Artificial cationic peptides that increase nuclease resistance of siRNA without disturbing RNAi activity

Abstract Properties of cationic peptides bearing amino or guanidino groups with various side chain lengths that bind to double stranded RNAs (dsRNAs) were investigated. Peptides with shorter side chain lengths effectively bound to dsRNAs (12mers) increasing their thermal stability. NMR measurements suggested that the cationic peptide binds to the inner side of the major groove of dsRNA. These peptides also increased the thermal stability of siRNA and effectively protected from RNase A digestion. On the other hand, both peptides containing amino groups and guanidine groups did not disturb RNAi activity.

Since the discovery of RNA interference (RNAi) [1] , RNAi drugs have been a focus as one of the most promising nucleic acid therapeutics. [2][3][4] In the proposed mechanism of RNAi, double-stranded RNAs (dsRNAs) are first processed by Dicer into short interfering RNAs (siRNAs). [5,6] Then, the siRNA is subsequently guided to Argonaute proteins to form the RNAinduced silencing complex (RISC) that decomposes the complementary mRNA.
RNAi drugs would be promising therapeutic agents, however, they are not ready for practical applications due to their inherent characteristics. Since siRNAs are negatively charged and highly hydrophilic, their membrane permeability is significantly low. [6] Moreover, rapid enzymatic digestion of naked siRNAs in plasma interrupts their delivery to target cells. [7] Therefore, development of effective delivery systems for RNAi drugs are in high demand to overcome these problems. Many trials for the effective delivery of RNAi drugs have been reported. For example, chemical modification effectively increases the stability of siRNA in cells by inhibiting the recognition of nuclease; however, some types of chemical modification also inhibited RISC formation, resulting in a decrease of RNAi activity. [8] As an alternative, delivery systems utilizing cationic lipids are one of the most investigated strategies [9][10][11] and some cationic carriers successfully showed high transfection efficiency of siRNAs in vitro; however, use of a large excess of cationic molecules gives rise to cytotoxicity. [12] Therefore, compatibility of both nuclease resistance and RNAi activity is still one of the main problems for the development of a delivery system for siRNA.
We have focused on a strategy for delivery of siRNAs that uses cationic molecules that selectively bind to dsRNAs in a non-covalent manner. In our previous study, we designed cationic peptides [13] (Figure 1) and cationic oligosaccharides [14] as siRNA binding molecules, some of which significantly increased the thermal stability of dsRNAs. In addition, cationic saccharide derivatives conjugated with vitamin E were also designed to deliver siRNAs to liver cells. [15,16] Compared to the oligosaccharide derivatives, peptides are easy to synthesize and conjugate with transporting molecules. Cationic oligopeptides consisting of L-arginine (Arg) or L-lysine (Lys) have already been reported for siRNA delivery. [17] Specifically, Arg oligomers have high membrane permeability and were found to be effective for delivery of oligonucleotide therapeutics to target cells. [14] In this study, we investigated the properties of the artificial cationic peptides such as their binding affinity to siRNA by measuring the association constant (K a ), effects on nuclease resistance and RNAi activity.
In this study, we synthesized oligopeptide 11mers containing an N-acetyl-Tyr-Gly-Gly (AcYGG) residue at the N-terminus for UV detection and quantification that containing eight basic amino acids with amino or guanidino groups (Figure 1). These peptides were prepared by the standard Fmoc solid-phase peptide synthesis. [13] First, the binding affinity of cationic peptides bearing amino (A1 to A4) and guanidino groups (G1 to G4) with dsRNAs were evaluated by fluorescence anisotropy measurements. The cationic peptides were titrated into a solution of the annealed fluorescently-labeled self-complementary dsRNA 12mer, (FAM-rCGCGAAUUCGCG) 2 , and complex formation was observed by an increase of anisotropy values. The K a values between cationic peptides and dsRNAs were calculated by curve fitting the concentrationanisotropy curves. These data indicate that the affinity of cationic peptides with dsRNAs varied in accordance with their side chain lengths and functional groups on the side chains (Table 1). Peptides with shorter side chain lengths, specifically A2 and G1 (K a ¼ 1.4 Â 10 7 and 2.2 Â 10 7 M À1 ), bound to dsRNA stronger than peptides with natural amino acids A4 and G3 (K a ¼ 7.1 Â 10 6 and 6.3 Â 10 6 M À1 ). Shorter side chain lengths of peptides decreased the flexibility of the functional groups, resulting in a decrease in entropy loss. As a result, peptides such as A2 and G1 show higher affinity to dsRNAs, similar to that of RNA-binding proteins. [18,19] There was a clear relationship between the K a and the melting temperature (T m ) in all cases except for A1 ( Figure 2). [13] The peptides strongly bind to both RNA strands, effectively increasing the thermal stability of dsRNA. On the other hand, A1 bound to dsRNA (K a ¼ 4.8 Â 10 6 M À1 ) without increasing the thermal stability of dsRNAs; therefore, A1 bound to dsRNA with a different binding mode compared to the other cationic peptides.
Next, NMR measurements were performed to reveal the binding mode of cationic peptides to dsRNAs. The NMR signals of the imino protons, non-exchangeable base protons (H2/H5/H6/H8), and ribose H1' of (rCGCGAAUUCGCG) 2 were assigned using a well-established sequential NOE connectivity method. [20] The assignments of these signals were compared to the previously reported NMR data [21] and confirmed to be correct.  The signals from the RNA slightly shifted upon addition of the peptide A2, and the signals of the dsRNA in the complex of RNA-A2 were assigned using the sequential NOE connectivity method ( Figure 3).
Although NMR signals from A2 could not be assigned due to the overlap of NMR signals from the eight Dab residues, signals of the b (2.08 ppm, 2.16 ppm) and c (3.06 ppm) protons were clearly discriminated from other signals. When we measured the NOESY spectra (mixing time of 100 ms) at 10 C, intermolecular NOEs were observed between the b protons of A2 and C1H5, G2H8, C3H5, A5H8, and U8H5 of the dsRNA ( Figure 4). Furthermore, intermolecular NOEs were observed between the c protons and of A2 and C1H5, C1H6, G2H8, C3H5, C3H6, G4H8, A5H8, U8H5, and C9H5 of the dsRNA. These protons are found in the major groove of the dsRNA. In contrast, intermolecular NOEs were not observed between A2 and the protons in the minor groove, A5H2, A6H2, or ribose protons. Thus, it was clearly revealed that A2 binds to the major groove of the dsRNA. On the other hand, each peptide has a L-tyrosine via glycine at the N-terminus, but intermolecular NOEs were not observed between Tyr and dsRNA. Therefore, it is suggested that peptides  mainly bind to the dsRNA due to electrostatic interaction, and the side chain length effectively affects the binding ability of peptides to the dsRNA.
Since the cationic peptides, especially A2 and G1, were found to effectively bind to the dsRNA 12mer to increase thermal stability, we also investigated the effects of the peptides on siRNAs. Because siRNA has a longer major groove than the dsRNA 12mers, T m measurements in the presence of excess peptides were conducted to reveal the stoichiometry of siRNA-A2 or G1 ( Figure 5). In the case of A2, the shape of the UV melting curve changed relative to the amount of A2; the T m values of siRNA-A2 complexes increased by 11.8 C to 14.4 C compared to the siRNA as the amount of A2 increased from 1 equiv to 5 equiv (Table 2). Therefore, it was suggested that more than 3 molecules of A2 can bind to siRNA. In the case of G1, each amount of G1 showed almost same shape UV melting curve; however, in the presence of 5 equiv of G1 a decrease in the UV absorption was observed in the low temperature range. These phenomena may be caused by aggregation of siRNA-G1. For your information, we measured CD spectra of 12mer dsRNA ((rCGCGAAUUCGCG) 2 ) in the presence of various equivalents of G1 (see Figure S6). In these experiments, CD spectra of the dsRNA in the presence of 2 equiv. or more of G1 were quite different from those in the case of less than 2 equiv. These results also suggest that excess amount of G1 induce higher-order-structural changes of dsRNAs such as aggregation.
The effects of cationic peptides on the stability of siRNA towards nucleases was estimated. A similar stability profile of siRNA in human serum and with RNase A was reported. [22] Therefore, the RNase A resistance of siRNA was investigated to evaluate the protection ability of cationic peptides by forming peptides-siRNA complexes. Fluorescence resonance energy transfer (FRET) measurements were employed for the real-time analysis of siRNA digestion by RNase A (Figure 6). The results indicated that the peptides increased the RNase A resistance. As siRNA has a much longer major groove the A2 and G1 peptide, 1 equivalent of A2 and G1 did not showed effective RNase A resistance; however, more than 3 equivalents of A2 or G1 showed effective RNase A resistance.
Finally, in order to assess the gene silencing ability of siRNA-A2 or G1 complexes, the expression level of ApoB1 mRNA was evaluated using a quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) that correlated to RNAi activity. Figure 7 shows the relative amount of ApoB1 mRNA in the absence and presence of peptides. In both cases, all complexes showed a similar level of RNAi activity relative to siRNA itself. Argonaute protein binding to a siRNA through many arginine residues has been reported, [23] so the amino groups of A2 and guanidine groups of G1 did not disturb RNAi activities.
We investigated the properties of cationic peptides, revealing that amino and guanidino groups, especially A2 and G1, effectively interact with dsRNA to increase the thermal stability and nuclease resistance of dsRNAs. A clear relationship between the thermal stability and nuclease stability was observed. Since RNase A recognizes single strand RNA, increasing the thermal stability Table 2. DT m values of peptide-siRNA complexes.
of dsRNA inhibits recognition by RNase A. NMR measurements reveal that A2 binds to the major groove of dsRNA and the main chain of the peptides is located in the deep inside the major groove. These results suggest that even small molecules can increase the thermal stability of dsRNA, and possess the potential to increase the stability of siRNA. Furthermore, A2 and G1 did not disturb the RNAi activity. Therefore, amino and guanidine groupmodified peptides with short side chain lengths effectively increased nuclease resistance without disturbing RNAi activity, an important characteristic for application of these molecules as nucleic acid drugs.

Materials
Peptides were synthesized via conventional solid-phase method using the 9fluorenylmethyloxycarbonyl (Fmoc) strategy as described previously. [13] RNA, siRNA and fluorescence and Dabcyl labeled RNA were purchased from Japan Bio Services. The sequence of the scrambled siRNA (r(GCACGUCUCAACCAAUUAA)dTdT/r(UUAAUUGGUUGAGACGUGC) dTdT) was designed using siRNA Wizard Software (InvivoGen).  Melting temperature (T m ) analysis Absorbance versus temperature profile measurements were carried out in quartz cells with a 1 cm path length using an eight-sample cell changer. Variation in the UV absorbance with temperature was monitored at 260 nm. The oligonucleotides were dissolved in 10 mM phosphate buffer containing 100 mM NaCl, pH 7.0. Solutions of (rCGCGAAUUCGCG) 2 (final conc: 4 lM) or r(GUCAUCACACUGAAUACCA)dTdT/r(U GGUAUUCAGUGUGAUGAC)dTdT (final conc: 1 lM) were first rapidly heated to 95 C, left for 10 min, then cooled to 10 C at a rate of 1 C/min. Then, 1, 3, 5 equivalents of peptides were added to the solution. The samples were left to equilibrate at the starting temperature for 30 min, the dissociation of the duplex was observed by heating the solution to 95 C at a rate of 0.5 C/min and data points were collected at every 0.1 C.

NMR measurements
The RNA sample (rCGCGAAUUCGCG) 2 was dissolved in 10 mM sodium phosphate (pH 7) containing 50 mM NaCl. The final concentration of the RNA was 1 mM. NMR spectra were measured using a Bruker AVANCE 600 spectrometer. Spectra were recorded at a probe temperature of 10 or 25 C. Exchangeable proton resonances were assigned by NOESY in H 2 O with mixing times of 150 ms using the jump-and-return scheme [24] for water suppression. Non-exchangeable proton resonances were assigned by NOESY (mixing times of 100 and 400 ms) in D 2 O.
In vitro siRNA activity assay To determine in vitro activity of siRNAs targeting human Apolipoprotein B (APOB) mRNA, Huh-7 cells were transfected with 10 nM siRNAs using Lipofectamine TM RNAiMAX (Invitrogen). The cells were harvested 24 h after transfection. Total RNA was extracted and the amount of endogenous APOB mRNA was measured by quantitative real-time polymerase chain reaction (qRT-PCR).
Quantitative RT-PCR (qRT-PCR) assay DNase-treated 500 ng of RNAs were reverse transcribed with Super Script III and Random Hexamers (Life Technologies, Carlsbad, CA). The cDNAs were amplified by the quantitative TaqMan system using the Light Cycler 480 Real-Time PCR Instrument (Roche Diagnostics, Mannheim, Germany). The primers and probes for human APOB (NM_000384) and human Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH; NM_002046.3) were designed by Applied Biosystems (Foster City, CA).

Statistical analysis
The student's two-tailed t-test was used to determine the significance of the differences between control and transfected groups in the quantitative RT-PCR assay. Data are presented as means ± standard error of the means (SEMs); P < 0.05 was considered significant.