Increasing the Analytical Sensitivity by Oligonucleotides Modified with Para- and Ortho-Twisted Intercalating Nucleic Acids – TINA

The sensitivity and specificity of clinical diagnostic assays using DNA hybridization techniques are limited by the dissociation of double-stranded DNA (dsDNA) antiparallel duplex helices. This situation can be improved by addition of DNA stabilizing molecules such as nucleic acid intercalators. Here, we report the synthesis of a novel ortho-Twisted Intercalating Nucleic Acid (TINA) amidite utilizing the phosphoramidite approach, and examine the stabilizing effect of ortho- and para-TINA molecules in antiparallel DNA duplex formation. In a thermal stability assay, ortho- and para-TINA molecules increased the melting point (Tm) of Watson-Crick based antiparallel DNA duplexes. The increase in Tm was greatest when the intercalators were placed at the 5′ and 3′ termini (preferable) or, if placed internally, for each half or whole helix turn. Terminally positioned TINA molecules improved analytical sensitivity in a DNA hybridization capture assay targeting the Escherichia coli rrs gene. The corresponding sequence from the Pseudomonas aeruginosa rrs gene was used as cross-reactivity control. At 150 mM ionic strength, analytical sensitivity was improved 27-fold by addition of ortho-TINA molecules and 7-fold by addition of para-TINA molecules (versus the unmodified DNA oligonucleotide), with a 4-fold increase retained at 1 M ionic strength. Both intercalators sustained the discrimination of mismatches in the dsDNA (indicated by ΔTm), unless placed directly adjacent to the mismatch – in which case they partly concealed ΔTm (most pronounced for para-TINA molecules). We anticipate that the presented rules for placement of TINA molecules will be broadly applicable in hybridization capture assays and target amplification systems.


Introduction
The stability of double-stranded DNA (dsDNA) is naturally limited to allow cellular processes that require helix dissociation such as gene transcription, gene regulation and cell division. However, the sensitivity of DNA diagnostic assays depends upon the stability of dsDNA helices. The analytical sensitivity of an assay can be improved by decreasing stringency, but at the risk of crossreactivity to other targets.
In addition, we report the synthesis of a novel ortho-TINA amidite using the phosphoramidite approach. Until now, ortho-TINA containing oligonucleotides has been synthesized by postsynthetic oligonucleotide modification using the Sonogashira Pd-catalyzed coupling reaction [12,16]. Although this approach is advantageous for the screening of different intercalators [14], to achieve sufficiently high coupling yields the reaction has to be repeated several times with fresh portions of the Sonogashira mixture. This can affect the subsequent oligonucleotide purification process. The phosphoramidite approach permits the production of a large number of oligonucleotides with several ortho-TINA molecule insertions in the sequences.
We find that inclusion of paraas well as ortho-TINA molecules in an oligonucleotide is capable of improving the analytical sensitivity of probe hybridization without increasing crossreactivity in a competitive antiparallel duplex hybrization capture assay. We anticipate that TINA molecules will enable a general improvement in the performance of future clinical diagnostic assays based upon conventional hybridization, as well as Polymerase Chain Reaction (PCR) and other primer-based enzymatic target amplification systems.

Synthesis of ortho-TINA amidite
As an alternative to the traditional method of postsynthetic oligonucleotide modification, we prepared the ortho-TINA monomer for use on a DNA synthesis platform via the more convenient phosphoramidite approach. The newly designed ortho-TINA phosphoramidite was synthesized in two steps from a known starting compound [12].
Full details of the synthesis procedure are provided in the materials and methods section, and in Supplementary Data S1. In brief (Figure 2), the starting compound (3) was prepared (80% overall yield) in three steps from commercially available compounds S-(+)22,2-dimethyl-1,3-dioxalane-4-methanol (1) and 2-iodobenzylbromide (2). In the first step of the ortho-TINA phosphoramidite synthesis, 1-ethynylpyrene was coupled to compound 3 using the Sonogashira coupling mixture [14]. To eliminate oxygen, the reaction mixture was degassed with nitrogen prior to the addition of tritylated compound 3; when the reaction mixture was not degassed, the product yield decreased significantly. DMT-protected ortho-TINA (4) was obtained as yellow foam (85% yield), and its structure was confirmed by NMR spectrometry. Finally, the secondary hydroxyl group was phosphatized. Signals in the 31 P NMR spectrum with chemical shifts of 148.9 and 149.3 ppm, respectively, confirmed the formation of the phosphoramidite (5).
Thermal stability of para-and ortho-TINA modified oligonucleotides To determine the optimal placement of paraand ortho-TINA molecules for stabilizing antiparallel DNA duplexes, we used a fluorescence resonance energy transfer (FRET) based high-speed melting curve method, as described and validated previously [17,18]. An 18-mer oligonucleotide from the Escherichia coli (E. coli) rrs gene (base pair 772-789) was used as the target. Table 1 shows the melting points (Tm) of paraand ortho-TINA modified oligonucleotides and changes in Tm associated with mismatches in the target strand (DTm). The full data set can be found in Table  S1. Orthoand para-TINA molecule insertions in the oligonucleotide increased Tm when placed terminally on the oligonucleotide, although the para-TINA molecule produced the greater increase. Maximum stability was reached when there was a modification at both termini. Placed internally, para-TINA molecules decreased Tm in all positions, especially when at the center of the oligonucleotide, whereas the positive effect of ortho-TINA molecules on Tm was neutralized towards the center of the oligonucleotide. The combination of a terminal paraor ortho-TINA molecule with an internal paraor ortho-TINA molecule showed the highest increases in Tm when the two modifications were separated by six or twelve nucleotides, equaling a half or complete helix turn. Both ortho-TINA and (especially) para-TINA molecules were found to partly conceal the DTm of a mismatch immediately next to them, but when the mismatch was moved one or more nucleotides away, they had no effect on DTm. The stabilizing effect of paraand ortho-TINA molecules increased when the oligonucleotide sequence was shortened from eighteen to sixteen nucleotides.   Effect of ionic conditions on dsDNA E. coli rrs gene PCR product capture by para-and ortho-TINA containing oligonucleotides Until now, the effects of TINA molecules have only been evaluated by Tm analyses, which are good model systems, but do not provide information on how TINA-modified oligonucleotides will perform as competitive annealing probes. To address this issue, we used the LuminexH 200 TM instrument to analyze the capture of denatured biotinylated E. coli rrs PCR product by magnetic microspheres coated with oligonucleotide sequences targeting base pairs 772-789 from the E. coli rrs gene. Figure 3 and Figure S1 show the capture of biotinylated rrs PCR product (in two-fold dilution series from 2.5 mL to 0.0098 mL rrs PCR product) by unmodified DNA oligonucleotides and oligonucleotides terminally modified with paraor ortho-TINA molecules in buffers of increasing ionic strength (100-1,000 mM monovalent cation). The overall level of median fluorescence intensity (MFI) was generally higher at greater ionic strength. In 150 mM buffer, the ortho-TINA modified oligonucleotide increased the analytical sensitivity 27-fold and the para-TINA modified oligonucleotide increased the analytical sensitivity seven-fold, compared with the unmodified DNA oligonucleotide. In 300 mM buffer, ortho-TINA modified oligonucleotide increased analytical sensitivity eleven-fold and para-TINA modified oligonucleotide six-fold, and even at 1,000 mM, a four-fold increase in analytical sensitivity was observed with both modified oligonucleotides compared with the unmodified equivalent.
To ensure that the increased analytical sensitivity was target sequence independent, the capture sequence was changed to base pairs 446-463 of the E. coli rrs gene. The corresponding sequence from Pseudomonas aeruginosa (P. aeruginosa) rrs gene is the most closely related sequence among the human pathogens. Consequently, P. aerugionsa was used as cross-reactivity control and contains a cluster of four mismatches to the E. coli sequence. A helper oligonucleotide (targeting E. coli rrs gene base pairs 464-483) was also added to prevent secondary structure formation (not required for base pairs 772-789). Changing the target sequence did not change the capture curves for the unmodified DNA and ortho-TINA modified oligonucleotides, whereas para-TINA modified oligonucleotides did not perform as well as for the 772-789 base pair target. There was no cross-reactivity with the P. aeruginosa control sequence.
Effect of hybridization temperature on dsDNA E. coli rrs gene PCR product capture by para-and ortho-TINA containing oligonucleotides To investigate whether the modulating effect of TINA molecules was temperature specific, the DNA hybridization assay was repeated at annealing temperatures from 42-62uC at three different ionic strengths and with two different concentrations for the E. coli rrs gene 446-463 base pair target sequence. P. aeruginosa was used as a cross-reactivity control sequence. As shown in Figure 4, the relative MFI of the terminally modified orthoand para-TINA and unmodified DNA oligonucleotides remained unchanged between 42uC and 52uC (temperature used in the ionic experiments), with the modified oligonucleotides generally providing the highest MFI. Above 52uC the difference in MFI rapidly diminished due to loss of signal. As expected, the level of cross-reactivity with the P. aeruginosa oligonucleotides rose with increasing ionic strength as the annealing temperatures decreased.
Effect of unlabeled helper oligonucleotide on dsDNA E. coli rrs gene PCR product capture  Table 1. Cont.
helper nucleotides that prevent the formation of secondary structures in the RNA [19]. This is in contrast to 16S E. coli rRNA nucleotide 772-789 for which no secondary structure has been found. Accordingly, in the studies reported here, we included an unlabeled DNA helper oligonucleotide targeting E. coli rrs gene base pairs 464-483 when capturing the E. coli rrs gene base pair 446-463 sequence, to avoid formation of secondary structures in the denatured single stranded DNA. As shown in Figure 5, we also examined the individual effect of this DNA helper oligonucleotide on analytical sensitivity when targeting E. coli rrs gene base pairs 446-463. Addition of the helper oligonucleotide increased the analytical sensitivity of the unmodified DNA and orthoand para-TINA modified oligonucleotides by approximately two-fold. As shown in earlier experiments (Figure 3), targeting base pair 446-463 with TINA/DNA modified oligonucleotides plus the helper nucleotide (to relieve secondary structure), gave similar levels of capture sensitivity to those obtained when targeting base pair 772-789 (no secondary structure).

Discussion
In the current paper, we have characterized the stabilizing effect and established design rules for placement of orthoand para-TINA molecules into Watson-Crick based antiparallel DNA duplexes. According to thermal stability analyses, both paraand ortho-TINA molecules should be placed terminally in the nucleotide sequence, and preferably on both the 59 and 39 terminal positions to achieve a maximum increase in Tm. Placement of para-TINA molecules at the 59 and 39 termini gave the most pronounced increase in Tm compared to ortho-TINA molecules. The stabilizing effect of paraand ortho-TINA molecules changes when they are placed internally in the oligonucleotide sequence. Ortho-TINA molecules have either a positive effect or no effect on Tm, whereas para-TINA molecules decrease Tm when placed internally. However, neither paranor ortho-TINA molecules interfere with mismatchinduced DTm, unless they are placed internally directly adjacent to the mismatch. Overall, when several TINA molecules are placed in an oligonucleotide, the highest increase in Tm is observed if they are placed at the 59 and 39 terminal positions (preferable) or, if placed internally as well, with the modifications separated by a half or whole helix turn.
The present thermal stability study was done using a single target sequence (the E. coli rrs gene base pair 772-789). The validity of the design rules are therefore still to be established, but the design rules suggested in this paper are in concordance with previously published thermal stability data on nucleic acid intercalator molecules in other target sequences [11][12][13][14][15][16]20]. The design rules identified in this study are also identical to the design rules we established previously for placement of para-TINA molecules into Hoogsteen based parallel DNA triplex formations [18]. Since thermal stability data for a number of different nucleic acid intercalating molecules are in perfect agreement with the herein presented design rules [11][12][13][14][15][16]20], we speculate whether these design rules might represent general design rules for placement of intercalator molecules into Watson-Crick based antiparallel duplex and Hoogsteen type triplex formations.
Previously, para-TINA has been tested for triplex and quadruplex hybridization in cellular systems [21,22], but the present study is the first evaluation of paraand ortho-TINA molecules in antiparallel DNA duplex based hybridization capture assays. The pronounced increase in analytical sensitivity conferred by paraand ortho-TINA molecules in antiparallel DNA duplex hybridization is note-worthy, especially since the increased analytical sensitivity is seen for two different target sequences, with and without a helper oligonucleotide. In addition, the specificity of the signal is maintained without cross-hybridization under a wide range of ionic conditions (100 mM to 1 M monovalent cations).
As previously stated, the corresponding sequence from the P. aeruginosa rrs gene was used as a cross-reactivity control in the hybridization capture assay, as it is the most closely related sequence among the known human pathogens. This sequence contains a cluster of four mismatches to the E. coli sequence, so a closer related sequence would have been desirable from a pure ''cross-reactivity control'' point of view. However, we decided to use the P. aeruginosa rrs gene sequence as cross-reactivity control, since we wanted the capture of biotinylated PCR product in the hybridization capture assay to reflect the clinical diagnostics reality the most. So, the true impact of TINA molecules on oligonucleotide cross-reactivity is still to be established.
The E. coli rrs gene base pair 772-789 target sequence was used in both the thermal stability study as well as the antiparallel duplex based hybridization capture assay. In the thermal stability study, placement of para-TINA molecules at the 59 and 39 termini gave the most pronounced increase in Tm compared to ortho-TINA molecules, but for capture of denatured E. coli rrs PCR product the analytical sensitivity was highest for ortho-TINA modified oligonucleotides. The thermal stability study reflects the temperature at which the fluorescence signal is changing at the highest rate, whereas the analytical sensitivity established in the hybridization capture assay reflects the hybridization to the target sequence in competitive annealing with the complementary strand of the PCR product. So even though the para-TINA modified oligonucleotides caused the highest Tm, the ortho-TINA modified oligonucleotides performed better in the competitive annealing hybridization capture assay.
Since addition of ortho-TINA in particular to the oligonucleotides increases the analytical sensitivity, we expect that ortho-TINA molecules, in particular, will be beneficial for increasing sensitivity, without compromising target specificity, in future clinical diagnostic assays, based on target hybridization capture as well as in target amplification systems. An example could be placement of an ortho-TINA molecule at the 59 end of PCR primers to increase efficacy of primer annealing, and thereby the overall efficacy in quantitative as well as end-point PCR reactions.

Synthesis of ortho-TINA amidite
Solvents were dried prior to use. All chemicals were obtained from Sigma-Aldrich (Brøndby, Denmark) and were used as purchased. The silica gel (0.040-0.063 mm) used for column chromatography was purchased from Merck & Co Inc. (Whitehouse Station, NJ, USA). Solvents used for column chromatography were distilled prior to use.
NMR spectra were measured on a Varian Gemini 2000 spectrometer at 300 MHz for 1 H using TMS (d: 0.00) as an internal standard, at 75 MHz for 13 C using CDCl 3 (d: 77.0) as an 1-Ethynylpyrene coupling (iv) was accomplished using the Sonogashira coupling mixture [14]. The reaction mixture was degassed with nitrogen prior to addition of tritylated compound 3. DMT-protected ortho-TINA (4) was obtained as yellow foam, and its structure confirmed by NMR spectrometry. The second step (v) was also performed in an inert nitrogen atmosphere, in the dark at 0uC to RT. NMR spectrometry confirmed the forma- Oligonucleotides and fluorescence resonance energy transfer (FRET) system All oligonucleotides were purchased from IBA GmbH (Göttingen, Germany) or DNA Technology A/S (Risskov, Denmark) on a 0.2 mmol synthesis scale with high performance liquid chromatography (HPLC) purification and subsequently quality control.

Melting curve acquisition
Melting curve experiments were performed on a LightCyclerH 2.0 using 20 mL LightCyclerH capillaries. 0.5 mM of each oligonucleotide was mixed with sodium phosphate buffer (50 mM NaH 2 PO 4 /Na 2 HPO 4 , 100 mM NaCl and 0.1 mM EDTA) at pH 7.0. Tm measurements were carried out using a standard program: (i) dissociation at 37 to 95uC, ramp rate 0.2uC/ sec, 5 min hold at 95uC; (ii) annealing at 95 to 37uC, ramp rate 0.05uC/sec, continued measurement of fluorescence; (iii) 5 min hold at 37uC; and (iv) denaturation at 37 to 95uC, ramp rate 0.05uC/sec, and continued measurement of fluorescence. Tm was determined using fluorescence data from both the annealing and denaturation curves. No hysteresis was observed. Using Light-CyclerH Software 4.1 for melting curve analysis, Tm was defined as the peak of the first derivative. All melting curve determinations were conducted as single capillary measurements. A setup control (matching oligonucleotides D-624 and D-643) was included in all runs. Prior to Tm identification, runs were color compensated by subtraction of the fluorophore background fluorescence.

Coupling of oligonucleotides to LuminexH MagPlexH microspheres
Conventional DNA oligonucleotides were coupled to Mag-PlexH-C magnetic carboxylated microspheres following the carbodiimide coupling procedure for amine-modified oligonucleotides, as recommended by Luminex Corporation. In short, 2.5610 6 microspheres were activated in 0.1 M MES, pH 4.5, followed by addition of 0.2 nmol oligonucleotide and 25 mg EDC. The coupling reaction was incubated for 30 min in the dark, followed by addition of 25 mg EDC and another 30 min incubation. 1.0 mL of 0.02% Tween-20 was added and the supernatant was removed after magnetic separation for 1 min on a DynaMag TM -2 magnetic particle concentrator (Invitrogen A/S, Tåstrup, Denmark). 1 mL of 0.1% SDS was added and vortexed, followed by magnetic separation and resuspension in 100 mL Tris-EDTA buffer, pH 8.0, and refrigerated storage.
For orthoand para-TINA modified oligonucleotides, a novel inhouse carbodiimide/sulpho-NHS coupling procedure was followed. In a low retention microcentrifuge tube (Axygen, Union City, CA, USA), 2.5610 6 microspheres were washed and activated in 100 mL of 0.1 M MES, pH 6.0, then resuspended in 35 mL buffer. 125 mg sulpho-NHS was added, followed by 625 mg EDC, incubation in the dark for 15 min, addition of another 625 mg EDC and 15 min incubation. Activation buffer was removed and 97 mL of 0.1 M phosphate buffer, pH 7.2, was added followed by 0.3 nmol oligonucleotide. Microspheres were incubated for 2 hours at RT on a Thermo-shaker TS-100 (BioSan, Riga, Latvia) at 900 rpm, followed by optional overnight incubation, without shaking. Microspheres were washed once in 100 mL of 0.1 M phosphate buffer, pH 7.2, blocked in 0.1 M phosphate buffer with 50 mM ethanolamine, pH 7.2, and incubated for 15 min at RT on the Thermo-shaker at 900 rpm. Microspheres were separated and resuspended in 100 mL Tris-EDTA buffer, pH 8.0, and stored at 5uC. All separation steps involved placing the microcentrifuge tube in the magnetic separator for 1 min, with low speed vortexing for 20 sec after each addition of buffer or reagent.
To ensure equal coupling efficiency for the carbodiimide coupling procedure, and the carbodiimide/sulpho-NHS coupling procedure used for the orthoand para-TINA modified oligonucleotides, a biotinylated oligonucleotide with or without terminally para-TINA modifications was included in each coupling protocol. The coupling efficiency was evaluated by incubation of 0.2 mL microspheres with 0.5 mg Streptavidin-R-PhycoErythrin Premium Grade (S-21388, Invitrogen A/S) with 10 mg albumin fraction V (Merck & Co Inc.), 0.03% Triton X-100 and 10 mM phosphate buffer, pH 6.4 with 200 mM NaCl. The reaction mixture was incubated for 15 min in an iEMSH Incubator/Shaker HT (Thermo Fisher Scientific) at 25uC and 900 rpm. After three washes in 10 mM phosphate buffer, pH 6.4, with 200 mM NaCl and 0.03% Triton X-100, 350 microspheres were counted on the LuminexH 200 TM instrument. Similar coupling efficiencies were found using both procedures. Microspheres from a single coupling round were used in all experiments.
PCR products were pooled and purified using NucleoSpinH Extract II PCR clean-up (Macherey-Nagel GmbH). The purified product was evaluated by gel electrophoresis on a 1.5% agarose gel in TAE buffer with ethidium bromide staining with GeleRuler TM 100 bp Plus DNA Ladder (Fermentas GmbH, St. Leon-Rot, Germany). DNA concentration was 54.8 ng/mL, as determined by OD50 measurement on the NanoDrop TM 1000. The pooled PCR product was used in all experiments.
Biotinylated PCR products were detected on the LuminexH 200 TM instrument (Luminex Corp.). A 70 mL premix of microspheres, PCR product, Triton X-100 and helper oligonucleotide (for E.coli rrs gene base pair 446-463 capture) was mixed in an EppendorfH twin.tec 96-well PCR plate and incubated at 95uC for 10 min in a SensoQuest Labcycler (SensoQuest GmbH, Göttingen, Germany). The PCR plate was immediately transferred to ice for 2 min and 50 mL was transferred to a conical bottom 96 MicroWell TM Plate (NUNC, Thermo Fisher Scientific, Roskilde, Denmark) on ice, and 50 mL of a cold 2x hybridization buffer added. The final mixture consisted of 0.2 mL of the relevant microsphere (approximately 2,500 microspheres/well), a two-fold dilution series of biotinylated E. coli rrs gene PCR product from 2.5-0.0098 mL, 0.03% Triton X-100, and 1x hybridization buffer (20 mM NaH 2 PO 4 /Na 2 HPO 4 adjusted with NaCl to monovalent cation concentrations of 100, 150, 200, 300, 400, 500 and 1000 mM at pH 7.0 (52uC)). The mixture was incubated for 15 min in an iEMSH Incubator/Shaker HT (Thermo Fisher Scientific) at 900 rpm and 52uC, or at 42, 46, 50, 54, 58 or 62uC in the temperature experiments. After incubation, the plate was washed three times by using a 96-well magnetic separator (PerkinElmer, Skovlunde, Denmark), removing the supernatant, and adding 20 mM NaH 2 PO 4 /Na 2 HPO 4 adjusted with NaCl to 50 mM monovalent cation concentration and 0.03% Triton X-100 at pH 7.0. Next, 0.5 mg Streptavidin-R-PhycoErythrin Premium Grade (S-21388, Invitrogen A/S, Tåstrup, Denmark) with 10 mg albumin fraction V (Merck & Co Inc.), 0.03% Trition X-100 and 1x hybridization buffer, was added to each well. Plates were incubated for 15 min at 52uC (or relevant experimental temperature), and washed three times as previously described. Wash buffer was added, and incubated for 30 min at RT before LuminexH 200 TM analysis, counting 300 of each microsphere set. The final step at RT avoided decreasing background fluorescence in the LuminexH analysis due to sedimentation of unevenly sized microspheres [23]. All dilution series were run in triplicate, with results presented as mean of MFI and 95% confidence intervals. Analytical sensitivity was defined as the limit of detection (LOD), calculated by adding three standard deviations to the mean background MFI. Differences in analytical sensitivity were defined as the ratio between the LOD of DNA and orthoor para-TINA modified oligonucleotides.

Supporting Information
Table S1 Change in Tm and DTm of Watson-Crick based antiparallel duplexes stabilized by para (X)-and/or ortho (Y)-TINA monomers. Tm was determined using 0.5 mM of each strand in 50 mM phosphate buffer, pH 7.0, with 100 mM NaCl and 0.1 mM EDTA. Tm was defined as the peak of the first derivative using both annealing and dissociation curves. Base mismatches are underlined and marked in bold blue. *Mismatch adjacent to TINA. (XLS) Figure S1 Competitive annealing of orthoor para-TINA terminally modified oligonucleotides compared with unmodified DNA oligonucleotide to denatured PCR products in buffer of increasing ionic strength -complete data. E. coli rrs biotinylated PCR product was captured by unmodified DNA oligonucleotide