Sensitive and specific quantification of antisense oligonucleotides using probe alteration-linked self-assembly reaction technology

Pharmacokinetics, especially the quantification of drug concentrations in biological samples such as plasma and tissues, is extremely important in drug development during both the nonclinical and clinical stages. Antisense oligonucleotides (ASOs) have been actively researched in recent years for use as therapeutics that selectively regulate the expression of genes involved in disease pathogenesis [1]. ASOs act through sequence-specific binding of target RNAs and inhibition of gene expression [2]. A variety of biological assays have been developed to help accelerate the development of ASOs. However, although advances have been reported with the evolution of the structural and pharmacokinetic properties of ASO drugs, developing methods for higher sensitive quantification, to allow the use of low doses of ASOs that can address toxicity caused by ASO structure and administration routes, remains challenging.

Bioanalytical methods for oligonucleotide drug quantification like ASOs mainly include LC–MS/MS [3–7] and hybridization-based assays (also known as ligand-binding assay or hybridization-ELISA) [6,8–16]. LC–MS/MS is less sensitive than most hybridization assays but can specifically quantify full-length ASOs and their metabolites [6,7]. In addition to increased sensitivity, one of the main advantages of hybridization assays is the simple pretreatment requirement that is economical and does not require the extraction of nucleic acids or a lot of time and effort [6].

Hybridization assays include those using enzymes (enzymatic hybridization assays), such as nucleases or ligases [8–14], and those not requiring enzymes (nonenzymatic hybridization assays) [15,16]. Enzymatic hybridization assays are highly efficient at detecting relatively short target oligonucleotides, such as siRNAs and ASOs, because they use a capture probe (CP) that captures approximately the entire target length. The specificity of the enzyme enables effective metabolite discrimination and crossreactivity with metabolites lacking a single base (3′n-1) is approximately 5–6% [9,13]. However, these methods are not without limitations. For instance, enzyme-based assays are strongly dependent on the specificity of the enzyme and sequence context [14]. Furthermore, the complexity of the reactions as well as the protocol, in general, can pose issues for researchers. For example, enzymatic reaction efficiency can be reduced or inhibited by matrix components and modified nucleic acids [17–22], which have been extensively developed in recent years.

Nonenzymatic hybridization assays, also known as dual-probe hybridization or sandwich hybridization assays, are simpler than enzymatic hybridization because the detection process consists of only the hybridization step. In addition, as they are enzyme-independent, nonenzymatic assays are relatively easy to adapt to various inhibitory components contained in biological samples. However, these methods are not particularly efficient at discriminating n-1 metabolites [15]. To the best of our knowledge, there are no reports of nonenzymatic hybridization-based bioanalytical quantification methods with both high sensitivity and high specificity for the detection of ASOs in bioanalytical samples. Therefore, in this study, the probe alteration-linked self-assembly reaction (PALSAR) short probe assay (PALSAR-SPA) was developed and validated to solve the inherent problems of the existing bioanalytical methods described above. We adopted PALSAR technology, a signal amplifier, to achieve high sensitivity and verified that the appropriate probe design is effective for the high specificity of metabolites. Signal amplification of PALSAR technology is achieved via the self-assembly of a pair of DNA probes, called honeycomb probes (HCPs), containing three complementary regions that bind to target nucleic acids [23]. In previous studies, PALSAR technology was applied to detect mRNA with high sensitivity [23]. This technique has also succeeded in sensitizing an antibody bridging assay as a signal amplifier [24]. Our combined method highlights new advances in the bioanalytical quantification of ASOs based on nonenzymatic hybridization assays that address both high sensitivity and specificity.

Materials & methods

Target ASOs & probes

Table 1 shows the sequences of the target ASOs, detection probes and HCPs. Two types of 15-mer ASOs with a common nucleic acid sequence were used as targets. ASOs had a general gapmer structure. Both ASOs contained a phosphorothioated gapmer structure in the central region. The three bases at both ends of the wing region were modified by locked nucleic acid (LNA; LX-A0236), amido-bridged nucleic acid and 2′-O, 4′-C-spirocyclopropylene bridged nucleic acid (scpBNA) (LX-A0237) [17,18]. The ASOs modified by LNA, amido-bridged nucleic acid and scpBNA were synthesized by conventional solid-phase synthesis using corresponding phosphoramidites. Unless otherwise specified, the probes used in the experiments were a seven-mer long CP with three DNA to LNA substitutions (7 m-L3) and an eight-mer long assist probe (AP) with all DNA substituted by LNA. LNA insertion increases the melting temperature and affinity of the oligonucleotide [25,26]. Other CPs and APs were used to verify the balance of sensitivity and specificity in probe design. These probes and HCPs were synthesized by Nihon Gene Research Laboratories, Inc. (Sendai, Japan). CP was chemically bound to solid-phase microspheres (MicroPlex Microspheres; Luminex Japan Co., Ltd, Tokyo, Japan, catalog no. LC10015-01) according to the instructions provided by Luminex Japan Co., Ltd.

Preparation of biological matrices

The plasma and tissues (mouse liver, kidney, brain and muscle) of C57BL/6J mice were purchased from The Jackson Laboratory Japan, Inc. (Yokohama, Japan). Mouse plasma was centrifuged at 1700×g and 4°C for 10 min, and the supernatant was used as a biological matrix to spike in ASOs. The lysis buffer for mouse tissue was adjusted by adding Proteinase K (Kanto Chemical, Tokyo, Japan, catalog no. 34060-96) to Direct PCR Lysis Reagent (Viagen Biotech, CA, USA, catalog no. 101-T) at a final dilution of 1:100. Each tissue (liver, kidney, brain and muscle) was shredded into 2–5 mm squares and combined with the adjusted lysis buffer to obtain a concentration of 200 mg/ml. Each tissue fragment was dissolved in a heat block at 55°C for 16–20 h (800 r.p.m.). After lysis, Proteinase K was inactivated at 85°C for 45 min (800 r.p.m.). The lysate was then diluted tenfold with RNase-free water to prepare the tissue homogenate that was used as a biological matrix to spike in ASOs.

First hybridization

The first hybridization was carried out with 35 μl of a mixed solution of CP, AP and target ASO at 40°C for 5 min and subsequently at 25°C for 1–16 h. The composition of the reaction solution for the first hybridization was adjusted to a total volume of 35 μl, with a final concentration of 1.5 × supplement (50 mM Tris-HCl/pH 8.0 [Thermo Fisher Scientific, MA, USA, catalog no. 15568-025], 0.8% N-lauroyl sarcosine [Sigma-Aldrich, MO, USA catalog no. L7414-50ML], 4 mM EDTA/pH 8.0 [Thermo Fisher Scientific, catalog no. 15575-038]), 1.5 M Tetramethylammonium chloride (Sigma-Aldrich, catalog no. T3411-50ML), 100 fmol or 1 pmol AP, and 500 microspheres coupled with CP. The ASO used in each experiment was diluted in 0.01% Tween water or various biological matrices to obtain each concentration and 10 μl was added to each reaction.

Probe alteration-linked self-assembly reaction

The PALSAR solution was prepared with 1.6 × supplement, 1.5 M Tetramethylammonium chloride, 35 pmol HCP-1 and 31.5 pmol HCP-2 in 50 μl. After the first hybridization reaction, 15 μl of the PALSAR solution was added to each reaction mixture containing 500 microspheres. The PALSAR reaction was carried out with a total volume of 50 μl at 25°C for 60 min. After PALSAR, the reaction solution was centrifuged at 1000× g for 1 min to remove the supernatant. Next, 100 μl of 1 × wash buffer (1 × PBS [Nippon Gene, Tokyo, Japan, catalog no. 314-90815], 0.02% Tween 20 [Calbiochem, Darmstadt, Germany, catalog no. 655206] and 1.5 p.p.m. ProClin 300 [Sigma-Aldrich, catalog no. 48912-U]) (1 × PBS-TP) was added to the microspheres and centrifuged at 1000× g for 1 min. The supernatant was then removed.

Detection

After the removal of the supernatant, 50 μl of a solution diluted with 1 × PBS-TP to 5 μg/μl of streptavidin-phycoerythrin conjugate (SA-PE) (Agilent Technologies Japan, Tokyo, Japan, catalog no. PJ31S) was added to the microspheres and stirred. The mixture was lightly spun down, shielded from light and allowed to stand for 10–60 min to react with the labeled biotin on HCP with SA-PE. The mixture was then centrifuged at 1000× g for 1 min at 20–25°C. The supernatant was removed and the microspheres were washed twice with 100 μl of 1 × PBS-TP. Finally, 75 μl of 1 × PBS-TP was added and vortexed. The Luminex 100/200 System (Luminex Japan Co., Ltd, Tokyo, Japan) was used for fluorescence detection. Special Luminex system-specific microspheres stained with fluorescent dye were used to detect microsphere-specific fluorescence and fluorescence signals by SA-PE using two onboard lasers (excitation wavelengths: 635 nm and 532 nm). The reacted microspheres (100 beads) were counted and the median fluorescence intensity (MFI) was measured. All samples were measured in duplicate from which the average positive signal was calculated. The positive signal is shown as a signal by subtracting the background (BG) from MFI. The accuracy and precision of the analysis were assessed by back-calculating concentrations from logarithmic plots of ASO concentrations and positive signals. Accuracy is determined by measuring five quality controls (QCs) extrapolated from the standard curve and expressed as the percentage error or bias. The precision is defined as the percentage CV.

Results & discussions

Sensitivity improvement of conventional assays by PALSAR technology

Figure 1 shows a schematic diagram of PALSAR-SPA applying PALSAR technology for ASO detection. ASOs were captured by CP and AP. Conventional hybridization assays used biotin-labeled AP for signal detection, whereas PALSAR-SPA used AP with polyadenine sequences as a scaffold for binding a pair of HCPs (HCP-1 and HCP-2). Through these polyadenine sequences, signal amplification was achieved by self-assembly of multiple biotin-labeled HCPs. The detection performance of PALSAR-SPA and a conventional hybridization assay were compared in mouse plasma. Figure 2 summarizes the dose–response curves in the range of 1.5–150 pg/ml. In the conventional assay, one AP acquired a signal with one biotin, whereas in PALSAR-SPA, multiple HCPs labeled with biotin on one AP contributed to signal acquisition. As a result, by incorporating PALSAR technology, the signal increased by approximately tenfold compared with that in the conventional hybridization assay, clearly demonstrating the effect of signal amplification by PALSAR technology. In addition, comparing the signal/noise (S/N) of PALSAR-SPA and the conventional assay in the low concentration range in Figure 2, the S/N of 1.5 pg/ml of the ASO concentration was 1.6 in the conventional assay and 2.8 in the PALSAR-SPA. In the intermediate concentration range of 6 pg/ml of the ASO concentration, the S/N of these assays were 3.3 and 8.5, respectively (data not shown). These results suggest that PALSAR-SPA can measure ASO more stably at lower concentrations than the conventional method.

Quantitative range of ASO by PALSAR-SPA

The quantification ranges of LX-A0236 and LX-A0237 with different modified nucleic acids in mouse plasma were investigated (Table 2). The limit of quantification was 1.5 g/ml for LX-A0236 and 6 pg/ml for LX-A0237, indicating a high sensitivity of several pg/ml even in ASO with different modified nucleic acids. In addition, the R² value was over 0.999.

Specificity performance of PALSAR-SPA in mouse plasma

The specificity performance of PALSAR-SPA with metabolites in mouse plasma was investigated. Full-length LX-A0236 (n) and metabolites containing a single-base deletion at the 3′ or 5′ ends (3′n-1 or 5′n-1) were measured at concentrations in the quantitative range of 1.5 pg/ml (lower limit of quantification [LLOQ]), 15 pg/ml and 150 pg/ml (upper limit of quantification [ULOQ]), respectively, and their crossreactivity examined. For the CP design, the 5′ end was coupled to the microsphere. Using this CP design, the crossreactivity of 3′n-1 was reduced to less than 1% at each ASO concentration (Table 3A), indicating that metabolites differing by one base were rarely detected. Conversely, the crossreactivity of the 5′n-1 metabolite was high, exceeding 88% (Table 3A). In this study, we focused on CP to investigate whether it is possible to distinguish 3′n-1 metabolites only by designing probes without using enzymes. Therefore, the AP was intentionally not designed to discriminate 5′n-1 metabolites. This high 5′n-1 crossreactivity is likely due to the placement of the 5′n-1 single-base deletion site on the AP side, where all DNA was substituted by LNA. Taken together, these findings suggest that crossreactivity suppression requires consideration of the single-base deletion side during probe design. Next, we mixed the 3′n-1 metabolite with the full-length LX-A0236 (n) at concentration ratios of 0–100% and investigated the effect of the full-length LX-A0236 (n) on the standard curve in the presence of 3′n-1 (Table 3B). Our data shows that even when the abundance ratio of the 3′n-1 metabolite was 100%, MFI was nearly the same as that of BG. Furthermore, when 3′n-1 was mixed with LX-A0236 (n) of 1.5 pg/ml (LLOQ) or 150 pg/ml (ULOQ) at several mixing ratios, the accuracy of the full-length LX-A0236 (n) standard curve was 100.5–124.7% and the measured MFI of the full-length LX-A0236 (n) was almost completely unaffected. These results indicate that when used with a suitable designed CP probe, the PALSAR-SPA is highly specific and sensitive for the quantitative measurement of full-length ASOs in the presence of their 3′n-1 metabolite.

Attaining a balance between sensitivity and specificity is an important challenge in the development of bioanalytical methods for ASO quantification. In selecting and using CPs of seven-mer length with three-LNA incorporation (7 m-L3) in this study, with respect to CP design, we examined how the number of LNAs and length of CP affects sensitivity and specificity (Figure 3). The design of CP was from six to ten mer (10 m, 9 m, 8 m, 7 m and 6 m) in length, and from zero to four LNAs (L0, L1, L2, L3 and L4) were appropriately incorporated. The detection signal (MFI) and S/N of full-length ASO were first compared. BG tended to increase as the number of LNA insertions increased. Therefore, in 7 m-L4, although MFI was high, the S/N decreased because BG was also high (Figure 3A). Furthermore, crossreactivity between full-length ASO and its 3′n-1 metabolite was examined using these CPs. The results of crossreactivity with 3′n-1 metabolites showed suppression of crossreactivity to less than 1%, except for 10 m-L0 (Figure 3B). The CP design that enables high specificity while maintaining MFI and S/N was shown to be 9 m-L0, 8 m-L1 and 7 m-L3. In this study, 7 m-L3 with a high signal value was selected.

Taken together, these results indicate that the probe design itself can achieve higher specificity and that metabolite discrimination is possible in nonenzymatic hybridization assays.

Intra-/interday accuracy & precision

To evaluate the intra-assay accuracy and precision of PASLAR-SPA, the ULOQ at 600 pg/ml, maximum QC at 300 pg/ml, middle QC at 60 pg/ml, low QC at 15 pg/ml and LLOQ at 6 pg/ml were established in mouse plasma as a spike in QC samples of LX-A0237.

Similar to the plasma sample, the dilution series of spiked QC samples in mouse tissue homogenates (the liver, kidney, brain and muscle) according to the respective quantitation ranges was also established. QC spikes in samples were prepared in triplicate, and each concentration was measured in duplicate (duplicate n = 3; total n = 6) to evaluate the intra-assay accuracy and precision. Furthermore, interassay variation was evaluated over three different days (total n = 18). The intra- and interassay accuracies in mouse plasma were 98.0–106.1% and 95.0–103.5%, and the precision was 4.6% or less and 12.1% or less, respectively (Table 4). In addition, the intra- and interassay accuracies for the four types of tissue homogenates were 86.8–119.1% and 88.1–113.1%, and the precision was <15.2 or <17.2%, respectively (Table 5). These results met the US FDA's Bioanalytical Method Validation Guidance for Industry [27] criteria and were considered satisfactory.

Overall, our results demonstrate that the challenge (the suppression of metabolite crossreactivity) in nonenzymatic hybridization assays can be effectively addressed by probe design without compromising sensitivity and accuracy. In nonenzymatic hybridization assays, the length of the probe is generally short and there is a tendency for problems in maintaining sensitivity and accuracy. However, with the use of appropriate probes and PALSAR technology, we have demonstrated that in addition to high specificity, high sensitivity and high precision at several pg/ml levels can be stably maintained.

Conclusion

Using PALSAR technology as a signal amplifier, we achieved ten-times more sensitization than conventional hybridization assays and provided a method to quantify ASO with high sensitivity and precision in several biological samples. In addition, we demonstrated an approach to suppress crossreactivity to less than 1% without compromising specificity, which has long been difficult in nonenzymatic hybridization assays. In the drug development process, it is not uncommon to have to change bioanalytical methods due to various problems. The approach presented here supplies a relatively simple alternative yet enables simultaneous access to sensitivity and specificity, which has been difficult to achieve in existing diversified bioanalytical methods. We expect that our new bioanalytical tool will aid in the assessment of preclinical pharmacokinetics.

Future perspective

Our approach in PALSAR-SPA may be useful for pharmacokinetic studies of oligonucleotide drugs, such as GTI-2040, which are known to be metabolized mainly through the 3′ end with a minor metabolism at the 5′ end [28]. In this study, PALSAR-SPA had no specificity for 5′n-1 where the deletion site is located on the AP side. This is because we have not optimized the AP design, and we are currently investigating the design of AP to address a wide range of drug metabolic pathways and approach the simultaneous discrimination of both metabolites (3′n-1 and 5′n-1). Furthermore, the development of an assay that detects only this pure, full-length ASO (n) may enable quantification of the n-1 metabolite in a nonenzymatic hybridization method. That is, quantification of the n-1 metabolite may be possible by measuring the sum (n plus n-1) using a conventional probe that also detects the n-1 metabolite and deriving a difference from the measurement value of the full-length ASO only. To respond to ASO with an arbitrary sequence, we also believe that it is necessary to examine whether a specific base placed at the end (G and C, or A and T) affects the distinction of metabolites and sensitivity. Dynamic range is also one of the parameters of bioanalysis development; therefore, the verification of combinations with instruments with a wider dynamic range of detection, such as electrochemiluminescence, is important for developing better bioanalytical methods. Additionally, the evaluation of pretreatment methods suitable for the physical and chemical properties of various oligonucleotide drugs is also an important issue to consider.