Skip to main content

Evaluation of the interactions between oligonucleotides and small molecules by partial filling–nonequilibrium affinity capillary electrophoresis


Recently, high-throughput analysis with minimal reagent consumption has been desired to assess interactions between drug candidates and disease-related oligonucleotides. To realize an ideal assay for drug screening, a rapid assay based on affinity capillary electrophoresis was generated to reduce the consumption of samples/reagents by a partial-filling technique under nonequilibrium conditions. In the proposed method, the first sample, oligonucleotide as a ligand, and second sample zones were injected into a capillary with spacers of background solution between the samples and oligonucleotide zones. After applying voltages, only the second sample zone passed through the partially filled oligonucleotide zone, resulting in variations in the peak parameters owing to this interaction. The electropherograms obtained were analyzed using equilibrium, reaction kinetics, and moment theories. In the interaction analyses between small molecules and DNA aptamers, only the small molecules binding to the aptamer showed significant changes in their peak heights and shapes. The estimated kinetic parameters were in good agreement with the reported values, indicating the applicability of the proposed method for drug screening. When interactions between drug candidates and disease-related RNAs were analyzed, one of the candidates showed remarkable variation in the peak parameters upon the addition of potassium ions. Consequently, the proposed method could be one of the ideal assays for drug screening.

Graphical abstract


Recently, analyses of interactions between small molecules and oligonucleotides have been important because disease-related oligonucleotides are expected to be new targets for drug discovery. A variety of analytical methods are available to measure the interactions between biomolecules, including surface plasmon resonance (SPR) [1,2,3], quartz crystal microbalance (QCM) [4,5,6], and isothermal titration calorimetry (ITC) [7, 8]. However, there are several disadvantages; for example, SPR and QCM require to immobilize the ligands. Therefore, handling is very complicated, and the structure of the ligand is changed by immobilization and modification; thus, it is difficult to measure the direct interactions between biomolecules in bodily fluids. In ITC, by contrast, no immobilization process is necessary for the assay because ITC analyses can be performed only by mixing solutions. Moreover, ITC has a great advantage in that kinetic and thermodynamic parameters can be evaluated by a single measurement without labeling. However, these methods require long analysis times and large sample consumption, which sometimes prevents their use for rare biological samples from patients.

Capillary electrophoresis (CE), liquid-phase separation method based on electrophoresis in a narrow hollow capillary (50–100 µm, ID), allows small sample and reagent consumption, short analysis time, and simplified handling by automation. CE is also applicable to the analysis of interactions between analytes and ligands in solution on the basis of electrophoretic separation of sample molecules, ligands, and complexes. Thus, CE plays an important role in the analysis of interactions between biomolecules. Over the past 10 years, affinity capillary electrophoresis (ACE) [9], which is a CE mode that employs a background solution (BGS) containing ligand molecules, was developed by Kanayama [10], who reported separation based on the difference in single nucleotide polymorphisms by ACE. In another report, Dinges reported interaction analysis between antithrombin-III and low-molecular-weight heparins and determined the dissociation constant [11]. However, ACE requires a slightly large amount of ligand solution because the inlet and outlet vials and capillary must be filled with a background solution containing a ligand (total 1–2 mL). Thus, it can be difficult to analyze a few, rare biomolecules, such as disease-related oligonucleotides, from patients. In addition, ACE often exhibits low sensitivity because of the background absorption of the ligands in the BGS. In addition, ACE needs to be measured under multiple conditions to accurately evaluate the interaction, which causes low throughput screening for drug discovery. Nonequilibrium capillary electrophoresis of equilibrium mixtures (NECEEM) [12, 13] and capillary electrophoresis reactor (CER) [14] are also powerful tools for the assessment of interactions between analytes and ligands. However, for NECEEM and CER, sample and ligand solutions must be mixed before the analysis, which increases the consumption of samples/ligands by at least 50 µL for each CE analysis. Thus, it is necessary to prepare as many mixture solutions as possible, therefore, these techniques are not suitable for drug screening because of the requirements of pre-mixing of the solutions.

Recently, a partial-filling technique (PF) was applied to ACE to decrease sample consumption and improve sensitivity. In PF–ACE, analytes are separated during migration in a partially filled ligand-solution zone in a capillary [15]. For example, the separation of the enantiomers of basic drugs by PF–ACE was reported by Tanaka [16]. In another report, the interaction between thrombin and thrombin inhibitors was analyzed by PF–ACE, which provided a successful determination of the dissociation constant [17]. PF–ACE has been widely used in this way because it has many advantages, such as short analysis time, small amounts of sample/ligand consumption (sub-µL order), and improved sensitivity by partial filling owing to suppression of the background absorption by the ligands. However, ACE and PF–ACE must be measured under experimental conditions in which the concentration of ligands is sufficiently higher than that of the analytes, which often makes it impossible to assess the interactions of rare biomolecules with significantly low concentration.

To overcome this drawback, we propose a new method for high-throughput screening of interactions between small molecules and low-concentration oligonucleotides by applying PF to ACE under nonequilibrium conditions (partial-filling technique with nonequilibrium affinity capillary electrophoresis, PF–NEACE, Fig. 1). In this study, the first sample, ligand, and second sample zones were injected into a capillary with spacers of a background solution (BGS) between these zones (Fig. 1a). After applying the separation voltage, only the second sample zone could pass through the partially filled ligand zone (Fig. 1b, c), resulting in the variation of peak parameters due to the interaction with the ligands, as compared to the first sample peak with no interaction (Fig. 1d). As a result, the interaction could be easily estimated by comparing the peak parameters of the first and second sample zones. The proposed PF–NEACE is expected to be an ideal interaction assay with a simple procedure and allow for minimal consumption of samples/ligands without pre-mixing and interference in the detection by the ligand. In this study, the interactions between model analytes and DNA were measured using PF–NEACE to evaluate the fundamental performance. PF–NEACE analyses of drug candidates and disease-related RNAs were also performed to assess their applicability for high-throughput drug screening.

Fig. 1
figure 1

Schematic illustration of the concept of PF–NEACE. a Initial condition of the capillary after injecting the zones. b Separation of bound samples from free ones while the 2nd sample zone overtakes the ligand zone. c Broadening of the 2nd zone after overtaking. d Evaluation of the peak ratio and peak shift from the typical result in PF–NEACE


Equilibrium theory

Generally, the following equilibrium between the sample molecules (S), ligands (L), and their complex (X) is considered to evaluate the association or dissociation of S and L.

$$S + L\mathop \rightleftarrows \limits_{{k_{{{\text{off}}}} }}^{{k_{{{\text{on}}}} }} X$$

Under the following experimental conditions, the initial concentrations of S1 and S2 ([S1]0 = [S2]0 = [S]0) are larger than that of L ([L]). The injected length of L (ZL0) is greater than that of the S (ZS). When the free sample molecules under the initial condition (Sfree) enter the L zone at the first contact, the competitive association of S with L begins (Eqs. 2 and 3).

$${S}_{\mathrm{free}}+L \stackrel{{k}_{\mathrm{on}}}{\to } X+{S}_{\mathrm{unbound}}$$
$$\left[ X \right]_{1} = k_{{{\text{on}}}} \left[ S \right]_{0} \left[ L \right]$$
$$\left( {\left[ S \right]_{0} \, > \,\left[ L \right], \, \left[ {S_{{{\text{unbound}}}} } \right]_{1} = \left[ S \right]_{0} {-}\left[ X \right]_{1} } \right),$$

where Sunbound is the free S molecule without interacting with L. [Sunbound]1 and [X]1 are the concentrations of Sunbound and X, respectively, in the first contact zone. Under the PF–NEACE condition, the apparent electrophoretic velocity of S (vS), is faster than that of L (vL), and that of X (vX). Thus, Sfree are immediately separated from X after complexation. Sfree can also penetrate the next L zone, not containing the X, because of the longer ZL0 than ZS. In the next L zone, the remaining Sfree interacts with unbound L molecules, which causes a new equilibrium condition (Eqs. 4 and 5).

$${S}_{\mathrm{unbound}, 1}+L \stackrel{{k}_{\mathrm{on}}}{\to } X+{S}_{\mathrm{unbound},2}$$
$$\left[ X \right]_{2} = k_{{{\text{on}}}} \left[ {S_{{{\text{unbound}}}} } \right]_{1} \left[ L \right]$$
$$\left( {\left[ {S_{{{\text{unbound}}}} } \right]_{1} > \left[ L \right],\;\left[ {S_{{{\text{unbound}}}} } \right]_{2} = \left[ {S_{{{\text{unbound}}}} } \right]_{1} {-}\left[ X \right]_{2} } \right),$$

While [Sunbound]n is maintained sufficiently higher than [L], the rate equation related to [S], which is the concentration of the unbound S in the L zone, is expressed as follows (Eq. 6).

$$- \frac{{{\text{d}}\left[ S \right]}}{{{\text{d}}t}} = k_{{{\text{on}}}} \left[ L \right]\left[ S \right]$$

Equation (6) is transformed into Eq. (7) by the method of the separation of variables.

$$\frac{\mathrm{d}[S]}{[S]}=-{k}_{\mathrm{on}}\left[L\right] \mathrm{d}t$$

When Eq. (7) is integrated on both sides from time t = 0, when [S] is [S]0, to time t, when the molar concentration of S becomes [S], the following equations can be derived:


In CE, peak heights with/without the interaction (h2/h1) are related to their concentration [S] and [S]0; thus, the values of [S] and [S]0 can be easily calculated from the obtained electropherogram. The reaction time, t, can also be easily estimated from ZL0 and the difference between vS and vL. Therefore, the association rate constant kon between the S and L can be measured from the plot of ln(h2/h1) against t.

Moment theory

The application of moment theory to PF–CE analyses was evaluated by Miyabe [18]. It is a kinetic analysis method that considers the balance between complex formation and dissociation, and information on the chemical equilibrium, mass transfer rate, and reaction rate can be obtained from the elution peaks that were experimentally measured. The moment analysis method for PF–ACE evaluate the interaction between S and L, with minimal sample consumption. From the reported equations related to the first moment of the PF–ACE analysis, the following equation can be obtained (Eq. 9).

$${v}_{S} {t}_{\mathrm{app}}-\mathrm{Z}=\frac{{v}_{S} - {v}_{S,L}}{{v}_{S,L} - {v}_{L}}{Z}_{L}^{0}$$

where vS,L is the apparent velocity of S with interactions in the ligand zone and Z is the effective length of the capillary. To simplify the equation, the concentration ratio of the free sample molecule and complex ([S]:[X]) in the L zone is assumed to be 1: k, and vS,L should be described as follows:


Under the following experimental conditions, vX can be approximately equal to vL because DNAs/RNAs as ligands are much larger than small sample molecules. As a result, Eq. (9) can be simplified as follows:

$${v}_{S} {t}_{\mathrm{app}}-Z=k{Z}_{L}^{0}$$
$$k=\frac{{v}_{S} {t}_{\mathrm{app}}-Z}{{Z}_{L}^{0}}$$

where ZB0 is the length of the spacer zones between zones S1, L, and S2, which can be calculated by the injection of zone L without any ligands. Under these experimental conditions, the dissociation equilibrium constant, Kd should be estimated by the following equation:

$${K}_{\mathrm{d}}=\frac{\left[S\right] [L]}{[X]}=\frac{{k}_{\mathrm{off}}}{{k}_{\mathrm{on}}}$$

Based on the assumption that [S]:[X] = 1:k, Eq. (14) can be rewritten as follows.


where [L]total and [S]total are the initial concentrations of L and S, respectively, in the L zone that has reached association/dissociation equilibrium. Under these experimental conditions, the X formed at first (X1) should be immediately separated from the main peak of Sfree, which shows the tailing of the peak, as shown in Fig. 1d. After separation, X1 reaches a new equilibrium in the L zone and the concentration balance becomes.

$$\left[ L \right]_{{{\text{total}}}} = \left[ L \right]_{0}$$
$$\ln \left[ X \right]_{1} = \ln \left[ S \right]_{0} {-}\ln \left[ S \right]_{1} = k_{{{\text{on}}}} \left[ L \right]t_{1}$$
$$t_{1} = Z_{S} /v_{S}$$

Thus, the apparent shift, Δt, and the migration time of S interacting with L, tapp, could be measured by the tailed part of the observed peak whose height (h′) is half the height of the initial complex X1 estimated by Eqs. (17) and (18). From the above equations, the interaction between the ligands and sample molecules can be evaluated using PF–NEACE.



Capillary electrophoresis was carried out by CE 7100 system (Agilent Technologies, Santa Clara, CA, USA). A fused silica capillary (inner diameter, 50 µm; outer diameter, 375 µm) was purchased from Polymicro Technologies (Phoenix, AZ, USA). The capillary was cut to 33 cm, and a detection window was opened at a length of 24.5 cm by removing the polyimide coating of the capillary.

Chemicals and reagents

DNA aptamer for tyrosinamide (Apt49-T) with the sequence 5′-AAT TCG CTA GCT GGA GCT TGG ATT GAT GTG GTG TGT GAG TGC GGT GCC C-3′ [19], and disease-related RNAs sequences (GGGGCC)6 and (GGGGCC)8 were artificially synthesized by Tsukuba Oligo Service Co. (Ibaragi, Japan). l-Tyrosinamide (Tyr-Am) was purchased from Combi-Blocks Co. (San Diego, CA, USA). l-Tyrosine (Tyr), l-tryptophan (Trp), l-phenylalanine (Phe), l-phenylalanin-amide (Phe-Am), p-acetamidophenol (AAP) and tert-butylphenol (t-BP) were purchased from Fujifilm Wako Pure Chemical Co. (Osaka, Japan). Biosynthesized disease-related RNA sequence containing (GGGGCC)20 and compounds A [20] and B [21] were provided by Sumitomo Dainippon Pharma Co. (Osaka, Japan). Tris(hydroxymethyl)aminomethane (Tris) and magnesium chloride hexahydrate were purchased from Fujifilm Wako Pure Chemical Co. (Osaka, Japan). Stock solutions of Tyr-Am, Tyr, Trp, Phe and Phe-Am were prepared in deionized water (18.2 MΩ cm) using a Milli-Q system (direct-Q UV 3, Merck, KGaA, Darmstadt, Germany). Stock solutions of AAP and t-BP were prepared in methanol (Kanto Chemical Co., Osaka, Japan).

Solution and sample preparation

Apt49-T and (GGGGCC)6 were dissolved in deionized water to prepare 100 µM stock solutions. Stock solutions of Apt49-T and (GGGGCC)6 were then dispensed in 5 µL aliquots and stored at − 25 °C. A stock solution of 30 µM (GGGGCC)20 dissolved in deionized water was dispensed in 100 µL aliquots and stored at − 80 °C.

Stock solutions of Tyr-Am, Tyr, Trp, Phe, and Phe-Am were prepared in deionized water to a concentration of 1 mM, and stock solutions of AAP and t-BP were prepared in methanol to a concentration of 10 mM. The samples were then stored at 4 °C. As a background solution for the assays to measure the interactions between Apt49-T and the Tyr-Am analogs, 10 mM Tris–HCl buffer (pH 7.5) containing 1 mM MgCl2 was prepared as necessary. In CE experiments using Apt49-T, the samples and aptamer solutions were prepared by diluting the stock solutions with background solutions.

Stock solutions of compounds A and B at a concentration of 1 mM were provided as DMSO solutions and stored at − 25 °C. As a background solution for the assay to evaluate the interactions between (GGGGCC)n and compounds A and B, 10 mM Tris–HCl buffer (pH 7.5) supplemented with 10% (v/v) DMSO was prepared as necessary. In CE experiments using (GGGGCC)n, stock solutions of RNAs were diluted with the background solution. Sample stock solutions dissolved in DMSO were diluted with the buffer solutions without adding DMSO to control the final concentration of the added DMSO, similar to the background solution.

Procedures for CE experiments

Before each CE measurement, the capillary was preconditioned for 200 s with 1 M NaOH, for 200 s with deionized water, and for 200 s with the background solution by applying a high pressure (> 900 mbar). Then, the capillary was filled with the background solution by applying low pressure (50 mbar, 200 s) for conditioning.

The DNA-sample compounds for which Kd values for Apt49-T were reported (Tyr-Am: Kd = 45 µM, Phe-Am: Kd = 80 µM, Tyr: Kd = 520 µM, and the Kd of the other samples were no data) [19] were selected as model analytes. To prepare the 100 µM sample solutions, 1 mM or 10 mM stock solutions were diluted 10- or 100-fold with the background solution, respectively.

In the PF–NEACE analyses using Apt49-T, each solution was introduced into the capillary as follows: the first sample zone (50 mbar, 5 s), the background solution as a spacer (50 mbar, 10 s), the DNA zone (50 mbar, 40 s), the spacer (50 mbar, 10 s), and the second sample zone (50 mbar, 5 s). After partial injection of these solutions, a separation voltage of 15 kV was applied to both ends of the capillary via vials filled with the background solution. UV detection was performed at 210, 220, and 230 nm for the DNA-sample compounds (210 nm for Phe, 220 nm for Phe-Am, and 230 nm for Tyr-Am, Tyr, Typ, AAP, and t-BP).

In the PF–NEACE analyses using (GGGGCC)n, each solution was introduced into the capillary as follows: the first sample zone (50 mbar, 5 s), background solution as a spacer (50 mbar, 20 s), RNA zone (50 mbar, 20 s), spacer (50 mbar, 20 s), and second sample zone (50 mbar, 5 s). After partial injection of these solutions, a separation voltage of 15 kV was applied to both ends of the capillary via vials filled with the background solution. UV detection was performed at 234.5 nm for compounds A and B.

Calculation of ratio of peak height and peak shift

As peak parameters to estimate the interactions between the model DNA/RNAs and samples, the differences in the peak heights and shapes of the first and second sample zones with/without the DNA/RNAs were analyzed. The ratio of the peak height was calculated by the second peak height corresponding to the first peak height (h2/h1), and the obtained values of the ratio were normalized by the ratio obtained in the PF–NEACE analysis without the ligands. Regarding the peak shapes, the apparent peak shift was measured from the tail part of the 2nd peak as illustrated in Fig. 1d. The calculated peak height of the samples forming the first complex, hx, was calculated using Eqs. (17) and (18).

Results and discussion

Analyses of interactions between small molecules and Tyr-Am aptamer

PF–NEACE was performed to evaluate the interactions of the Tyr-Am aptamer (Apt49-T) with the Tyr-Am analogs. As shown in Fig. 3a, only Tyr-Am and Phe-Am showed a remarkable decrease in the peak heights with large tailing, whereas the other five analogs, Tyr, Phe, Trp, AAP, and t-BP, did not change their peak heights and shapes significantly. The normalized values of the peak ratio are summarized in Fig. 3b, which clearly indicate the specific interaction between Apt49-T and Tyr-Am/Phe-Am. The Kd values between Apt49-T with Tyr-Am, Phe-Am, and Tyr were reported to be 45, 80, and 520 µM, respectively, by equilibrium dialysis [19], whereas the Kd values of the other analogs were not due to their weak or non-interaction. From the reported Kd values, the amount of Tyr-Am, Phe-Am, and Tyr bound to Apt49-T were calculated to be ca. 6.8, 5.4, and 1.6 µM, respectively. However, the obtained results did not show an interaction between Tyr and Apt49-T (Fig. 2A(c)) under nonequilibrium conditions. These results suggest that the presence or absence of an amide group has a significant effect on the association reaction rate under nonequilibrium conditions. Consequently, it is demonstrated that the proposed PF–NEACE can easily distinguish molecules that rapidly interact with the ligand from those that do not interact by the variation in the height and shape of the 2nd peak.

Fig. 2
figure 2

Typical results in PF–NEACE for tyrosinamide analogues. A Electropherograms obtained by PF-NEACE (a′) without and (a)–(h) with Apt49-T. Samples: (a) and (a′) Tyr-Am, (b) Phe-Am, (c) Tyr, (d) Trp, (e) Phe, (f) AAP, and (g) t-BP. B Normalized values of the peak heights (h2/h1)

Evaluation of the association rate constant, k on, by PF–NEACE

As explained in Fig. 1 and the Experimental section, a higher concentration of the sample solution (100 µM) was introduced into the capillary than the injected DNA (10 µM). Therefore, the free sample molecules in the DNA zone without binding should be rapidly separated from the bound one, thereby decreasing the height of the 2nd peak. From Eq. (10), the association rate constant kon can be evaluated by varying the reaction time. Under experimental conditions, the reaction time can be estimated as the time required for the 2nd sample zone to overtake the DNA zone, and can be easily controlled by changing the injected length of the DNA zone. In the case of Tyr-Am, the values of injected length and reaction time were calculated to be ca. 1.45–11.6 cm, and ca. 6.9–56.9 s, respectively when the injection time of the DNA zone changed from 10 to 80 s. In the case of Phe-Am, the values of injected length and reaction time were calculated to be ca. 1.23–6.15 cm, and ca. 5.8–30.5 s when the injection time of the DNA zone changed from 10 to 50 s. The height of the 2nd peak decreased with increasing injection time/length of the DNA zone owing to the increase in complex formation. These results are summarized in Fig. 3, and the kon values were determined from the slopes of the approximate lines (Eq. 8), and the kon values of Tyr-Am and Phe-Am against Apt49-T were estimated to be 2300 ± 0.99 and 4300 ± 0.92 M−1 s−1, respectively.

Fig. 3
figure 3

Evaluation of association rate constants of Tyr-Am and Phe-Am against Apt49-T

Evaluation of the dissociation constant, K d, by PF–NEACE

In PF–NEACE, the first complex formed during the penetration of the sample solution into the ligand zone is rapidly separated from the remaining free sample molecules without interactions (Fig. 1b). After separation, the complex reaches a new association/dissociation equilibrium in the ligand zone. As a result of the successive formation of the complex and its separation from the free sample molecules, the 2nd sample zone shows a significant tailing shape (Fig. 1d). As described above, the apparent migration time of S interacting with L is measured by the tailed part of the observed peak, whose height (h′) is half the height of the initial complex X1 estimated by Eq. (3) using the estimated values of kon. When the baseline drift was observed in the electropherogram, the baseline was adjusted by referring to the baselines before and after detecting the 1st and 2nd sample peaks, respectively. When the tailing of the 2nd peak was observed without the ligand, the difference in the detection time of tail parts whose heights were h′ with/without the ligand were measured after adjusting the baselines. Typical results of the calculated average values of kon, Kd, and koff (n = 3) are summarized in Table 1.

Table 1 Kinetic parameters obtained by PF–NEACE

As described above, a smaller Kd value (stronger binding) was estimated for Tyr-Am, despite the larger kon value of Phe-Am. This could be explained by differences in the reaction rates and stabilities of the complexes. The only difference between Tyr-Am and Phe-Am is the presence or absence of a hydroxy group on the aromatic ring. Thus, the faster binding of Phe-Am to Apt49-T can be attributed to the hydrophobic interactions related to the aromatic ring without hydroxy groups. However, the formation of the more stable complex of Tyr-Am with Apt49-T is expected to be due to the hydrogen bonding of the hydroxy group. Consequently, it was confirmed that the proposed PF–NEACE can assess the interaction between the analytes and ligands with a short analysis time and minimal consumption of the sample and ligands, which will contribute to drug screening of rare biological samples as a high-throughput method.

Application of PF–NEACE on screening for drug candidates with disease-related oligonucleotides

To confirm the applicability of PF–NEACE in the screening of drug candidates against disease-related oligonucleotides, the RNA of the expanded repeat ((GGGGCC)n), which possibly causes frontotemporal dementia and amyotrophic lateral sclerosis, was selected as a model RNA sample. As model ligands, artificially synthesized RNAs, (GGGGCC)6 and (GGGGCC)8, and biosynthesized RNA, (GGGGCC)20, were prepared in 0.25–10 µM solutions containing 10 mM Tris–HCl buffer (pH 7.5) and 10% (v/v) DMSO. As model target samples, compounds A and B were prepared at 100 µM by diluting stock solutions with BGS without DMSO. "Upon checking, it was noticed that there are panels in the caption and citation of Figure [3]; however, they were not found in the corresponding image. Please provide us with an updated figure with corresponding panels matching their description in the figure caption."?> "Thanks for careful check. Please revise the Figure caption from "Evaluation of association rate constants of (a) Tyr-Am and (b) Phe-Am" to "Evaluation of association rate constants of Tyr-Am and Phe-Am against Apt49-T"."?>

Figure 4 shows the electropherograms obtained by PF–NEACE analyses of compounds A and B. Negative peaks were often observed because there was a slight difference in their components between the solutions due to the addition of 10% (v/v) DMSO. Compound A showed a significant decrease in the peak height (Fig. 4A), whereas compound B showed a small decrease (Fig. 4B). The calculated values of the peak ratio of compounds A and B against RNA are summarized in Fig. 4C. These results indicate that there were multiple compounds bonded to a single RNA molecule, and the number of molecules bound to the RNA was higher in compound A than in compound B. In the case of one-to-many interactions, it is very difficult to calculate accurate binding constants, whereas it is possible to roughly estimate the difference in interactions, as shown in Fig. 4. Compound A is known to interact with the (GGGGCC)8 structure with a Kd of 16 µM [20], and compound B has been reported to interact with the G–G mismatch structure [21]. Thus, the observed results related to compound A indicated that both RNAs, (GGGGCC)6 and (GGGGCC)20, have a similar structure to the reported RNA, (GGGGCC)8, with a G-quadruplex providing a remarkably strong interaction between compound A and (GGGGCC)6/(GGGGCC)20, which may have G-quadruplex structures. On the other hand, the lower binding of compound B suggests that these RNAs do not have many G–G mismatch structures. Consequently, these results indicate the applicability of PF–NEACE to not only the screening but also the evaluation of the structure of DNAs/RNAs in solution.

Fig. 4
figure 4

Typical results in PF-NEACE analyses of the interactions between the model target with disease-related RNAs. A S, compound A; L, (GGGGCC)6; B S, compound B; L, (GGGGCC)6; C difference in the normalized ratio of the peak heights of the compounds A and B due to the sequence of the RNAs

It is also well known that potassium ions can stabilize G-quadruplex structures. To evaluate the effect of potassium ions on the interaction between compounds A and RNAs, the concentration of RNAs was varied with and without adding KCl in the BGS. The results are summarized in Fig. 5. This clearly showed a decrease in the peak ratio wiht increasing concentrations of each disease-related RNA. For each RNA, the molar binding ratios were also calculated from the decrease in the peak intensities and the initial molar ratio of [S] and [RNA]. Without adding KCl, it was clarified that (GGGGCC)6 and (GGGGCC)8 could bind 17–26 molecules of compound A, while (GGGGCC)20 could only bind up to 17, despite the longer sequence.

Fig. 5
figure 5

Effect of the addition of KCl on the interaction between compound A and RNAs

On the other hand, when KCl was added in the BGS, the peak heights became larger as compared to those without KCl. The estimated numbers of binding molecules were reduced by 9–11 for (GGGGCC)6, 12–17 for (GGGGCC)8, and 3–5 for (GGGGCC)20. These results indicate that the binding of RNAs stabilized by potassium ions reduced the binding sites for compounds A as compared to that without stabilization. Thus, it was suggested that compound A bound to the RNAs at the binding sites related not only to the stabilized planar G-quartet structures but also to the various gaps of the G-quadruplex structures without stabilization by potassium ions. Further and more detailed investigations are required to clarify their binding structure; however, the proposed PF–NEACE analyses provide important insights related to the interactions between compounds A/B and RNAs with simple operation, short analysis time, and minimal consumption of the reagents.

Comparison of PF–NEACE to conventional methods

As discussed above, PF–NEACE has great potential as a useful analysis tool in the interaction screening of small molecules and oligonucleotides. On the other hand, there are various conventional methods to assess the interactions between biological samples as described in the Introduction. Table 2 shows the reported analysis time and reagent consumption using conventional methods and the proposed PF–NEACE. With respect to the analysis time, the proposed method and NECEEM were shorter than those of the other methods. These methods also have a great advantage in that they do not require the immobilization of a sample/ligand, whereas SPR and QCM do. ITC also measures the interactions without any immobilization process, but it requires a large amount of samples/reagents and complicated procedures. Estimations of the interactions between DNA and a target molecule using magnetic nanoparticles (MNP) have been also reported [22]. The MNP method is very easy to use but requires a long analysis time and a large consumption of reagents. In NECEEM, mixed solutions of samples and ligands reaching equilibrium were analyzed in CE within a short analysis time. However, NECEEM needs to wait until the mixed solutions reach equilibrium and prepare tens of microliters of each sample solution for use in the CE apparatus. PF–NEACE also requires at least several tens of microliters of sample and ligand solutions in the vials, but only a small portion of them (several tens of nanoliters) is introduced into the capillary to measure the interaction. Thus, PF–NEACE saves both the mixing time and volume of the sample and ligands. Conventional PF–ACE also provides interaction assays with minimal consumption of samples/reagents. However, PF–ACE often requires a higher concentration of the ligand than that of the target to maintain the stable complexation of the whole target molecules, which sometimes interferes with the measurement of rare, dilute, and difficult to obtain ligands such as biomolecules from patients. On the other hand, PF–NEACE is applicable to the condition that the concentration of ligands is lower than that of targets as described above. Therefore, high-throughput screening of disease-related oligonucleotide-target drugs is expected to be realized by PF–NEACE.

Table 2 Comparison of PF–NEACE to conventional interaction assays


PF–NEACE was proposed to screen interactions between drug candidates and oligonucleotides. The interactions between the Tyr-Am aptamer (Apt49-T) and Tyr-Am analogs were evaluated by combining kinetic and equilibrium analyses based on equilibrium and moment theories. As a result, the specific interactions between Apt49-T and Tyr-Am or Phe-Am could be evaluated from the results of PF–NEACE, indicating the applicability of the proposed method for screening the affinity between oligonucleotides and drag candidates. The developed method was also applied to analyze the interactions between disease-related oligonucleotides and reported drug candidates. As a result, the specific binding of multiple molecules of compound A with each RNA and the effect of the addition of KCl on the interactions were confirmed by PF–NEACE. Compared to conventional interaction assays, PF–NEACE can provide a high-throughput assay with a minimal consumption of reagents, which will contribute to drug screening for which very few samples and reagents can be obtained from patients.


  1. H. Bhardwaj, G. Sumana, C.A. Marquettea, Food Chem. 307, 125530 (2020)

    CAS  Article  Google Scholar 

  2. B. Nguyen, F.A. Tanious, W.D. Wilson, Methods 42, 150–161 (2007)

    CAS  Article  Google Scholar 

  3. A. Litovchick, X. Tian, M.I. Monteiro, K.M. Kennedy, M. Guié, P. Centrella, Y. Zhang, M.A. Clark, A.D. Keefe, Molecules 24, 2026 (2019)

    CAS  Article  Google Scholar 

  4. G. Bayramoglu, C. Ozalp, M. Oztekin, U. Guler, B. Salih, M.Y. Arica, Talanta 191, 59–66 (2019)

    CAS  Article  Google Scholar 

  5. A. Osypova, D. Thakar, J. Dejeu, H. Bonnet, A.V. Heyden, G.V. Dubacheva, R.P. Richter, E. Defrancq, N. Spinelli, L. Coche-Guérente, P. Labbé, Anal. Chem. 87, 7566–7574 (2015)

    CAS  Article  Google Scholar 

  6. V.C. Ozalp, Analyst 136, 5046–5050 (2011)

    Article  Google Scholar 

  7. Z. Wang, H. Yu, J. Canoura, Y. Liu, O. Alkhamis, F. Fu, Y. Xiao, Nucleic Acids Res. 13, e81 (2018)

    Article  Google Scholar 

  8. A. Das, K. Bhadra, G.S. Kumar, PLoS ONE 8, e23186 (2011)

    Article  Google Scholar 

  9. C. Zhang, A.G. Woolfork, K. Suh, S. Ovbude, C. Bi, M. Elzoeiry, D.S. Hage, J. Pharm. Biomed. Anal. 177, 112882 (2020)

    CAS  Article  Google Scholar 

  10. N. Kanayama, T. Takarada, A. Kimura, H. Shibata, M. Maeda, J. Sep. Sci. 31, 837–844 (2008)

    CAS  Article  Google Scholar 

  11. M.M. Dinges, K. Solakyildirim, C.K. Larive, Electrophoresis 35, 1469–1477 (2014)

    CAS  Article  Google Scholar 

  12. M. Berezovski, A. Drabovich, S.M. Krylova, M. Musheev, V. Okhonin, A. Petrov, S.N. Krylov, J. Am. Chem. Soc. 127, 3165–3171 (2005)

    CAS  Article  Google Scholar 

  13. S. Kochmann, A.H. Le, R. Hili, S.N. Krylov, Electrophoresis 39, 2991–2996 (2018)

    CAS  Article  Google Scholar 

  14. Y. Sasaki, Y. Sato, T. Takahashi, M. Umetsu, N. Iki, Anal. Biochem. 585, 113406 (2019)

    CAS  Article  Google Scholar 

  15. V. Villareal, J. Kaddis, M. Azad, C. Zurita, I. Silva, L. Hernandez, M. Rudoluph, J. Moran, F.A. Gomez, Anal. Bioanal. Chem. 376, 822–831 (2003)

    CAS  Article  Google Scholar 

  16. Y. Tanaka, Chromatography 23, 13–23 (2002)

    CAS  Google Scholar 

  17. E. Farcas, C. Bouckaert, A.C. Servais, J. Hanson, L. Pochet, M. Fillet, Anal. Chim. Acta 984, 211–222 (2017)

    CAS  Article  Google Scholar 

  18. K. Miyabe, M. Nakajima, Electrophoresis 42, 2528–2537 (2021)

    CAS  Article  Google Scholar 

  19. E. Vianini, M. Palumbo, B. Gatto, Bioorg. Med. Chem. 9, 2543–2548 (2001)

    CAS  Article  Google Scholar 

  20. Z. Su, Y. Zhang, T.F. Gendron, P.O. Bauer, J. Chew, W. Yang, E. Fostvedt, K. Jansen-West, V.V. Belzil, P. Desaro, A. Johnston, K. Overstreet, S. Oh, P.K. Todd, J.D. Berry, M.E. Cudkowicz, B.F. Boeve, D. Dickson, M.K. Floeter, B.J. Traynor, C. Morelli, A. Ratti, V. Silani, R. Rademakers, R.H. Brown, J.D. Rothstein, K.B. Boylan, L. Petrucelli, M.D. Disney, Neuron 83, 1043–1050 (2014)

    CAS  Article  Google Scholar 

  21. K. Nakatani, S. Sando, I. Saito, Nat. Biotechnol. 19, 51–55 (2001)

    CAS  Article  Google Scholar 

  22. X. Ma, W. Wang, X. Chen, Y. Xia, N. Duan, S. Wu, Z. Wang, Food Control 47, 545–551 (2015)

    CAS  Article  Google Scholar 

Download references


The authors greatly appreciate Prof. Kanji Miyabe (Rikkyo University, Tokyo, Japan) for his advice on the arrangement of the moment theory. This study was partially supported by Sumitomo Dainippon Pharma Co., Ltd.: Joint Research Program, Partnership to Realize Innovative Seeds and Medicine (PRISM). This study was partially supported by JST, PRESTO, Japan (Grant Number JPMJPR19H7).

Author information

Authors and Affiliations


Corresponding author

Correspondence to Kenji Sueyoshi.

Ethics declarations

Conflict of interest

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Mitsuno, E., Endo, T., Hisamoto, H. et al. Evaluation of the interactions between oligonucleotides and small molecules by partial filling–nonequilibrium affinity capillary electrophoresis. ANAL. SCI. 38, 851–859 (2022).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


  • Affinity capillary electrophoresis
  • Moment analysis
  • Nonequilibrium capillary electrophoresis
  • Partial-filling technique
  • Partial filling–nonequilibrium affinity capillary electrophoresis