is of both fundamental and practical importance. ,

Various molecular recognition strategies have been developed to target LPS, including lipopolysaccharide-binding protein (LBP), antilipid A antibodies, and peptide-based ligands such as polymyxin derivatives. − Many of these approaches focus on lipid A because of its relative structural conservation. However, lipid A accessibility varies depending on species-specific chemical modifications and supramolecular organization. , Meanwhile, the O-antigen region exhibits substantial structural variability across bacterial strains, limiting its suitability as a universal recognition site. Consequently, strategies that target either highly conserved minimal motifs or highly variable surface structures may not fully capture the structural and organizational complexity of intact LPS assemblies. Together, these considerations suggest that structural features proximal to the core oligosaccharide region may offer a useful compromise between conservation and chemical complexity for cross-species analytical recognition of intact LPS.

Nucleic acid aptamers are single-stranded DNA or RNA molecules selected in vitro through systematic evolution of ligands by exponential enrichment (SELEX). , They bind a wide range of molecular targets with high affinity and specificity, offering several advantages as molecular recognition elements, such as chemical synthesizability, structural adaptability, and ease of modification. Beyond their practical applications, aptamers provide a chemically defined, sequence-controlled system for examining how sequence composition and folding influence target recognition. This makes them particularly suitable for investigating recognition mechanisms toward structurally complex and heterogeneous targets such as LPS.

Because the analytical motivation of this study is broad LPS monitoring rather than strain typing, the desired recognition profile differs from that of species-specific aptamers. Species-specific aptamers are valuable when the analytical target is a single defined bacterial strain or chemotype. However, such narrow species-level selectivity is less advantageous for heterogeneous LPS analytes encountered in endotoxin monitoring, pharmaceutical quality control, or broad-spectrum bioprocess surveillance. In these contexts, a recognition element that tolerates interspecies structural variation while still engaging conserved or recurrent features presented by intact LPS assemblies is more practically useful than one restricted to a narrow species-specific epitope.

To the best of our knowledge, although an increasing number of studies have reported LPS-binding aptamers, − a systematic understanding of how aptamer sequence features relate to epitope preference and cross-species recognition remains lacking. Reported LPS-binding aptamers have mainly been developed in the context of endotoxin detection, separation, or broad-spectrum capture. − Although these studies established the feasibility of aptamer-based LPS recognition, they generally focused on individual binding sequences or assay formats rather than on how sequence composition, folding behavior, target presentation, and LPS structural heterogeneity jointly determine recognition. Although cytosine-rich or i-motif-related architectures have been reported for aptamers against non-LPS targets, , previous LPS-binding aptamer studies have primarily focused on individual binding sequences, endotoxin capture, or assay development. − To our knowledge, contiguous C-tracts have not been systematically established as recurring functional features of LPS-binding aptamers. Thus, whether contiguous cytosine tracts can reproducibly emerge as functionally required sequence features for recognition across structurally diverse LPS preparations remains unclear. Because LPS structures vary widely across species and even strains, comparisons across diverse LPS types are needed to distinguish target-specific interactions from more general recognition modes. , Moreover, the influence of selection conditionsparticularly target immobilization and presentationon the emergence of epitope preference remains incompletely understood. , Additionally, cytosine-rich, pH-responsive DNA architectures remain relatively underexplored as functional aptamer scaffolds, despite isolated examples of i-motif-derived aptamers. , At the same time, the field has accumulated individual LPS-binding sequences without an accompanying relationship that links sequence composition, folding behavior, target presentation, and epitope preference across structurally heterogeneous LPS.

To examine this possibility, this study investigates how sequence composition, folding behavior, and target presentation govern aptamer recognition across a structurally diverse LPS panel. In the present study, this question is addressed at the level of broad recognition across the structurally diverse LPS panel tested here, rather than by claiming a priori identification of a fully universal binder for all LPS chemotypes. Using magnetic bead-based SELEX against LPS derived from multiple Gram-negative bacteria, we independently identified aptamers that converged on cytosine-rich sequence motifs. This convergence is itself notable from a recognition-design perspective, because it suggests a reproducible sequence solution rather than a selection-specific outcome of a single selection. Through mutational analysis, pH-dependent structural characterization, and epitope-probing experiments, we evaluated how cytosine continuity and assay format shape aptamer recognition of intact LPS. These results define a sequence–structure–recognition relationship and support intact LPS presentation in a core-proximal context as a plausible design principle for analytical molecular recognition of heterogeneous LPS analytes.

Experimental Section

Preparation of LPS-Immobilized Magnetic Beads

LPS derived from Escherichia coli O157, E. coli O55, Pseudomonas aeruginosa PAO1, and Salmonella enterica serovar Minnesota 1114 (Fujifilm Wako, Japan) were dissolved at 1 mg/mL in 100 mM acetate buffer (pH 5.5). The carbohydrate moieties of LPS were oxidized by adding one-tenth volume of freshly prepared 100 mM sodium periodate, followed by incubation for 1 h at 25 °C in the dark. Excess oxidant was removed by ultrafiltration (Amicon 10K, Merck, USA) with three buffer exchanges into 10 mM acetate buffer (pH 4.0). Carboxyl-modified magnetic beads (Dynabeads MyOne Carboxylic Acid, Thermo Fisher Scientific, USA) were activated by incubation with EDC (20 mg/mL) and NHS (10 mg/mL) in 100 mM MES buffer (pH 5.0) for 1 h at 25 °C. After washing, hydrazide groups were introduced by incubation with 5 mM carbohydrazide in PBS (pH 7.4) for 1 h at 25 °C. Oxidized LPS (200 μg/mL in 10 mM acetate buffer, pH 4.0) was then added to the beads and incubated for 2 h at 25 °C to enable hydrazone bond formation. After washing, unreacted sites were blocked with 1 M ethanolamine for 30 min. The LPS-immobilized beads were washed and stored in PBS at 4 °C until use.

Oligonucleotide

All DNA oligonucleotides were synthesized by Eurofins Genomics (Japan), Sigma-Aldrich (USA), or Thermo Fisher Scientific (USA). The ssDNA library comprised a 72-nt sequence containing a 30-nt randomized region (N30) flanked by two 18-nt primer-binding regions, each separated from the randomized region by a 3-thymidine (TTT) spacer (18 nt–TTT–N30–TTT–18 nt). Before each selection round, the ssDNA library was heat-denatured at 95 °C for 5 min and allowed to cool to 25 °C for at least 20 min to facilitate proper folding. The sequences of the library and primers are listed in Table S1.

Magnetic Bead–Based SELEX

Magnetic bead–based SELEX (MB-SELEX) was performed for 10 rounds using LPS-immobilized magnetic beads. LPS derived from E. coli O157, E. coli O55, P. aeruginosa PAO1, and S. enterica serovar Minnesota 1114 (S. minnesota 1114) were immobilized separately, and independent SELEX procedures were conducted for each LPS species. To compare selection outcomes under common conditions, the amount of immobilized target was not deliberately changed during the initial 10 rounds; each round used 50 μL of LPS-immobilized bead suspension. No counter-selection step was included in the present protocol. Folded ssDNA libraries were incubated with the LPS-immobilized beads in PBS containing 1 mM MgCl₂ (binding buffer) for 1 h at 25 °C. Beads were washed three times with washing buffer (PBS containing 1 mM MgCl₂ and 0.05% Tween-20) in each round; washing stringency was not progressively increased during these rounds. Bound ssDNA was eluted twice using elution buffer (PBS containing 0.05% Tween-20, 10 mM EDTA, and 4 M urea) at 80 °C for 5 min. Eluted DNA was ethanol-precipitated and amplified by PCR using PrimeSTAR GXL DNA polymerase (Takara Bio, Japan). PCR was performed with an initial denaturation at 98 °C for 30 s, followed by 20 cycles of 98 °C for 10 s and 68 °C for 20 s. Double-stranded DNA was captured using streptavidin-coated magnetic beads, and ssDNA was regenerated by alkaline treatment with 0.1 M NaOH for 5 min. After neutralization and ethanol precipitation, the regenerated ssDNA pool was carried forward to the subsequent round as completely as practical. Because the amount of selected DNA recovered after each round was near the reliable quantification limit, the regenerated ssDNA pool was not separately normalized to a fixed absolute input amount before each round. Late-round pools from rounds 8–10 were prepared for indexed sequencing and analyzed using a MiSeq system (Illumina). Sequence analysis and candidate identification were performed using AptaSUITE and AptCompare. , Candidate aptamer sequences were subjected to primary binding screening against multiple LPS species using an enzyme-linked oligonucleotide assay (ELONA). Based on the overall preliminary ELONA response profiles, including signal intensity and detectable binding across multiple LPS preparations, three representative aptamers were selected for further characterization.

Biotinylation of LPS

LPS were oxidized as described above (periodate oxidation: 10 mM NaIO₄, 1 h at 25 °C in 100 mM acetate buffer, pH 5.5) to generate aldehyde groups on carbohydrate moieties. The oxidized LPS were buffer-exchanged into 10 mM acetate buffer (pH 4.0) by ultrafiltration (Amicon Ultra-0.5, 10 kDa cutoff; 14,000g, 4 °C, 10 min) with three buffer exchanges. Biotinylated LPS were prepared by hydrazide coupling. Biotin-PEG3-hydrazide (Fujifilm Wako, Japan) was added to the oxidized LPS at a final concentration of 5 mM in 10 mM acetate buffer (pH 4.0) and the reaction mixture was incubated for 2 h at 25 °C with gentle agitation. Because this labeling strategy targets aldehyde groups generated by periodate oxidation of oxidizable carbohydrate moieties, biotinylation is expected to occur primarily at accessible carbohydrate positions within the glycan region of each LPS preparation, rather than within the hydrophobic acyl-chain portion of lipid A. Owing to the structural heterogeneity of LPS, the exact labeling sites were not assigned to single defined sugar residues. After coupling, excess reagent and salts were removed by ultrafiltration (Amicon Ultra-0.5, 10 kDa cutoff; 14,000g, 4 °C, 10 min) with five buffer exchanges into PBS.

ELONA-Based Binding Assay

LPS derived from E. coli O157, E. coli O55, P. aeruginosa PAO1, S. minnesota 1114, and S. enterica serovar Minnesota R595 (Re) (S. minnesota R595) (Fujifilm Wako, Japan) were biotinylated as described above. The biotinylated LPS were diluted to 10 μg/mL in PBS and immobilized (100 μL per well) on SuperBlock-treated streptavidin-coated 96-well plates (Thermo Fisher Scientific, USA) by overnight incubation at 4 °C. The ELONA washing buffer consisted of PBS containing 1 mM MgCl₂ and 0.05% Tween-20. After LPS immobilization, wells were washed three times with ELONA washing buffer. Folded 5′-FAM–labeled aptamers at varying concentrations were added and incubated for 1 h at 25 °C. After aptamer incubation, wells were washed three times, and bound aptamers were detected using an HRP-conjugated antifluorescein antibody (Abcam, UK; 1:10,000 dilution in ELONA washing buffer). Chemiluminescence was developed using an enhanced chemiluminescent substrate (SuperSignal ELISA Pico Chemiluminescent Substrate, Thermo Fisher Scientific, USA) and measured with a microplate reader. Background signals from the corresponding control wells were subtracted before curve fitting. Apparent dissociation constants (K _d) were determined by fitting the background-subtracted concentration-dependent ELONA signals to a one-site saturation binding model using nonlinear regression in GraphPad Prism (version 10.0):

$$Y=\frac{B_{\max }\times X}{K_{\mathrm{d}}+X}$$

where X is the aptamer concentration and Y is the background-subtracted ELONA signal.

Mutational Analysis of C-Rich Sequences

To evaluate the contribution of cytosine tracts to LPS binding, mutant aptamers were designed in which three or more consecutive cytosine residues were replaced with adenine or thymine while maintaining overall sequence length. These substitutions were used as a representative C-tract-disruption strategy rather than as an exhaustive mutational scan of all possible replacement patterns. Guanine-containing substitutions were avoided to minimize the possibility of introducing alternative G-dependent secondary-structure features or additional base-pairing interactions that could complicate interpretation of the mutational effects. Binding of each mutant aptamer to LPS was assessed using the ELONA procedure described above at a concentration of 100 nM. Binding signals were normalized to those of the corresponding wild-type aptamers and expressed as percentages of wild-type binding.

UV Melting Analysis

Aptamers and their mutants were dissolved at a final concentration of 10 μM in acetate buffer (pH 5.0), MES buffer (pH 6.0), or PBS (pH 7.4). UV absorbance at 295 nm was monitored from 0 to 95 °C at a heating rate of 1 °C/min using a UV-1900i spectrophotometer (Shimadzu, Japan). Blank-corrected absorbance values were normalized to a minimum of 0 and a maximum of 1 for each sample. The melting temperature (T _m) was defined as the temperature at which the normalized absorbance reached 0.5.

Native Polyacrylamide Gel Electrophoresis

DNA samples (2 μM) were prepared as described above. LPS was not included in the native PAGE samples. The analysis was performed to assess the intrinsic pH-dependent conformational behavior of the aptamers in the absence of target. Native polyacrylamide gel electrophoresis was performed using 10–20% gradient gels (E-T1020L, ATTO, Japan) on ice for 3 h using running buffers adjusted to pH 5.0, 6.0, or 7.4. The running buffers were 10 mM acetate buffer at pH 5.0, 10 mM MES buffer at pH 6.0, or PBS at pH 7.4, each supplemented with 1 mM MgCl₂. After electrophoresis, gels were stained with SYBR Gold (Thermo Fisher Scientific, USA) diluted 1:10,000 for 30 min at 25 °C and visualized using a GelDoc imaging system (Bio-Rad, USA). Previously reported cytosine-rich DNA sequences were included as reference controls for native PAGE analysis. The reference DNA sequences used for native PAGE analysis are listed in Table S3 in the Supporting Information.

pH-Dependent Binding Analysis

Biotinylated LPS derived from S. minnesota 1114 or S. minnesota R595 were immobilized on streptavidin-coated plates as described in the ELONA-based binding assay section. Plates were equilibrated with buffers adjusted to pH 5.0, 6.0, or 7.4, and folded 5′-FAM–labeled aptamers (50 nM) were added and incubated for 1 h at 25 °C. After washing with the corresponding pH-adjusted buffer, plates were rinsed with PBS before incubation with HRP-conjugated antifluorescein antibody to minimize pH-dependent effects on the secondary antibody reaction.

Immobilization-Dependent LPS Binding Assays

To evaluate the influence of LPS presentation on binding behavior, LPS-binding assays were performed using three different plate surfaces: Polysorp microplates, Maxisorp microplates (Thermo Fisher Scientific, USA), and SuperBlock-treated streptavidin-coated plates with immobilized biotinylated LPS. Polysorp and Maxisorp plates were used as passive adsorption-based immobilization surfaces with different adsorption properties, allowing us to examine whether changes in passive LPS presentation influence aptamer binding relative to biotin–streptavidin immobilization. For adsorption-based assays, LPS solutions (10 μg/mL in PBS) derived from E. coli O55 and S. minnesota R595 were added to Polysorp or Maxisorp plates (100 μL per well) and incubated overnight at 4 °C. Wells were washed with washing buffer and blocked with SuperBlock for 1 h at 25 °C. Biotinylated LPS were immobilized on streptavidin-coated plates as described above. For aptamer binding measurements, folded 5′-FAM–labeled aptamers (50 nM) were added and incubated for 1 h at 25 °C. After washing, bound aptamers were detected using an HRP-conjugated antifluorescein antibody. Signals were developed using a TMB substrate, stopped with acid, and measured at 450 nm. For comparison, lipopolysaccharide-binding protein (LBP), a well-characterized LPS-binding protein, was evaluated as a reference LPS-binding protein under the same immobilization conditions rather than as a negative control. Plates were blocked with 5% skim milk for 1 h at 25 °C prior to the assay. His-tagged LBP (Sino Biological, China) diluted to 1 μg/mL in washing buffer containing 3% skim milk was added to the wells and incubated for 1 h at 25 °C. After washing, bound LBP was detected using an HRP-conjugated anti-His-tag antibody (Medical & Biological Laboratories, Japan; 1:5000 dilution in washing buffer) followed by TMB colorimetric detection at 450 nm.

Competition ELONA and Interference Assays

Competition ELONA and interference assays were performed using the immobilized-LPS ELONA format described above. Briefly, biotinylated E. coli O157 LPS (10 μg/mL) was immobilized on streptavidin-coated plates as described in the ELONA-based binding assay section. FAM-labeled aptamers were preincubated with the indicated soluble competitors or representative non-LPS molecules in binding buffer (PBS containing 1 mM MgCl₂) for 1 h at 25 °C. The mixtures were then added to the LPS-coated wells and incubated for 1 h at 25 °C. After incubation, wells were washed three times with ELONA washing buffer (binding buffer containing 0.05% Tween-20), and bound aptamers were detected using an HRP-conjugated antifluorescein antibody as described above. For epitope-related competition assays, FAM-labeled O55-16 aptamer (50 nM) was preincubated with lipid A derived from E. coli (Fujifilm Wako, Japan; 0–5 μM) or free Kdo (3-deoxy-d-manno-oct-2-ulosonic acid; Sigma-Aldrich, USA; 0–500 μM). These assays were used to test whether isolated lipid A or an isolated inner-core sugar could compete with aptamer binding to immobilized LPS. For interference assessment using representative non-LPS molecules, FAM-labeled O55-16, PAO1-26, and Minne-27 aptamers (50 nM) were evaluated in the presence of glucose, dextran, or lipoteichoic acid (LTA). Glucose and dextran were used as soluble carbohydrate controls and were tested at 10 mM and 1 mg/mL, respectively. LTA (Sigma-Aldrich, USA) was included as a representative Gram-positive cell-envelope polymer and was tested at 1, 10, and 100 μg/mL to evaluate concentration-dependent interference.

Biolayer Interferometry

Biolayer interferometry (BLI) measurements were performed using an Octet R8e system (Sartorius, Germany). Biotinylated O55–16 aptamer (1 μM) was immobilized on streptavidin-coated biosensors for 300 s, followed by baseline stabilization in PBS containing 1 mM MgCl₂ for 60 s. Association was monitored by immersing the sensors in solutions containing lipid A (1 μM) or intact LPS (100 μg/mL) for 300 s, followed by dissociation in PBS containing 1 mM MgCl₂ for 300 s. All measurements were conducted at 25 °C with agitation at 1000 rpm.

Safety Considerations. No unexpected or unusually hazardous conditions were encountered. Standard laboratory precautions were used when handling sodium periodate, sodium hydroxide, urea, and bacterial lipopolysaccharide preparations.

Results and Discussion

Selection of LPS-Binding Aptamers and Evaluation of Their Affinity

Independent SELEX experiments were conducted against four distinct LPS species under fixed selection conditions. To evaluate late-round enrichment, NGS was performed for pools from rounds 8–10. Because enrichment was not monitored in the earlier rounds, the precise round at which sequence enrichment first became detectable was not determined. Recurrent sequence families were evident in the late-round pools and became prominent by round 10, consistent with substantial convergence under the applied selection conditions. Therefore, the round-10 pools were used for candidate ranking and for reporting the top-abundance sequences (Table S4). Sequence analysis of the round-10 pools identified multiple candidate sequences from each selection. Candidates were prioritized based on sequencing abundance, motif features identified using AptaSUITE and AptCompare, sequence diversity, and preliminary single point ELONA responses rather than abundance rank alone. The selected aptamers were assigned internal screening IDs (O55-16, PAO1-26, and Minne-27), which do not correspond to abundance ranks in the NGS data sets. Candidates from all four round-10 pools were considered during the initial screening stage; however, no O157-derived candidate showed sufficient preliminary binding to be included among the ten candidates reported in Table S5 and Figure S4. Ten prioritized primer-trimmed candidates were then evaluated by preliminary single-point ELONA screening against the LPS panel as an exploratory triage step (Table S5 and Figure S4). Three representative aptamers were then selected for detailed characterization based on their overall preliminary response profiles, including signal intensity and detectable binding across multiple LPS preparations. Notably, all three independently selected sequences contained extended cytosine tracts, indicating recurring enrichment of C-rich motifs under the applied selection conditions rather than an isolated outcome from a single SELEX experiment. To assess affinity and recognition breadth, concentration-dependent ELONA measurements were performed against five structurally distinct LPS samples. This panel was selected to cover LPS preparations from different Gram-negative bacterial genera and to include both O-antigen-containing smooth-type LPS and O-antigen-deficient rough-type LPS, while using commercially available and well-characterized materials for reproducible comparison. The panel included smooth-type LPS possessing O-antigen structures (E. coli O157, E. coli O55, P. aeruginosa PAO1, and S. minnesota 1114) as well as rough-type LPS lacking O-antigen (S. minnesota R595). All aptamers exhibited nanomolar apparent dissociation constants (K _d) under the immobilized-LPS assay conditions used here (Table and Figure S1). The K _d values reported here are ELONA-derived apparent K _d values obtained from individual fits to concentration-dependent binding curves under immobilized-LPS assay conditions. ELONA was used as the comparative affinity assay across the aptamer–LPS panel because it provided concentration-dependent binding curves for all selected aptamers and LPS preparations under a single consistent assay format. Aptamer O55-16 showed the most consistent low-nanomolar apparent affinity across the tested panel, with K _d values from 1.3–5.0 nM. PAO1-26 and Minne-27 also retained nanomolar apparent affinity across species (3.6–26.0 nM), although moderate species-dependent variation was observed. Importantly, robust binding to rough-type LPS (R595), which lacks O-antigen, indicates that O-antigen structures are not required for binding under these assay conditions. Collectively, these data indicate cross-species recognition within the tested LPS panel and are consistent with engagement of shared structural features in intact LPS assemblies.

1. Apparent Dissociation Constants (K _d) of Each Aptamer for Various LPS Species .

	apparent K d values (nM)
	LPS sources
aptamer name	E. coli O157	E. coli O55	P. aeruginosa PAO1	S. minnesota 1114	S. minnesota R595
O55-16	2.0 ± 0.6	2.9 ± 1.1	1.3 ± 0.7	5.0 ± 2.2	4.6 ± 1.9
PAO1-26	5.9 ± 1.8	9.6 ± 3.2	6.0 ± 2.1	26.0 ± 13.2	10.6 ± 4.4
Minne-27	5.3 ± 1.9	8.5 ± 3.5	3.6 ± 2.0	12.2 ± 5.1	10.2 ± 3.6

^aValues are best-fit ELONA-derived apparent K _d values with 95% confidence intervals, obtained from nonlinear regression of three replicate measurements.

Functional Importance of Continuous Cytosine Tracts

To evaluate the functional significance of continuous cytosine residues, mutant aptamers were designed as targeted C-tract-disruption constructs in which consecutive cytosines (CCC or longer) were substituted with adenine or thymine while maintaining overall sequence length (Table S2). Binding of these mutant aptamers to LPS was assessed by ELONA. In all cases, disruption of continuous cytosine tracts resulted in a marked reduction in binding signal compared with that of wild-type aptamers, which were normalized to 100%. Regardless of the LPS species tested, binding signals of the mutants decreased to approximately 10% or less of that of the wild-type levels (Figure ).

1.

Binding of wild-type and mutant aptamers to various LPS species. (a–c) Comparison of binding between original (black bars) and mutant (white bars) aptamers for O55-16 (a), PAO1-26 (b), and Minne-27 (c). Data represent the mean ± SD (n = 3). Binding signals were normalized to the corresponding wild-type aptamer and are shown as percentage of wild-type binding. LPS, lipopolysaccharide; SD, standard deviation.

Disruption of the native contiguous C-tracts strongly impaired binding across the tested LPS preparations. However, because the substitutions also reduced the total cytosine content, the present targeted mutants do not resolve the relative contributions of tract continuity and overall cytosine content. Nevertheless, the repeated emergence of C-rich motifs across independent selections against structurally distinct LPS targets, together with the pronounced mutational sensitivity, supports the functional importance of the native C-rich sequence architecture under the present conditions. At neutral pH, this functional dependence is unlikely to reflect formation of a canonical acidic i-motif based on hemiprotonated C:C+ pairs. , Instead, the native C-rich sequence architecture may help bias the aptamer toward a distinct neutral-pH conformational ensemble through local base stacking, hydrogen-bonding potential, and hydration-mediated interaction patterns. Such an ensemble could present functional groups in a geometry compatible with structural features displayed by intact LPS assemblies in a core-proximal context, rather than through simple electrostatic attraction or recognition of an isolated component alone. This interpretation provides a plausible structural rationale rather than a residue-level assignment, because the precise neutral-pH topology and molecular contacts remain unresolved.

pH-Dependent Structural Behavior and Its Functional Relevance to LPS Binding

Cytosine-rich DNA sequences undergo pH-dependent folding transitions, including formation of i-motif-like conformations under acidic conditions. , To determine whether the C-rich LPS-binding aptamers identified here exhibit such behavior and to clarify its functional relevance, we characterized their pH-dependent structural transitions and evaluated LPS binding across the same pH range. UV melting profiles recorded at 295 nm revealed cooperative, sigmoidal transitions at pH 5.0–6.0 for all three wild-type aptamers, consistent with previously reported signatures of i-motif formation (Figure a). , In contrast, no cooperative transition was observed at pH 7.4. Mutant aptamers lacking continuous cytosine tracts did not display sigmoidal melting behavior under any pH condition (Figure b), indicating that continuous C-tracts are essential for the acidic pH–dependent transition. Native PAGE provided complementary evidence for conformational differences between wild-type and mutant aptamers (Figure c). Across pH 5.0, 6.0, and 7.4, wild-type aptamers migrated consistently more slowly than the corresponding mutants, consistent with C-tract–dependent folding. The duplex control further supports that the neutral-pH species observed here are not simply intermolecular dimers under the tested conditions. A previously reported, C-rich reference sequence exhibited characteristic pH-dependent mobility shifts and served as a comparative control for interpreting electrophoretic behavior (Figure c). Hence, these results indicate that the aptamers undergo an acidic pH-favored, i-motif-like transition at pH 5.0–6.0, whereas the pH 7.4 state lacks the cooperative melting behavior associated with this acidic transition, consistent with a distinct neutral-pH conformational ensemble. Secondary structure prediction using m-fold under neutral conditions did not indicate formation of highly stabilized canonical hairpin structures (Figure S2), consistent with the absence of cooperative melting behavior at pH 7.4. Next, we examined how these structural transitions relate to functional LPS recognition. Binding assays showed maximal aptamer–LPS interaction at pH 7.4, with progressively reduced binding at pH 6.0 and 5.0 for both smooth-type S. minnesota 1114 LPS and rough-type S. minnesota R595 LPS (Figures d and S3). Although the i-motif-like transition indicates pH-dependent stabilization of the C-rich motif under acidic conditions, this acidic state did not correspond to enhanced LPS binding; LPS binding was maximal at neutral pH and decreased at pH 6.0 and 5.0. Formation of the acidic i-motif-like state is therefore neither required nor advantageous for effective LPS binding under near-physiological conditions. Reduced binding at lower pH likely reflects a shift toward conformations less compatible with LPS recognition and may also involve pH-dependent changes in electrostatic interactions or LPS organization. These results distinguish the acidic folded state from the neutral-pH state associated with LPS binding. Although the precise topology of this neutral-pH state remains structurally unresolved, the combined evidence indicates that the acidic transition itself is not the binding-favored state under the present assay conditions. Additional structural characterization, including CD spectroscopy and higher-resolution structural approaches, will be important for defining the detailed topology of this neutral-pH ensemble associated with LPS binding.

2.

pH-dependent structural behavior of cytosine-rich LPS-binding aptamers. (a) UV melting profiles of wild-type aptamers at 295 nm and pH 5.0, 6.0, and 7.4. (b) UV melting profiles of mutant aptamers containing disrupted cytosine tracts, showing no cooperative transitions under any pH conditions tested. (c) Native PAGE analysis of wild-type and mutant aptamers, together with cytosine-rich reference sequences, at pH 5.0, 6.0, and 7.4. Lane assignments: M, DNA marker; lanes 1–2, O55-16 WT and mutant; lanes 3–4, PAO1-26 WT and mutant; lanes 5–6, Minne-27 WT and mutant; lanes 7–11, cytosine-rich reference sequences previously reported to adopt i-motif or related structures (Table S3); and lane 12, duplex formed between O55-16 and its fully complementary strand. (d) pH dependence of O55–16 binding to S. minnesota LPS. Black, white, and gray bars indicate pH 5.0, 6.0, and 7.4, respectively. Data represent the mean ± SD (n = 3). 1114 and R595 denote S. minnesota strains 1114 and R595 (Re), respectively. LPS, lipopolysaccharide; SD, standard deviation.

Presentation-Dependent LPS Binding and Epitope-Probing Analysis

To clarify whether the observed binding behavior could be explained primarily by lipid A, O-antigen accessibility, or presentation-dependent exposure of core-proximal LPS features, we compared aptamer binding across multiple LPS immobilization formats and orthogonal binding assays. Binding signals were first compared under three LPS immobilization conditions: Polysorp microplates, Maxisorp microplates, and streptavidin-coated plates with immobilized biotinylated LPS. Under all conditions, the aptamers exhibited detectable binding to LPS, whereas LBP showed only weak reactivity on Polysorp plates (Figure a). Notably, for rough-type LPS (S. minnesota R595), the aptamers produced the highest binding signals on Maxisorp plates, while LBP displayed strong signals predominantly on streptavidin-coated plates (Figure b). These trends were consistent across all three aptamers examined. The observed differences in binding behavior likely reflect variations in LPS presentation under distinct immobilization conditions. To evaluate whether isolated LPS-related components could account for the observed interaction, we first performed competition ELONA using free lipid A and free Kdo. Free lipid A did not substantially inhibit O55-16 binding to immobilized LPS in the ELONA format (Figure c). Free Kdo also did not substantially reduce O55-16 binding to immobilized LPS over the tested concentration range (Figure d), indicating that isolated Kdo alone is not sufficient to recapitulate the aptamer-recognized LPS feature. BLI was then used as an orthogonal response-based assay to compare O55-16 responses toward isolated lipid A and intact LPS, rather than as a quantitative affinity-determination method. When O55-16 aptamer was immobilized on the biosensor surface, the response to lipid A was comparable to that of PBS alone. In contrast, clear binding signals were detected for intact LPS derived from E. coli O55 and S. minnesota R595 (Figure e). Together, these observations indicate that the measured interaction is not dominated by isolated lipid A and is not competed by free Kdo. In addition, the retained binding to O-antigen-deficient R595 indicates that O-antigen structures are not required for binding under the present assay conditions. Rather, the results are most consistent with recognition of structural features presented by intact LPS assemblies in a core-proximal context. Because this interpretation is based on immobilized-LPS assays, soluble-component competition experiments, and BLI response comparisons rather than direct structural mapping, the precise molecular contacts remain unresolved. Thus, the proposed core-proximal recognition context is best viewed as a working model supported by functional epitope-probing data, as summarized in Figure .

3.

Presentation-dependent LPS binding and epitope-probing analysis of C-rich aptamers. (a, b) Immobilization-dependent LPS binding assays performed using Polysorp, Maxisorp, and streptavidin-coated plates. Signals in (a) and (b) were measured by colorimetric HRP detection at 450 nm. (a) Binding of FAM-labeled aptamers to LPS immobilized on different plate surfaces. (b) Binding of His-tagged LBP to LPS immobilized on different plate surfaces. (c) Lipid A competition assay for O55-16 binding to immobilized LPS. (d) Kdo competition assay for O55-16 binding to immobilized LPS. Signals in (c,d) were measured by chemiluminescent HRP detection. (e) Representative BLI sensorgrams showing binding of O55-16 to lipid A and intact LPS. Data in (a–d) represent the mean ± SD (n = 3). LPS, lipopolysaccharide; LBP, lipopolysaccharide-binding protein; Kdo, 3-deoxy-d-manno-oct-2-ulosonic acid; BLI, biolayer interferometry; SD, standard deviation.

4.

Schematic illustration of LPS and a proposed core-proximal recognition context consistent with the present functional data. The schematic summarizes a working model in which aptamer binding is associated with intact LPS presentation in a core-proximal context, rather than being explained by isolated lipid A, free Kdo, or O-antigen alone. The precise molecular contacts remain unresolved. LPS, lipopolysaccharide; Kdo, 3-deoxy-d-manno-oct-2-ulosonic acid.

General Implications for LPS Recognition

Taken together, these results point to an important measurement-science consideration for heterogeneous LPS analytes: aptamer performance depends not only on sequence composition but also on how the target is presented during selection and assay. Independent selections against distinct LPS preparations converged on cytosine-rich sequences, disruption of the native C-rich sequence architecture strongly impaired binding, and the selected aptamers retained nanomolar apparent affinity across a structurally diverse LPS panel. At the same time, changes in immobilization format altered binding behavior, indicating that target presentation can influence the apparent epitope preference that emerges from SELEX. The lack of inhibition by isolated lipid A or free Kdo, together with the weak BLI response to lipid A, argues against a model in which the observed binding is determined primarily by isolated minimal LPS components. Instead, the present evidence is most consistent with recognition of structural features presented by intact LPS assemblies in a core-proximal context, under the present assay formats, although the precise molecular contacts remain unresolved. Additional interference assays with representative non-LPS molecules were performed to define the analytical boundary of the selected aptamers (Figure S5). Glucose and dextran did not appreciably reduce binding to immobilized LPS, indicating that the responses are not readily explained by generic soluble carbohydrate recognition. In contrast, LTA reduced binding at 10–100 μg/mL, identifying this Gram-positive cell-envelope polymer as a potential interferent in the current assay format. For analytical molecular recognition, these observations are important: intact-LPS presentation in a core-proximal context may provide a useful balance between strict conservation and extreme variability when designing recognition elements for heterogeneous LPS. More broadly, recognition of heterogeneous analytes may depend less on maximizing affinity to a single purified motif than on sequence features and selection conditions that preserve binding across structural and presentation differences in the target. The present study therefore provides a sequence-structure-recognition relationship for designing aptamer-based recognition elements for chemically heterogeneous LPS and related glycolipid assemblies.

Conclusion

This study identifies cytosine-rich DNA aptamers capable of recognizing a structurally diverse panel of LPS and clarifies the sequence- and assay-dependent factors underlying this behavior. Independent selections consistently converged on contiguous C-tracts, and mutational analysis showed that disruption of the native C-rich sequence architecture strongly impaired binding under the tested conditions. Acidic i-motif-like folding was observed, but LPS binding was favored at neutral pH, indicating that the acidic folded state is distinct from the neutral-pH state associated with LPS binding; the precise topology of this neutral-pH ensemble remains unresolved. Epitope-probing experiments indicate that recognition is not explained by isolated lipid A or free Kdo alone. In addition, retained binding to O-antigen-deficient R595 indicates that O-antigen structures are not required for binding under the assay conditions. Together, these findings support a working model in which intact LPS presentation and a core-proximal context contribute to binding. Additional interference assays showed that the responses were not attributable simply to generic soluble carbohydrate recognition, although LTA defined an analytical boundary for the current assay format. Because the apparent affinities were obtained using immobilized-LPS assays and the epitope-probing analysis is indirect, these results should be interpreted as guiding principles for the design of analytical molecular recognition elements rather than as a direct assessment of practical endotoxin-sensing performance. Overall, our findings support intact LPS presentation in a core-proximal context as a plausible design principle and provide a sequence–structure–recognition relationship to guide the development of analytical elements for heterogeneous glycolipid analytes such as LPS.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmeasuresciau.6c00157.

• DNA library and primer sequences; selected aptamer and C-tract-disrupted mutant sequences; oligonucleotide sequences used for native PAGE analysis; top-abundance NGS sequence tables; ELONA binding curves and apparent dissociation constant fitting; predicted secondary structures of wild-type and mutant aptamers; additional pH-dependent binding data for S. minnesota 1114 and R595 LPS; and preliminary single-point ELONA screening of prioritized candidates; and interference assays using representative non-LPS molecules (PDF)

A.T.: Conceptualization, methodology, investigation, data curation, writingoriginal draft. N.K.: Investigation, data curation, writingreview and editing. M.Y.: Supervision, project administration, writingreview and editing. All authors reviewed and approved the manuscript.

This study was financially supported by Kirin Holdings Company, Ltd.

The authors declare no competing financial interest.