Properties of artificial cationic oligodiaminosaccharides and oligopeptides that bind to A-type oligonucleotide duplexes

Abstract A critical strategy to improve the properties of oligonucleotide therapeutics is using cationic molecules as carriers. We developed artificial cationic molecules that bind to A-type oligonucleotide duplexes, such as siRNAs, in a stoichiometric ratio. In this study, we investigated the properties of oligo 2,6-diamino-D-galactoses (ODAGals) and L-2,4-diaminobutanoic acid oligomers (Dabs) and revealed their thermal and biological stabilization effects on A-type duplexes and their chemical stability. As a result, ODAGal and Dab with the same number of amino groups had the commensurate ability for the biological stabilization effect, whereas Dab enhanced the thermal stability of A-type duplexes more effectively than ODAGal.


Introduction
In recent years, double-stranded oligonucleotide therapeutics with A-type duplex structures, such as short interfering RNAs (siRNAs) [1] and heteroduplex oligonucleotides (HDOs), [2,3] have gained considerable attention. Unmodified DNAs and RNAs are unstable in vivo and rapidly cleaved by nucleases. Therefore, it is essential to improve their in vivo stability while developing oligonucleotide therapeutics. Chemical modification of oligonucleotides is an effective method for improving biological stability, including phosphorus modification, such as phosphorothioate, and sugar modifications, such as 2′-OMe and 2′-O-MOE, are widely used to acquire substantial stability. [4] However, there are multiple challenges in developing oligonucleotide therapeutics, such as membrane permeability and delivery efficiency.
One strategy to improve oligonucleotide therapeutics' properties is using cationic molecules as carriers. Because nucleic acids are polyanionic compounds, they form non-covalent complexes with cationic molecules through electrostatic interactions. Various cationic compounds, including lipids, polymers, dendrimers, and sugars, bind nucleic acids and assist transfer into cells. One of the most widely used RNA carriers is the cationic lipid lipofectamine™, which is often used for RNAi experiments. [5] In addition, chitosan (poly-β-(1→4)-glucosamine), [6] a natural polysaccharide, and cell-permeable peptides (CPP) [7] consisting of basic amino acids, such as arginine and histidine, are also extensively used as nucleic acid carriers. These molecules have the advantage of being highly stable and are often easy to obtain. However, these cationic molecules could interact with negatively charged biomolecules in a nonspecific manner and cause cytotoxicity. [8] Existing carriers generally use an excess amount of oligonucleotide therapeutics to form complexes efficiently. Thus, developing carrier molecules that are efficiently stabilized with a small number of cations is highly required.
Under these circumstances, we have developed artificial cationic molecules that bind to A-type oligonucleotide duplexes and stabilize thermally and biologically in a stoichiometric ratio. One category of artificial molecules is oligodiaminosaccharide, in which a pyranose's 2 and 6-hydroxy groups were substituted with amino groups. [9][10][11] In particular, oligodiaminogalactoses (ODAGals) have a highly stabilizing effect on A-type double-stranded nucleic acids, including siRNAs and HDOs (Figure 1 (left)). [12,13] We recently synthesized up to hexasaccharide and showed that the longer saccharide effectively stabilized A-type duplexes. [14] We also reported the synthesis of ODAGal analogs conjugated with vitamin E for delivery of siRNA to hepatocytes. [15] The other category is cationic oligopeptides. Peptides with amino groups stabilized A-type duplexes, especially L-2,4-diaminobutanoic acid oligomers (Dabs) showed a notable stabilizing   (Figure 1 (right)). [16,17] The most characteristic feature of these artificial oligosaccharides and oligopeptides is that they can bind selectively to A-type nucleic acid duplexes, such as RNA/RNA and DNA/RNA duplexes, with high affinity and selectivity, without changes in duplex structures. This feature is attributed to the distance between adjacent positively charged amino groups: the distance is quite similar to the width of major grooves of A-type duplexes (7-14 Å in solution structures [18] ). Thus, these molecules are expected to bind to the major groove of an A-type duplex. This is supported by NOE correlation analysis using complexes of ODAGal4 and Dab8 and RNA/RNA duplex. [11,19] ODAGals and Dabs have similar structural features, but their flexibility is expected to differ. ODAGals have rigid pyranose ring structures, whereas Dabs have amino groups on flexible side chains. Although we have previously investigated the properties of each of these molecules, we have not directly compared them. We speculate that differences in flexibility would lead to differences in the stabilization of A-type duplexes. Thus, this study conducted a comprehensive investigation of the effect of ODAGals and Dabs on the properties of A-type duplexes to reveal their differences.
All measurements were conducted under physiological conditions using a 10 mM phosphate buffer containing 100 mM NaCl at pH 7.0. The UV melting curves of the duplexes were measured using 4 µM of 12mers and 16mers or 2.5 µM of 20mers with and without 1 or 2 equivalents of ODAGals and Dabs. Analytical samples were prepared by annealing the oligonucleotides and adding an aqueous solution of ODAGals or Dabs. The melting temperature (T m ) values were calculated from the intersection of the absorbance curve at 260 nm and the midline of the two tangents. Figures 2 and S1-S5 and Tables S1 show T m values for the various duplexes, and Figures 3, 4, and S6-S19 show the representative UV melting curves.
The addition of the cationic molecules increased the T m values of all the duplexes examined in this study. However, the T m values of some samples could not be calculated because of the deformation of the melting curve ( Figure 2 and Tables S1). The effect of increasing the T m value (ΔT m ) by adding the cationic molecules was more significant for nucleic acid duplexes with low T m values. In the absence of cationic molecules, the T m values of the phosphorothioate-modified nucleic acids were lower than the corresponding unmodified nucleic acids. On the other hand, when these cationic molecules were added, the ΔT m values were higher for the phosphorothioate-modified nucleic acids. These tendencies have been confirmed in the previous reports. [12] Also, the higher the molecular equivalent and the number of cations of a molecule, the higher the ΔT m value (Tables S1 and Figures S1-S5). Notably, there was a tendency that the presence of Dabs had a more pronounced effect on increasing the T m value of the duplex than that of the corresponding ODAGals. Also, the changes in the shape of the melting curves with the addition of ODAGals and Dabs were different (See Supplementary information). Although ODAGals and Dabs stabilize A-type oligonucleotide duplexes, there would be subtle distinctions in their binding, and the differences may be reflected in the melting curves.
In addition, we also investigated the formation of an aggregate of the nucleic acid duplex. We have reported that the existence of aggregates can be evaluated by focusing on the absorbance at 320 nm. [14] Generally, nucleic acids have negligible absorbance than the aggregates of nucleic acids and cationic molecules at 320 nm. Thus, in the thermal denaturation tests, the absorbance at 320 nm was recorded to elucidate the formation of aggregates (Figures 3B, D and 4B, D, and S6-S19 right). In the presence of ODAGal4 and ODAGal5, Dab8 and Dab10, there was negligible absorbance at 320 nm, indicating the addition of up to 2 equivalents of these molecules did not cause severe aggregation. On the other hand, in the presence of 2 equivalents of ODAGal6, the absorbance at 320 nm was detected for all the samples tested in this study, which was in good agreement with our previous report. [14] The addition of 2 equivalents of Dab12 also caused the absorbance at 320 nm with some oligonucleotide sequences. These phenomena were more remarkable with oligonucleotide with phosphorothioate modifications than with the unmodified counterparts. The order of absorbance at 320 nm caused by the addition of ODAGal6 and Dab12 was dependent on the phosphorus modification of a duplex, and it was indicated that unmodified and fully phosphorothioate duplexes were prone to form aggregates with ODAGal6 and Dab12, respectively. Furthermore, by comparing the change in the absorbance at 320 nm with increasing temperature, it was suggested that the aggregations caused by Dab12 were dissolved at a temperature of about 40 °C. In contrast, the aggregations caused by ODAGal6 tended to be resolved at a higher temperature. (Figure 4B and D, and Supplementary information). These behaviors might be attributed to the differences in the interaction of ODAGals and Dabs with oligonucleotide duplexes.

Thermal stability study of Dabs
We found an interesting phenomenon through the denaturation tests when the thermal denaturation test was conducted several times with the same sample. Namely, after the melting curve was measured in a forward direction (0 °C→90 °C, +0.5 °C/min), the experiment was carried out in a backward direction (90 °C→0 °C, −0.5 °C/min), followed by a forward direction (0 °C→90 °C, +0.5 °C/min). In the presence of ODAGals, the T m values of the duplex (Seq 2-1) was increased for the first and second forward direction measurement ( Figure 5 and Supplementary information). In sharp contrast, the T m values of the duplex were almost the same in the presence and absence of Dabs for the second forward direction measurement. We hypothesized that while ODAGals were relatively stable even at high temperatures, Dabs were unstable. Therefore, we investigated Dabs' chemical stability in various conditions (pH, buffer, and temperature). A hydrochloride or trifluoroacetic acid salt of Dab8 was used for the experiments. First, we conducted thermal stability tests in water. A 1 mM aqueous solution of Dab8 was heated at 30 °C, 40 °C, 60 °C, or 90 °C for 1 h, and the resulting solutions were analyzed by reverse-phase high-performance liquid chromatography (RP-HPLC). The HPLC profiles are shown in Figure  6 (left) and S24 (left). No peaks corresponding to decomposition products were detected even after heating at 90 °C for 1 h. These results suggested  Reverse-phase high-performance liquid chromatography (RP-HPlc) profiles of l-2,4diaminobutanoic acid 8mer (dab8) in water (left) and 1× phosphate buffer (right) after heating for 1 h. RP-HPlc analyses were performed using a linear gradient of 0%-30% cH 3 cN/0.05% tFA in water/0.05% tFA at 30 °c for 30 min with a flow rate of 0.5 ml/min using a c18 column, and detection at 280 nm.
that Dab8 was stable in water. Next, a 0.9 mM Dab8 solution in 1× phosphate buffer (10 mM NaH 2 PO 4 -Na 2 HPO 4 , 100 mM NaCl, pH 7.0) was heated at 40 °C, 50 °C, 60 °C, or 90 °C for 1 h (Figure 6 (right) and S24 (right)). In the RP-HPLC profile of the sample heated at 40 °C, there was only a peak corresponding to Dab8. However, few new peaks were observed for the sample kept at 50 °C. The number of peaks corresponding to the degraded products increased with the temperature, and the sample heated at 90 °C showed significant decomposition. The degraded products formed by heating Dab8 solution in the phosphate buffer at 90 °C were analyzed by mass spectrometry. As a result, peaks which were corresponded to lactam derivatives were detected ( Figure S29), suggesting that Dab8 was decomposed by the intermolecular nucleophilic attack by an amino group on a carbonyl group forming a lactam with five-membered ring ( Figure  7). In addition, a 1 mM Dab8 solution in 1 × PBS (137 mM NaCl, 8.1 mM Na 2 HPO 4 , 2.68 mM KCl, 1.47 mM KH 2 PO 4 , pH 7.5) showed a similar behavior ( Figure S25). Furthermore, 0.9 mM Dab8 solutions in Tris buffer (100 mM Tris·HCl, 100 mM NaCl) at various pH values were heated at 40 °C or 90 °C for 1 h and analyzed by RP-HPLC. Although there was no sign of decomposition of Dab8 kept at 40 °C at below pH 7.0, the peaks corresponding to the degraded product were observed for the sample at 40 °C under basic conditions. The analysis of samples heated at 90 °C indicated that Dab8 was decomposed under mild acidic and basic conditions, and the degradation was notable under basic conditions ( Figures  S26 and S27).
To summarize, it was revealed that Dabs were stable up to approximately 40 °C under neutral conditions, but Dabs are sensitive to heat, especially under high salt concentration and/or pH value.

Circular dichroism (CD) spectroscopy
Next, the CD spectrum of DNA/RNA and RNA/RNA duplexes were measured in the presence and absence of ODAGals to elucidate the structural changes caused by the cationic molecules. In the previous reports, ODAGals and Dabs interact with DNA/RNA and RNA/RNA duplexes without altering their A-type duplex structures. [11,14,17] On the other hand, we found that the CD spectra changed appreciably in the presence of ODAGals at low temperatures. [14] Therefore, in this study, we investigated the structural changes of the duplexes and ODAGals complex due to annealing.
We used 5 µM 16mer DNA/RNA (Seq 2-1), RNA/RNA (Seq 5-1), and 2 equivalents of ODAGals for the experiments. All measurements were conducted under physiological conditions using a 10 mM phosphate buffer containing 100 mM NaCl at pH 7.0. The analytical samples were prepared according to two methods; in Method A, an aqueous solution of ODAGal4 was added after a pair of complementary strands annealed, whereas, in Method B, an aqueous solution of ODAGal4 was added before a pair of complementary strands annealed. The CD spectrum of obtained samples was measured at 10 °C and 37 °C. The results are shown in Figures 8, S28, and S29. It is known that the CD spectrum of the A-type duplexes showed a positive peak at around 265 nm and a negative peak at around 210 nm. The spectra of the sample prepared by Method A and B showed the characteristic feature of A-type duplexes. However, a remarkable increase in the intensities of the negative peak around 210 nm was observed for the samples prepared by Method A compared to Method B. It was revealed that the structures of the complex with ODAGals and duplexes differed depending on the sample preparation method. This may be due to the slight difference in the binding position of ODAGals, which may have led to the difference in ΔT m values of the first and second forward direction measurements in the thermal denaturation tests (Table S2).

RNase A resistance experiment
Next, RNase A resistance experiments were performed to investigate the effects of ODAGals and Dabs on the suppression of the RNA strand digestion by RNase A. RNase A is an enzyme that cleaves single-stranded RNA. However, it also acts on double-stranded RNA, such as siRNA. [20] DNA/RNA is generally less stable than RNA/RNA. Therefore, in this study, we used a DNA/RNA duplex easily cleaved by RNase to evaluate the difference in the biostabilizing effect of these cationic molecules. The 16mer DNA/RNA duplex (Seq 2-1) and RNase A from the bovine pancreas were used for the experiments. The final concentration of the solutions of 1 µM duplex and 0, 1, or 3 equivalents of ODAGals or Dabs in a 10 mM Tris buffer containing 100 mM NaCl with pH 7.1 were treated with 0.50 μg/ mL RNase A at 37 °C. [20] After the reaction, the solution of RNase inhibitor was added to deactivate RNase A. Then, the mixtures were analyzed by RP-HPLC. The HPLC profiles are shown in Figure 9 and Supplementary information. Figures S32 and S34 show the results in the presence of 1 equivalent of the cationic molecules, and Figures 9 and S34-S36 show the results in the presence of 3 equivalents of the cationic molecules.
The HPLC profiles indicated that the RNA strand was almost digested after treatment of RNase A for 30 min in the absence of the cationic molecules. In contrast, more intact RNAs were observed in the presence of 3 equivalents of ODAGals or Dabs ( Figure 9). Notably, in the presence of 3 equivalents of ODAGals or Dabs, the intact RNAs were observed even after 6 h ( Figure S36). These results suggested that the addition of ODAGals and Dabs effectively protected DNA/RNA duplexes against the cleavage by RNase A. The extent of the protection was notable with increasing the number of the amino groups and equivalents of the cationic molecules, and ODAGal and Dab with the same number of amino groups had the commensurate ability for the suppression of the digestion. In contrast, as stated earlier, Dabs enhanced the thermal stability of A-type duplexes more effectively than ODAGals in the thermal denaturation studies. These results indicated that the thermal stabilization effect of the cationic molecules on A-type duplexes was not always correlate well with the biological stabilization effect.

RNase H1 activity experiment
Finally, we investigated the effects of ODAGals and Dabs on human RNase H1 activity using some RNA/DNA duplexes. In most previous studies, we used E. coli RNase H1 to assess its activity and revealed that Dabs could enhance the activity of RNase H1. [21] Also, ODAGal4 is effective in mismatch discrimination by RNase H1 cleavage. [22] Recently, we have reported that the human RNase H1 activity was maintained to some extent in the presence of ODAGals. [14] In this study, we investigated the effect of ODAGals and Dabs on human RNase H1 activity using two RNA/DNA duplexes.
The experiment was designed based on the conditions described on the Abcam website [23] and our previous study. [14] We used a 24mer RNA based on the firefly luciferase gene and 16mer RNAs based on the mouse ApoB mRNA. Solutions of 4 µM duplex and 1 equivalent of cationic molecules in a 60 mM Tris buffer containing 60 mM KCl, 2.5 mM MgCl 2 , and 2.0 mM DTT with a pH of 7.5 were treated with RNase H1 for 30 min at 37 °C. After the reaction, the mixture was heated to 95 °C and rapidly cooled to 4 °C to denature RNase H1. RP-HPLC performed the analysis, and the peaks corresponding to cleaved products were identified by mass spectrometry and/or comparison to the purchased standards.
First, we performed the experiments with equal amounts of 24mer RNA (rCACUUACGCUGAGUACUUCGAAAU) and 12mer DNA (dCGAAGTACTCAG) DNA (Seq 6). The results are shown in Figure 10. When duplex Seq 6 was treated with human RNase H1, the HPLC profiles were complex, and some peaks corresponding to cleaved fragments of the RNA strand were detected. This result indicated multiple cleavage sites by human RNase H1 (Figures 10 and S42). The profile in the presence of ODAGal4 was almost identical to those in the absence, whereas ODAGal5 or ODAGal6 showed decreased cleavage rates of the RNA strand. On the other hand, in the presence of Dabs, the cleavage rate was not significantly reduced, and Dab8 and Dab10 improved human RNase H1 cleavage. However, concerning RNA cleavage sites, there was little difference in the cleavage sites in the presence and absence of ODAGals or Dabs. In addition, we performed the experiments using an excess amount of RNA. Seq 6 was used, and three equivalents of RNA and one equivalent of cationic molecules were mixed with DNA and treated with human RNase H1 ( Figure 11). The presence of ODAGal5 and ODAGal6 decreased the cleavage rate, and this result was similar to when equal amounts of DNA and RNA were used. On the other hand, the cleavage rate was increased in the presence of Dabs. The presence of Dab10 or Dab12 using equal amounts of DNA and RNA tended to slightly decrease the cleavage rate, whereas the presence of Dabs using DNA and RNA in a ratio of 1:3 increased the cleavage rate. Moreover, the rate of RNA cleavage in the presence of Dab12 exceeded that of Dab10. No significant difference was observed for the cleavage site.
Moreover, we performed these experiments with 16mer DNA/RNA (Seq 2-1). Comparing the HPLC profiles, there was a difference in the cleavage rate but not in the degree of cleaved fragments (Figures S43 and S44). These tendencies were consistent with the results mentioned above using Seq 6. On the other hand, the HPLC profile in Figure S44 suggests that the presence of ODAGal4 also increases the rate of cleavage when an excess amount of RNA is used for DNA. It was not observed in experiments containing equal amounts of DNA and RNA. The previous reports suggest that ODAGals and Dabs bind to the major groove of a DNA/RNA duplex and thus do not inhibit the activity of RNase H1 that binds to the minor groove of the duplex. [11,19,24] Therefore, when the experiments were conducted using human RNase H1, the same tendency as the E. coli RNase H1 used in the previous experiments was observed. These results support this hypothesis. In any case, the presence of these cationic molecules did not completely inhibit the RNase H1 activity. In particular, Dabs increased the cleavage rate of the RNA strand by RNase H1 when the amount of RNA was in excess to DNA. The fact that RNA cleavage by human RNase H1 is not inhibited when RNA is present in excess relative to DNA is beneficial for applications of these cationic molecules in oligonucleotide therapeutics.

Conclusions
We evaluated the properties of oligodiaminogalactoses (ODAGal) and L-2,4-diaminobutanoic acid oligomers (Dabs) that bind to A-type oligonucleotide duplexes. In the thermal denaturation studies, unmodified and phosphorothioate-modified nucleic acids were stabilized in the presence of cationic molecules, and the effects were more notable with Dabs. However, it was observed that ODAGals and Dabs with high amino group content sometimes form aggregates with duplexes. Moreover, Dabs were sensitive to heat in buffers. In addition, RNase A resistance studies revealed that these cationic molecules have a similar ability to protect the DNA/ RNA duplex from the digestion of RNase A.
Furthermore, human RNase H1 activity experiments elucidate that these molecules did not completely inhibit the RNA cleavage by RNase H1. In particular, when an RNA strand was used in excess of a DNA strand, the presence of Dabs enhanced the cleavage rate. Through vigorous investigation of the properties of ODAGals and Dabs, it was confirmed that these molecules were adequate for enhancing the thermal and biological stability of A-type duplexes in a stoichiometric ratio. Thus, these cationic molecules are helpful carriers for A-type duplexes, which are widely used as oligonucleotide therapeutics, such as siRNAs and HDOs. The insights obtained in this study would contribute to a pragmatic choice of these molecules for applying oligonucleotide therapeutics, depending on the purpose. Thermodynamic analysis using isothermal titration calorimetry (ITC) and nuclear magnetic resonance (NMR) is currently underway to elucidate the binding modes of these molecules to A-type oligonucleotide duplexes.

General methods and materials
ODAGal hydrochlorides and Dab hydrochlorides were used for all experiments unless otherwise noted. Dabs were purchased from SCRUM Inc. Oligonucleotides from Japan Bio Services Co., Ltd. Recombinant Human RNase H1 was purchased from Abcam (ab153634). A PROTEOSAVE™SS 0.5 mL Microtube (SumitomoBakelite Co., Ltd) was used as the reaction container for the RNase A and RNase H1 assay. RP-HPLC was performed using the Delta-Pak column (5 μm, C18, 100 Å, 3.9 × 150 mm, Waters). Degradation products of Dab8 were identified by electrospray ionization (ESI) mass spectrometry. RNA cleavage fragments of RNase H1 activity experiments were identified by electrospray ionization (ESI) mass spectrometry and/or comparison to the purchased standards.

Thermal denaturation study
The experiments were performed in a 10 mM NaH 2 PO 4 -Na 2 HPO 4 buffer containing 100 mM NaCl with a pH of 7.0. A solution containing a pair of complementary strands was heated at 95 °C for 5 min and cooled to 4 °C at a rate of 0.5 °C/min. 1 or 2 equivalents of ODAGal or Dab were added to the solution, and the final concentration of the duplex was 2.5 (for 20mer duplexes) or 4 µM (for 12mer and 16mer duplexes). After ODAGal or Dab was added to the solution, these samples were cooled to 0 °C and left for 10 min. Finally, the absorbance at 260 nm and 320 nm were recorded in the forward direction from 0 °C to 90 °C or 95 °C at a rate of 0.5 °C/ min unless otherwise noted.

CD spectroscopy
The experiments were performed in a 10 mM NaH 2 PO 4 -Na 2 HPO 4 buffer containing 100 mM NaCl with a pH of 7.0. A solution containing a pair of complementary strands was heated at 95 °C for 5 min and cooled to 4 °C at a rate of 0.5 °C/min. 2 equivalents of ODAGals were added to the solution, and the final concentration of the duplex was 5 µM. All CD spectra were recorded in a wavelength range of 200-320 nm at 10 °C and 37 °C. The following instrument settings were used: resolution: 1 nm; sensitivity: 5 mdeg; response: 1 s; speed: 100 nm/min; accumulation: 10.

RNase A resistance experiment
The experiments were performed in a 10 mM Tris buffer containing 100 mM NaCl with a pH of 7.1 at 37 °C. A solution containing a pair of complementary strands was heated at 95 °C for 5 min and cooled to 4 °C at a rate of 0.5 °C/min. 1 or 3 equivalents of ODAGals or Dabs were added to the solution, and the final concentration of the duplex was 1 µM. A solution of RNase A from the bovine pancreas in the Tris buffer was added to the duplex solution with or without ODAGals or Dabs to yield a final concentration of 0.50 μg/mL. After treatment with RNase A, the reaction was quenched by adding an RNase A inhibitor (TOYOBO) and then rapidly cooled to 4 °C. RP-HPLC analyzed the mixture.

RNase H1 activity experiment
The experiments were performed in a 60 mM Tris buffer containing 60 mM KCl, 2.5 mM MgCl 2 , and 2.0 mM DTT with a pH of 7.5 at 37 °C. A solution containing a pair of complementary strands was heated at 95 °C for 5 min and cooled to 4 °C at a rate of 0.5 °C/min. 1 equivalent of cationic molecules were added to the solution, and the final concentration of the duplex was 4 µM. To the solution of the duplex in the presence or absence of cationic molecules, 0.02 equivalent of human RNase H1 (Abcam, ab153634) was added. After 30 min, the mixture was rapidly heated to 95 °C, left for 5 min, and then rapidly cooled to 4 °C to denature RNase H1. RP-HPLC analyzed the mixture.