RNA knockdown by synthetic peptidyl-oligonucleotide ribonucleases: behavior of recognition and cleavage elements under physiological conditions

Abstract Sequence-specific protein-based ribonucleases are not found in nature. Absolute sequence selectivity in RNA cleavage in vivo normally requires multi-component complexes that recruit a guide RNA or DNA for target recognition and a protein–RNA assembly for catalytic functioning (e.g. RNAi molecular machinery, RNase H). Recently discovered peptidyl-oligonucleotide synthetic ribonucleases selectively knock down pathogenic RNAs by irreversible cleavage to offer unprecedented opportunities for control of disease-relevant RNA. Understanding how to increase their potency, selectivity and catalytic turnover will open the translational pathway to successful therapeutics. Yet, very little is known about how these chemical ribonucleases bind, cleave and leave their target. Rational design awaits this understanding in order to control therapy, particularly how to overcome the trade-off between sequence specificity and potency through catalytic turnover. We illuminate this here by characterizing the interactions of these chemical RNases with both complementary and non-complementary RNAs using Tm profiles, fluorescence, UV-visible and NMR spectroscopies. Crucially, the level of counter cations, which are tightly-controlled within cellular compartments, also controlled these interactions. The oligonucleotide component dominated interaction between conjugates and complementary targets in the presence of physiological levels of counter cations (K+), sufficient to prevent repulsion between the complementary nucleic acid strands to allow Watson-Crick hydrogen bonding. In contrast, the positively-charged catalytic peptide interacted poorly with target RNA, when counter cations similarly screened the negatively-charged sugar-phosphate RNA backbones. The peptide only became the key player, when counter cations were insufficient for charge screening; moreover, only under such non-physiological conditions did conjugates form strong complexes with non-complementary RNAs. Communicated by Ramaswamy H. Sarma


Introduction
Bioconjugates with controlled chemical properties can be designed with new biological functionality as a basis for therapeutic innovation to provide solutions to complex biological challenges, which are currently unmet by conventional drug discovery approaches (Scannell et al., 2012). One of the major health care priorities is to develop highly-selective treatments against abnormal gene expression in disease states, which currently rely on the cytotoxic effect of chemotherapies and/or radiotherapy and inevitably lead to high levels of toxicity and side effects in humans. The majority of available low molecular weight drugs only try to tackle disease downstream, when established in cell components at the level of expressed pathogenic proteins. Although longfavored, this therapeutic approach of treating or alleviating expressed symptoms often fails to meet therapeutic needs. For example, the high drug doses required often cause significant side-effects (Casi & Neri, 2015;Chari, 2008), typical for chemotherapy and anti-viral drugs; and many proteins (e.g. transcription factors) are not suited to targeting by small-molecule drugs due to their nuclear location, complexity and/or lack of suitable binding sites.
Recently, an alternative approach has been developed (Beloglazova et al., 2004;Bennett et al., 2017;Bennett & Swayze, 2010;Garzon et al., 2010;Kole et al., 2012) with focus instead on the upstream catalytic targeting of diseaserelevant RNAs, which include mRNAs encoding pathogenic proteins, non-coding RNAs involved in cellular signaling pathways (e.g. microRNAs) or viral genomic RNA. The advantage of RNA as a therapeutic target is recognized due to its pervasive role in macromolecular processes, including transcriptional and translational regulation, epigenetic memory (Holoch & Moazed, 2015), RNA splicing (Romero-Barrios et al., 2018) and retroviral replication (Zhang et al., 2018). Further, as RNA is the genetic material of some viruses, the RNA virus CONTACT Elena V. Bichenkova elena.v.bichenkova@manchester.ac.uk Supplemental data for this article can be accessed online at https://doi.org/10.1080/07391102.2020.1751711. genome could be selectively attacked. Key advantages of using RNA as a therapeutic target include the lack of repair mechanisms, accessibility of many RNAs in the cytoplasm and a less-intensive puckering structure with greater tertiary structural diversity than DNA, thus providing specific folding motifs as additional targets for site-specific drugs.
We have recently developed a new class of catalytic bioconjugates, as synthetic ribonucleases using amphipathic peptides, which only cleave RNA when conjugated to oligonucleotide recognition motifs (Mironova et al., 2002;Mironova, Boutorine et al., 2004;Mironova et al., 2006Mironova et al., , 2007Patutina et al., 2017Patutina et al., , 2019Pyshnyi et al., 1997;Staroseletz, Williams et al., 2017;Williams et al., 2015). These RNA-targeting synthetic nucleases are chemically-engineered by conjugation of short, catalytically-inactive peptides with oligonucleotide recognition motif components, which range from fully-through poorlyto non-complementary to the target RNA, in order to produce novel biologically-active molecules capable of recognizing and cleaving RNA sequences quantitatively. The most remarkable feature of these novel biocatalysts is that the conjugation of peptide and oligonucleotide seems to produce a new, hybrid type of molecule (Mironova, Boutorine et al., 2004;Mironova et al., 2006), with new properties, which are not found in nature. This synergistic combination of the individual properties of the two components yields a new catalytic entity, which is capable of catalyzing the cleavage of phosphodiester bonds in RNA under physiological conditions, with a rate enhancement factor of RNA cleavage reaching 10 8 in some cases, as compared to non-catalyzed reaction (Mironova et al., 2007). We have recently demonstrated that this type of biologically-active conjugates can selectively knock down highly-oncogenic miR-21 by irreversible cleavage and thereby down-regulate abnormal expression in lymphosarcoma cells, thus leading to a reduction in proliferative activity of tumor cells (Patutina et al., 2017(Patutina et al., , 2019. Insight into the different binding behaviors of the catalytic and recognition elements of these novel bioconjugates is essential for rational design to control their biological behavior, and to accelerate their translational development into therapies with clinical impact. Potent, selective and irreversible catalytic cleavage of specific mRNA sequences encoding disease-relevant proteins or different micro-RNAs associated with diverse types of human pathology (e.g. cancer, inflammation, neurodegeneration) has the potential to expand the scientific boundaries of drug discovery and allow new selective therapies to develop.
Peptidyl-oligonucleotide conjugates (POCs) designed to date have the ability to cleave RNA at two different extremes: (i) either sequence-specifically (Mironova et al., 2002;Patutina et al., 2017Patutina et al., , 2019Pyshnyi et al., 1997;Staroseletz, Williams et al., 2017;Williams et al., 2015) by site-directed cleavage of the regions adjacent to the RNA hybridization sites, but in a non-catalytic manner; or (ii) non-specifically at non-complementary regions, which were located distantly from the major RNA binding region, but with a high level of catalytic turnover (Mironova, Boutorine et al., 2004;Mironova et al., 2006Mironova et al., , 2007. These data provided some experimental indications that peptidyl-oligonucleotide conjugates may interact differently with the complementary and non-complementary regions of RNA sequences. Insight into the role of different structural elements of such bioconjugates in recognizing and cleaving RNA sequences is now vital, but challenging. Precise sequence-specific recognition of the target often opposes their ability to leave RNA after each cleavage event, and gain the high level of reaction by catalytic turnover, necessary for a potent drug. This study aims therefore to investigate the possible modes of the interactions of catalytic peptide and model POCs bearing 6-mer and 9-mer oligonucleotide recognition components (see C1, C2 or C3, respectively; Figure 1) with both complementary and noncomplementary 20-mer RNA targets using fluorescent, UV-visible and NMR spectroscopies. 2 0 -O-methyl oligoribonucleotide analogues were used as model RNA targets, in order to prevent any cleavage during hybridization experiments. We find that charge screening produces markedly different interactions under electrolyte conditions relevant to the intracellular environment, where high levels of monovalent cations (K þ ) and low levels of divalent cations (Ca þþ ) are tightly controlled in the cytosolic domain of transcribed RNA. NMR Diffusion Ordered Spectroscopy (DOSY) illustrated self-assembly of 20 Å monomeric into 40 Å dimeric peptide-oligonucleotide complexes, which may explain the non-linear enhancement of their catalytic activity with concentration, and the possible benefit of engineering assembly.

Materials
Reagents, of the highest quality available, were purchased from the following suppliers.

Analytical methods and instrumentation
Reversed phase HPLC. HPLC purification of compounds was performed using an Agilent 1100 system fitted with a thermostated column compartment, a quaternary pump with solvent degasser, and a diode-array detector module for multi-wavelength signal detection using an Agilent 1100 series of UV-visible and a fluorescent detector module. The system had a manual Rheodyne injector, and the instrument was operated through a PC running Agilent HPLC 2D ChemStation Software. Depending on the purpose indicated, the columns used were: semi-prep Phenomenex Luna 5 mm C18 column (250 Â 10 mm), or analytical Phenomenex Luna, 5 lm C18 column (250 Â 4.6 mm). NMR spectroscopy. 1 D NMR ( 1 H and 13 C) and 2 D NMR ( 1 H-1 H-NOESY, 1 H-1 H-COSY, 1 H-1 H-TOCSY and 1 H-13 C HSQC) spectra were recorded using Bruker Avance II þ spectrometers operating at proton frequencies of 400 or 500 MHz. 31 P NMR spectra were recorded using either a Bruker 300 MHz (Avance-300) spectrometer or Bruker Avance II þ spectrometers (400 MHz) operating at phosphorus frequencies of 121.5 MHz or 162 MHz, respectively. In all cases, spectra were acquired and processed using Bruker software TopSpinV R v2.0 or v2.1. Chemical structure of fluorescein (FAM) and mode of its attachment to the 5 0 -terminal phosphate of nucleic acid residues. (C-D) Six POC-RNA systems studied in the hybridization assays, which involved the conjugate NH 2 -Gly-[Arg-Leu] 4 -5 0 pUUACACACACUGGGAAGUUU 3 0 (FAM-RNA 2 ; green). The Watson-Crick base-pairs between the complementary oligonucleotide component and FAM-RNA 1 are shown aligned and indicated in black. All conjugates carry the same peptide moiety -[Leu-Arg] 4 -Gly-NH 2 (black) attached at the 5 0 -terminal position of the oligonucleotide component (blue). The orange and pink spheres represent the attached fluorescein (FAM) and indocarbocyanine (Cy3) dyes, respectively.
1D 1 H NMR data were collected using 64 transients into 65 k data points over a spectral width of 8 kHz (400 MHz spectrometer) or 10 kHz (500 MHz spectrometer), with a relaxation delay of 2.4 s between scans. 31 P NMR data were collected into 64 K data points over a spectral width of 48 KHz. For each spectrum up to 4095 transients were acquired, with 2 s relaxation delay between scans. Processing involved multiplication by an exponential window function prior to Fourier transform and phase correction. In all cases, the spectrometers were equipped with Microsoft Windows XP Professional running TopSpinV R NMR software (Version 2.0 or 2.1) for acquiring and processing NMR data. Chemical shifts (d) are reported in parts per million (ppm) with peak positions reported relative to trimethylsilyl-2,2,3,3-tetradeuteropropionic acid (0.00 ppm) and trimethyl phosphate (TMP; 3.1 ppm) internal standards for 1 H NMR and 31 P NMR experiments, respectively. Two-dimensional 1 H-1 H COSY NMR data were collected with quadrature detection into 2048 complex data points for 512 or 1024 t 1 increments with a number of transients appropriate for the sample concentration being acquired for each t 1 increment. Data were acquired over 3.2 KHz (400 MHz spectrometer) or around 4 KHz (500 MHz spectrometer) with optimized spectral width for each sample from the 1 D 1 H spectrum, and with 1.64 s relaxation delay.
2D 1 H NOESY spectra were recorded using 500 MHz Bruker Avance þ II spectrometer using noesyprf2gpph pulse program, with water pre-saturation and a phase sensitive acquisition mode State-TPPI. Spectra were collected into 2048 complex points in the t 2 dimension and 512 complex points in the t 1 dimensions. The mixing time was 0.5 s, and the relaxation delay between scans was 1.82 s. The spectral width was set to 8.0 ppm in both dimensions, and the sample spinning was turned off. The spectra were processed using TopSpinV R 2.1 by applying Fourier transformation for both dimensions using QSIN window function after zero-filling.
Data reported use the following convention: chemical shift (splitting patterns, integrated intensity, and assignment). Abbreviations used for splitting patterns are: s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet), dd (doublet of doublets), dt (doublet of triplets), br (broad), obs (obscured). 1 H DOSY experiments Valentini et al., 2004) were carried out on the 400 MHz Bruker Avance þ II spectrometer, using 5 mm MIC Z121725/0001DOSY z-diffusion probe with fixed 1 H/X configuration and 2 H lock to generate a gradient strenghth of 30 Gauss/cm/A. The combination of the DIFF30L probe with the 40 Amp gradient current amplifier (GREAT 40) allowed to produce a field strength upto maximum of 1200 Gauss/cm. The pulse program used was diffste, and the experiments were run at room temperature. The gradient (g), the length of the gradient pulse (d) and the diffusion time (D) were calibrated for each experiment using TopSpinV R DOSY script (diff). An array of experiments with different combinations of the three parameters was run to obtain sufficient signal attenuation. At the parameterization step, each experiment was run over 16 scans and 16 gradient steps. The typical value for the gradient field (g) was 400 Gauss/cm, whereas the optimal values for the gradient duration (d) and diffusion time (D) were 35 ms and 0.8 ms, respectively. Once the sufficient signal attenuation was achieved, the diffusion measurements were carried out using the original Bruker pulse sequences. The gradient pulse length and the diffusion time were kept at fixed values while gradually increasing the gradient strength. In each DOSY experiment, a series of 56 spectra were collected into 128 K data points, using a linear gradient ramp. The resulting FIDs were processed using MestReNova, and the diffusion coefficients and hydrodynamic radius were calculated by CONTIN algorithm (Provencher, 1982a(Provencher, , 1982bValentini et al., 2004).
Mass spectrometry. Mass spectra of the synthesized peptides and their oligonucleotide conjugates were recorded at the Manchester Interdisciplinary Biocentre (MIB, University of Manchester) using a Matrix Assisted Laser Desorption/ Ionization-Time of Flight-/Time of flight (MALDI-Tof/Tof) mass spectrometer (Ultraflex II V R ) with pulsed nitrogen laser of wave length 337 nm and Voyager software. Some MALDI-ToF experiments were carried out for us by EPSRC National Mass Spectrometry Service Centre (Swansea University, Swansea). To record MALDI-ToF mass spectra, a 1 mg/mL stock solution of a sample was prepared in water. For the peptide samples, gel matrix solutions were prepared by dissolving 2,5-dihydroxybenzoic acid in a mixture of solvent containing MeCN:water:TFA at 50:50:1 ratio, respectively, to obtain 10 mg/ml final concentrations. To record MALDI-ToF mass spectra for the oligonucleotide conjugates, a-cyano-4-hydroxycinnamic acid matrix and a solvent (1:1 mixture of MeCN and water) were used to avoid possible hydrolysis of the phosphoramidate bond by the TFA. A 0.5 mL sample was pipetted and spotted on to a MALDI sample plate and allowed to dry. The spotted samples were then covered using 0.5 mL matrix.
Sample quantification for hybridization experiments. The concentrations of oligonucleotides, 2 0 -O-methyl oligoribonucleotide analogues and oligonucleotide conjugates in stock solutions were measured by taking small aliquots (typically 2 ml; in triplicate) followed by the addition of appropriate volume of de-ionized water to obtain the final volume of 1 mL, when UV absorbance at 260 nm was measured, which was then used to estimate both the concentration of the test sample and the concentration of the original stock solution using the Beer-Lambert Law: where A 260 is UV absorbance measured at 260 nm; e 260 is the millimolar extinction coefficient of the corresponding oligonucleotide at 260 nm (mM À1 Ácm À1 ), C is the concentration of the sample (mM) and L is the pathlength of the cuvette (cm). The millimolar extinction coefficients for oligonucleotides and their analogues were calculated using an online commercial (ATDbio Ltd.) oligonucleotide analyzer, 1 which evaluates extinction coefficients based on the principle of the nearest neighbor thermodynamic model. The estimated e 260 values for conjugates C1 and C2 were found to be 56.9 and 81.2 mM À1 Ácm À1 , respectively. In the case of fluorescently-labeled conjugate C3 and RNA targets (FAM-RNA 1 and FAM-RNA 2 ), the millimolar extinction coefficients were calculated by taking into account the millimolar extinction coefficients of free FAM (e 260 ¼ 20.9 mM À1 cm À1 ) and Cy3 (e 260 ¼ 4.93 mM À1 cm À1 ). 2 The resulting e 260 values of the fluorescently-labeled analogues were 86.13 mM À1 Ácm À1 for C3, 244.6 mM À1 Ácm À1 for FAM-RNA 1 and 215.3 mM À1 Ácm À1 for FAM-RNA 2.
All stock solutions were additionally purified by gel filtration chromatography using Illustra TM NAP disposable columns prepacked with Sephadex G-25 (DNA grade resin) to remove any possible traces of counter cations from peptidyloligonucleotide conjugates, RNA and peptide samples.
UV-visible spectroscopy. UV-visible spectroscopy measurements were carried out using a Cary-4000 UV-Visible spectrophotometer from Varian (Australia) connected to a Cary Peltier temperature controller, operating under Varian Cary WinUV software. All spectra were recorded between 200 to 900 nm at 5 C using 1 cm pathlength matched quartz cuvettes. Hybridization of the oligonucleotide conjugates C1, C2 and C3 with 20-mer RNA target (FAM-RNA 1 or FAM-RNA 2 ) were studied in 50 mM Tris-HCl, 200 mM KCl, and 0.5 mM EDTA at pH 7.0. Initially, the UV-visible spectra of the oligonucleotide conjugates (2.5 mM) alone were measured in buffer at 5 C (k max , 260 nm). The target RNA was then added to achieve 2.5 mM concentration in the cuvette, and spectra were taken until no further change occurred.
Fluorescence spectroscopy. Fluorescence excitation and emission spectra were recorded in 1 cm pathlength quartz cuvettes at 5 C using a Cary-Eclipse fluorescence spectrophotometer fitted with Cary Peltier-thermostat cuvette holder. The instrument was operated with Cary Eclipse software for Windows. Hybridization of the oligonucleotide conjugates C1-C3 with complementary 20-mer RNA target (FAM-RNA 1 or FAM-RNA 2 ) were studied either in Tris buffer (50 mM Tris-HCl, 200 mM KCl, and 0.5 mM EDTA at pH 7.0) or in de-ionized water. First, the emission spectra of the free fluorescently-labeled component (either FAM-RNA 1 or FAM-RNA 2 ; 5 mM) were measured at 5 C following excitation at 495 nm. Then, the second component (C1, C2 or C3) was added to reach the final conjugate concentration of 5 mM in the cuvette, to achieve 1:1 molar ratio of both components. The emission spectra of the complex formed were then recorded following excitation at 495 nm. In the independent experiments with the dually labeled system, the emission spectrum of C3 (2.5 mM) was first recorded either in water or in Tris buffer on excitation at 552 nm, followed by addition either FAM-RNA 1 or FAM-RNA 2 to achieve 2.5 mM final concentration in the cuvette. The binding between conjugates and RNA sequences was monitored by recording fluorescence spectra at 570 nm (for Cy3 emission) following excitation at 552 nm.
Melting curves. Melting temperature experiments were performed in 1 cm pathlength quartz cuvettes using a Cary-4000 UV-visible spectrophotometer fitted with Cary Peltierthermostat cuvette holder. Thermal denaturation experiments monitored at 260 nm (oligonucleotide absorbance), 495 nm (FAM absorbance) and/or 550 nm (Cy3 absorbance) were recorded between 5 and 90 C, at a rate of 0.5 or 0.1 C/min with a data interval of 0.5 or 0.1 C. The melting temperature, T m , is defined as the temperature corresponding to the half-transition of the thermal denaturation process of the duplex. The T m values were calculated as the maxima of the first-order derivatives of the melting curves using Cary software.

Peptide synthesis
Catalytic peptide [Leu-Arg] 4 -Gly-NH 2 was synthesized by manual solid-phase Boc strategy using MBHA.HCl resin. 4methylbenzhydrylamine hydrochloride salt resin (1.02 g, MBHA.HCl (0.59 mEq amine/g, $ 0.602 mmol)) was washed with DCM, neutralized with DIPEA/DCM (1:19, ca. 12 mL) for 5 min and left to swell in DCM over two days, rewashed with DCM and DMF, and the first Boc-protected amino acid residue, Boc-Gly-OH, was added to perform the coupling procedure as described below. At each coupling step, 1.81 mmol of the Boc-amino acid to be coupled (3.0 eq. relative to resinbound amine) was pre-activated for 3-5 min in a small vial by combining it with O-benzotriazole-N,N,N 0 ,N 0 -tetramethyluronium-hexafluoro-phosphate (HBTU, 3 eq. relative to resinbound amine) and diisopropylethylamine (DIPEA, 9 eq. relative to resin-bound amine) in 10 mL of DMF and added to the glass vessel containing resin. The reaction vessel was shaken for 40 min, washed with DMF (2 Â 15 mL) and DCM (2 Â 15 mL). The resulting mixed Boc-aminoacyl polymer was assessed with ninhydrin using the Kaiser test (Stewart & Young, 1979). The aminoacyl resin was then treated with TFA/DCM (1:1) for 2 min, again with TFA/DCM (1:1) for 30 min, washed with DCM (Â4), neutralized with DIPEA/DCM (1:19) for 5 min and finally washed with DCM and DMF to complete the coupling cycle. The same cycle was repeated for coupling the remaining amino acid residues to obtain the desired sequence [Leu-Arg] 4 -Gly-NH 2 . After completion of the final coupling cycle, the peptide resin was washed twice with DCM followed by MeOH and dried.
The dried resin with the attached peptide chain was placed in a reactor equipped with a Teflon-coated stirring bar, when p-cresol (0.75 g) and of thiocresol (0.25 g) scavengers were added. The reactor was closed, connected to the apparatus and cooled down in an ice bath to 0 -4 C, when HF (15 mL) was distilled under vacuum into the reaction vessel and stirred for 2 h. The HF was evaporated under vacuum at low temperature (4 C). The cleaved peptide and resin were suspended in diethyl ether to precipitate the peptide, which was washed with diethyl ether and then with EtOAc (2 Â 30 mL) to remove the remaining scavengers. The peptide was extracted with 20% acetic acid (Â3), followed by 80% acetonitrile to ensure the complete removal of the peptide from the resin. Finally, the solution was lyophilized to give $500 mg of crude yield of pale green solid.

Coupling of the peptide to oligonucleotide
Prior to coupling of the peptide [Leu-Arg] 4 -Gly-NH 2 to the 5 0terminal phosphate group of the oligonucleotide, the latter was converted into the N-cetyl-N,N,N-trimethylammonium salt, soluble in DMSO, as described earlier (Staroseletz, Williams et al., 2017;Williams et al., 2015). The N-cetyl-N,N,Ntrimethylammonium salt of the oligonucleotide (0.12 mmol) was dissolved in anhydrous DMSO ($ 60 mL) and treated with an excess of activating agents: 2,2 0 -dipyridyl disulfide (10 mg, 45.5 mmol) and triphenylphosphine (10 mg, 38.1 mmol) for 5 min at room temperature followed by addition of N,Ndimethylaminopyridine (5 mg, 41 mmol) for another 10 min. Anhydrous DMSO (20 lL) and 4-(dimethylamino)pyridine (3 mg, 24.6 lmol) were added to a separate tube containing oligopeptide (1.4 mg, 1.2 mmol), and the resulting solution was transferred into the activated oligonucleotide solutions. The reaction mixture was incubated at 40 C for 3 h, followed by precipitation with 4% LiClO 4 in acetone as described in (Staroseletz, Williams et al., 2017;Williams et al., 2015).

RNA cleavage assays
Preparation of substrate RNAs. RNA HIV-1 was prepared by in vitro transcription with T7 RNA polymerase using Fok Ilinearized plasmid pHIV-1 (Milligan & Uhlenbeck, 1989). The reaction was carried out in 300 lL of 40 mM Tris-HCl, pH 7.5, containing 6 mM MgCl 2 , 2 mM spermidine, 10 mM sodium chloride, 10 mM dithiothreitol, 1 mM of each NTP, 30 lg of DNA template, and 100 U T7 RNA polymerase, for 2 h at 37 C. The reaction was quenched by phenol/chloroform (1:1, v/v) extraction, followed by ethanol precipitation. After centrifugation, RNA precipitate was rinsed twice with 80% ethanol and dissolved in water.
HIV-1 RNA transcript was dephosphorylated using bacterial alkaline phosphatase BAP (Fermentas) according to a described protocol (Silberklang et al., 1979). The reaction mixture, 50 lL of 50 mM Tris-HCl, pH 8.5, containing 1 mM EDTA, 0.2% SDS, 2% formamide, 2.5 mM DTT, 0.1 A 260 in vitro transcript of RNA HIV-1 and 2 U of bacterial alkaline phosphatase was incubated at 37 for 1 h. BAP was added to the reaction mixture at 0 and 30 min incubation time. The reaction was quenched by phenol/chloroform (1:1, v/v) extraction followed by extraction of water phase with ethyl ester and ethanol precipitation.
RNA 5 0 -labeling. 5 0 -End labeling of RNA HIV-1 was carried out using [c 32 P]-ATP and T4 polynucleotide kinase (Fermentas) according to a described protocol (Silberklang et al., 1979). 32 P-labeled RNAs were isolated by electrophoresis in 12% denaturing polyacrylamide gel with 8 M urea. RNAs were visualized by autoradiography using X-ray film. 32 P-Labelled RNAs were eluted from the gel with 300 lL 0.3 sodium acetate, h 5.5 followed by ethanol precipitation. Ribonuclease activity assay. The reaction mixture (10 lL) was prepared in such a way to contain the following components at the stated final concentrations: 50,000 cpm 5 0 -[ 32 P]labelled RNA, one of the oligonucleotide-peptide conjugates (C1 or C2) at concentration either 10 lM or 50 lM in 50 mM Tris-HCl buffer, pH 7.0, 0.2 M KCl and 1 mM ðDTA, in the presence of 100 lg mL À1 RNA carrier. The mixtures were incubated at 37 C (for various times) and quenched by precipitation of RNA with 2% lithium perchlorate in acetone (150 lL). RNA was collected by centrifugation and dissolved in loading buffer (6 M urea, 0.025% bromophenol blue, 0.025% xylene cyanol). RNA and RNA cleavage products were resolved in 12% polyacrylamide/8 M urea gel using TBE (100 mM Tris-borate, pH 8.3, 2 mM EDTA) as a running buffer. To identify cleavage sites, an imidazole ladder and G-ladder produced by partial RNA cleavage with 2 M imidazole buffer (pH 7.0) and with RNase T1, respectively, were run in parallel. Quantitative data were obtained by counting using Molecular Imager FX (Bio-Rad, USA). The total extent of RNA cleavage and extent of RNA cleavage at specific sites were determined as a ratio of radioactivity measured in the RNA fragment(s) to the total radioactivity applied onto the gel.

Conjugate design
Peptidyl-oligonucleotide conjugates C1-C3 were designed to report interactions with target RNA by fluorescence, UV-visible and NMR studies (see Figure 1 for chemical structures and nomenclature, and Supplementary Figure S6 for oligonucleotide sequences). FAM was used as a dye for fluorescent labeling of the model complementary (FAM-RNA 1 ) and non-complementary (FAM-RNA 2 ) targets (Figure 1(B)). Indocarbocyanine dye (Cy3) was also used as a fluorescent tag at the 3 0 -terminal position of the 9-mer conjugate to produce fluorescently labelled conjugate C3. Fluorescein has an absorption maximum around 495 nm and fluorescence emission maximum at 520 nm, whereas Cy3 has absorption maxima around 550 nm and emission maxima at 570 nm. Change in fluorescence of the two fluorophores should report binding between C3 and the FAM-labelled RNA sequences and provide useful indications of the mode of interactions at least by quenching, if not by F€ orster Resonance Energy Transfer (FRET). FAM and Cy3 often act as a Donor-Acceptor couple (Chen et al., 2010;Olejko et al., 2016;Olejko & Bald, 2017;Tsourkas et al., 2003;Zhao et al., 2010), as the fluorescence emission region of FAM (480-600 nm) substantially overlaps with the excitation region of Cy3 (470-570 nm).

Conjugate ribonuclease activity
Ribonuclease activities of the conjugates C1 and C2 were confirmed and compared in back-to-back studies against 5 0 -[ 32 P]-RNA-HIV-1 (Figure 2(A)), which does not contain regions that are fully complementary to the oligonucleotide components of the conjugates. The secondary structure of HIV-RNA-1 is shown in Figure 2(B), with red arrows indicating the major sites of RNA cleavage by C1 and C2 conjugates, whereas the total extent of 5 0 -[ 32 P]-RNA-HIV-1 cleavage is summarized by a diagram in Figure 2(C).
Overall efficiency of 5 0 -[ 32 P]-RNA-HIV-1 cleavage was found to be 30% and 37% for C1 and C2, respectively, at 10 mM conjugate concentration. An increase of the conjugate concentrations to 50 mM led to the enhancement of the overall cleavage efficiency to 50% for C1, whereas the efficiency of its elongated analogue C2 reached 100% over 24 h.
Analysis of the RNA digestion products showed that the C1 and C2 conjugates cleaved 5 0 -[ 32 P]-RNA-HIV-1 predominantly at Pyr-A sites. Indeed, C-A and U-A positions were cleaved more efficiently than C-G and U-G sites. Pyr-A motifs located within loops, bulge-loops and/or junctions seemed to represent the major cleavage sites (i.e. U7-A8, U11-A12, C22-A23, U41-A42, C54-A55, C68-A69; indicated by red arrows in Figure 2(B)). In contrast, the Pyr-G motifs and the sites located within stem regions were cleaved less efficiently (e.g. U32-A33 > C60-G61, C64-G65, U77-A78, C90-A91 ) U51-A52), and were detected mainly at higher conjugate concentrations (50 mM), thus signifying the lower propensity of these sites toward cleavage by this class of conjugates. Interestingly, the U32-A33 site located within the long stem region of HIV-RNA-1 was cleaved more efficiently than U51-A52 site, which was located within the opposite strand of the same stem region. This might be attributed to the variations in accessibility of these two regions and/or to the different levels of their structural tension, which may affect catalysis of transesterification of the RNA phosphodiester bonds. Indeed, our recent molecular dynamics simulations (Staroseletz, Nechaev et al., 2017) showed considerable changes in the fine structure of this stem regions. Electrophoretic mobility of fragments formed upon RNA cleavage induced by the conjugates corresponded to the mobility of fragments formed by RNA cleavage with RNase T1 (Figure 2(A)) and 2 M imidazole buffer (primary data not shown), which suggests similar products formed upon cleavage with POCs and RNase T1. It is known that RNase T1 produces fragments bearing 5 0 -hydroxyl and 2 0 ,3 0cyclophosphate, similar to the RNA fragments that can be generated either through alkaline hydrolysis or in the presence of 2 M imidazole buffer. We can conclude therefore that the cleavage reaction catalyzed by C1 and C2 at physiological pH occurs through usual transesterification of the 2 0oxyanion onto the adjacent phosphorous atom leading to the formation of a di-anionic phosphorane intermediate, followed by the departure of the 5 0 -linked nucleoside and formation of a 2 0 , 3 0 cyclic phosphate.
The hypothetical cleavage mechanism of RNA sequences by such peptidyl-oligonucleotide conjugates was discussed earlier (Lonnberg, 2011;Niittymaki & Lonnberg, 2006;Staroseletz, Williams et al., 2017;Williams et al., 2015), including the catalytic role of the guanidinium groups in promoting transesterification, as well as their synchronized action when they are present in the same molecular structure  Remarkably, the base-specificity and the cleavage efficacy of this class of the peptidyl-oligonucleotide conjugates strongly depend on the nature, sequence and structural features of the RNA target. Earlier, C1 and C2 conjugates were evaluated against the linear (e.g. non-structured) complementary oligoribonucleotide 5 0 -[P 32 ]-GAUUGAAAAUCCCC, which corresponded to a sequence from the anticodon arm of E.Coli tRNA Phe (Pyshnyi et al., 1997 and had some sequence similarity with RNA 1 studied in this research. In contrast to the data presented here (see Figure 2(A)) showing clear Pyr-A preference in cleavage of 5 0 -[ 32 P]-RNA-HIV-1, both conjugates demonstrated exclusive G-X base-specificity for the site-directed cleavage of this short, linear complementary target, mainly at G1-A2 and G5-A6 positions. C1 showed 80% cleavage activity, whereas the activity of C2 against this target was not reported. Similarly, G-X basespecificity was detected again for these C1 and C2 conjugates, when they were studied against the linear (i.e. non-structured) non-complementary 20-mer sequence 5 0 -[P 32 ]-UUACACACACU GGGAAGUUU , which had full homology with RNA 2 studied here. The overall cleavage efficiency of C1 and C2 against this target was found to be 80% and 63%, respectively, with the main cleavage sites seen at G12-G13, G13-G14, G14-A15 and G17-U18 positions . Neither peptide alone nor the mixture of the unconjugated peptide and oligonucleotide possessed cleavage activity, thus suggesting that only the hybrid peptide-oligonucleotide could promote catalysis.

'Single-label' hybridization
Clear evidence of strong interactions at 1:1 molar ratio between the complementary RNA sequence and conjugate C2 or C1 (Figure 3), both in the presence of monovalent ions (i.e. in Tris buffer) and in their absence (i.e. in water), was provided by a noticeable response from fluorescein tag located at the 5 0 -terminal phosphate of the 2 0 -O-methyl-RNA sequences, which was also supported by UV-visible spectroscopy (see Figure S7 in the Supplementary Material). In Tris buffer, this can be attributed to the formation of the C2:FAM-RNA 1 or C1:FAM-RNA 1 duplexes stabilized by Watson-Crick hydrogen bonding between the oligonucleotide moiety of the POC and the complementary region of the RNA 1 sequence (see Figure 3(A,E), respectively). The oligonucleotide recognition component seemed to play the dominant role in the interactions between the C2 or C1 conjugates and the complementary target in the presence of a high level of counter cations under buffered-electrolyte conditions, which was sufficient to minimize repulsion between the negatively charged oligonucleotide strands, as demonstrated by a pictorial diagram shown in Figure 4.
However, in the absence of electrolyte, the peptide component seemed to become the key player in the interactions with both complementary and non-complementary target, when counter cations were insufficient or negligible in de-ionized water, thus allowing repulsion between the negatively-charged sugar-phosphate backbones of the oligonucleotides. Indeed, the addition of the conjugate C2 or C1 to the complementary target FAM-RNA 1 in water led to considerable quenching (30%) of the fluorescent signal in both cases (Figure 3(B,F), respectively). Thus, a high degree of binding to RNA through the peptide component was indicated when the strong electrostatic repulsion between the sugar-phosphate backbones of the oligonucleotides cannot be compensated by usual Watson-Crick hydrogen bonding between the complementary nucleotide residues.  (blue) and after the addition of conjugate C2 or C1 (red) recorded in Tris-buffer (A and E, respectively) and in de-ionized water (B and F, respectively). Fluorescence emission spectra of the non-complementary RNA FAM-RNA 2 alone (blue) and after the addition of conjugate C2 or C1 (red) recorded in Tris-buffer (C and G, respectively) and in de-ionized water (D and H, respectively). The concentration of each component in the cuvette was 5 lM. The spectra were recorded at 5 C with the excitation wavelength set at 498 nm and 481 nm in Tris buffer and water, respectively. The slit width was 2.5 nm.
Similar levels of interaction were observed in water between these conjugates with the non-complementary target FAM-RNA 2 , which was evident from considerable (30-38%) fluorescence quenching (Figure 3(D,H)), thus illustrating the predominance of nonspecific (in terms of RNA sequence) interactions in water, in the absence of physiological electrolyte.
Such interactions were also seen by UV-visible spectroscopy (see Figure S7 in the Supplementary Material). For example, conjugate C2 interaction with the non-complementary FAM-RNA 2 target in water led to a 55% decrease in the intensity of the FAM absorption bands at around 360 -510 nm (hypochromic effect). In addition, the absorption maxima k max of this band was also blue shifted by 4 nm (to 454 nm) relative to the unbound FAM-RNA 2 target (458 nm). These observations were in agreement with the noticeable fluorescence response, thus confirming strong intermolecular interactions between C2 and FAM-RNA 2 in water.
In the absence of the shielding effect from monovalent cations, the electrostatic interactions between the negativelycharged sugar-phosphate backbones of the RNA and the positively-charged arginine residues of the conjugates seem to become strong enough to prevent repulsion between the oligonucleotide moieties and stabilize nonselective conjugate-RNA complexes in de-ionized water. Under such conditions, the peptide seems to act as an 'electrostatic anchor' to hold all the components together (Figure 4(B,D) illustrates proposed arrangements).
The role of the electrostatic forces in these interactions was also evident from the fact that physiological concentrations ($200 mM) of monovalent counter cations, whether predominantly Na þ (cf. extracellular) or K þ (cf. intracellular), completely suppressed the formation of any stable complexes between C1 or C2 conjugates and non-complementary RNA sequences (see Figure 3(C,G), respectively), presumably due to their competition for binding with the negatively-charged phosphate groups. Only negligible (<1%) or minor (<5%) fluorescence response was induced by addition of C2 or C1, respectively, to FAM-RNA 2 in the Tris-buffered electrolyte (Figure 3(C,G), respectively). Moreover, we demonstrated by fluorescence recovery (see Figure S8 in the Supplementary Material) that the conjugates bound to the non-complementary RNA in water can be fully displaced from the complexes by simple injection of a high concentration of Na þ or K þ ions (up to 1 M). We conclude therefore that this interaction with non-complementary RNA was mediated mainly by the peptide moiety via electrostatic interactions with the RNA strand, which seems to represent the only possibility for interactions between the noncomplementary RNA and the conjugates in water.
Our 1 H and 31 P NMR studies of the unconjugated peptide with FAM-RNA 1 and FAM-RNA 2 (see Supplementary Figures S9-S14) provided additional evidence of the predominant role of the electrostatic forces in their interactions. Relatively sharp 1 H or 31 P NMR signals of FAM-RNA 1 , typical for unbound molecular species, showed marked line broadening upon mixing of these two components together in water (Supplementary Figures S9 and S10, respectively). This was accompanied by a decrease of signal intensities, presumably due to formation of multiple conformational species of the peptide-RNA complexes formed. In addition, some peptide 1 H signals showed noticeable (0.04 -0.08 ppm) up-field shifts upon interaction with RNA. However, the addition of NaCl in excess (1 M final concentration) reversed signal broadening and led to substantial or full recovery of their intensities and chemical shifts. When FAM-RNA 1 (30 lM) was initially premixed with NaCl (250 mM), the subsequent addition of the 4fold excess of peptide (to achieve 120 lM) had no significant effect on 31 P NMR spectra of RNA (see Supplementary Figure  11) and led to only very minor changes in positions of some 1 H NMR peaks of FAM-RNA 1 (see Supplementary Figure 12). This was presumably due to the presence of the high concentration of counter cations in the sample, which blocked any potential arginine-phosphate nonspecific electrostatic interactions. Similar effect was observed for the non-complementary FAM-RNA 2 upon addition of the unconjugated peptide in Tris-buffered electrolyte (50 mM Tris-HCl, pH 7.0, 200 mM KCl and 0.5 mM EDTA). 1 H NMR and 31 P NMR spectra recorded for individual FAM-RNA 2 and peptide components showed no significant differences after mixing them together, thus eliminating any possibility of stable complex formation between these components in the presence of counter cations at sufficient concentrations (see Supplementary Figures 13 and 14, respectively).
Nevertheless, the contributions of some other molecular forces (e.g. hydrogen bonding between guanidinium groups of arginine and guanosine residues) cannot be fully excluded from the factors stabilizing RNA-peptide or RNA-conjugate interactions. Indeed, it was recently reported (Krishna et al., 2019) that a guanidinium moiety, when incorporated into the middle of the PNA chain, could serve as a hydrogen bond donor and potentially form two hydrogen bonds with the guanosine residue of the G-C base-pair within the double-stranded RNA target, provided that these interactions were additionally stabilized through triplex formation by the PNA nucleotides flanking such guanidinium groups. However, this cannot be assumed for single-stranded RNA sequences, because hydrogen bonds between guanidinium groups and unpaired guanosine residues appeared to be unstable. In fact, the binding affinity of the PNA incorporating guanidinium moiety toward single-stranded RNA (relevant to this study) was shown to be 14 times lower than that toward double-stranded RNA (Krishna et al., 2019). These results seemed to agree with our molecular modeling simulations, which demonstrated the possibility of hydrogen bonding interactions between the arginine sidechain with the N 7 and carbonyl O atoms of guanosine residues (Patutina et al., 2017). However, these interactions appeared to be ancillary, whereas the network of ionic interactions between arginine residues with backbone phosphate groups of RNA was shown to dominate such interactions.
The predominant modes of interactions indicated by these studies between the unlabeled conjugates C1 and C2 with the complementary and non-complementary RNA targets FAM-RNA 1 and FAM-RNA 2 , respectively, are depicted in Figure 4, for the two experimental extremes of buffered electrolyte (A and C) and de-ionized water (B and D).

'Dual label' hybridization
Dually-labelled systems (see Figure 1(C,D)) with two fluorescent tags, one attached to the 3 0 -terminus of the conjugate C3 (i.e. 3 0 -Cy3) and the other attached to the 5 0 -terminus of the RNA target (i.e. 5 0 -FAM) provided insight into the relative location of individual components and illuminated their particular roles in binding events, both in water and buffered electrolyte.
The overall response on binding was more pronounced (Figure 5) for the dually-labelled systems, but demonstrated very similar behavior patterns to those observed earlier for the unlabeled C1 or C2 conjugates. Indeed, the hybridization of the FAM-RNA 1 with C3 in Tris-buffered electrolyte led to an 87% decrease of the fluorescence in both cases: when the hybridization event was monitored by following the emission of FAM-RNA 1 (excited at 496 nm) upon the addition of C3 (Figure 5(A)); and, in the parallel studies, by looking at the emission of Cy3-labelled conjugate ( Figure 5(B)) upon the addition of FAM-RNA 1 (excited at 552 nm). In each case, this considerable quenching of fluorescence was accompanied by a small, but reproducible (1-2 nm) shift in the emission k max to a longer wavelength (bathochromic shift) seen for either FAM or Cy3 emission band.
Both the nature and the scale of these reciprocal responses from the two fluorophores within the duallylabelled system were somewhat unexpected. FAM and Cy3 have been often used in as a Donor-Acceptor couple in FRET, due to the substantial overlap of the emission spectrum of FAM with the excitation spectrum of Cy3 (Chen et al., 2010;Olejko et al., 2016;Olejko & Bald, 2017;Tsourkas et al., 2003;Zhao et al., 2010). In the case of F€ orster resonance energy transfer from FAM (acting as a Donor) to Cy3 (acting as an Acceptor), the observed decrease of FAM fluorescence is usually coupled with an increase of Cy3 fluorescence. Instead here with dually-labelled components, we detected very strong quenching of both emission bands triggered by their self-assembly. A possible explanation of such mutual quenching of the fluorophores could be a formation of a non-fluorescent complex between the excited fluorophore (FAM) and another fluorophore molecule (Cy3) via 'contact quenching' (also known as a 'static quenching') (Crisalli & Kool, 2011;Johansson et al., 2002;Marras et al., 2002).
In 'contact quenching', the donor and acceptor molecules interact via a proton-coupled electron transfer through the formation of hydrogen bonds at their ground-state, which are often controlled by electrostatic and/or hydrophobic forces. When such a ground-state hetero-dimer is excited using a resonant electromagnetic radiation wavelength, the resulting excited state immediately returns to the ground state without emission of a photon. As a result, such groundstate complex does not emit fluorescence upon excitation. A characteristic feature of such 'contact quenching' is a change in the UV-visible absorption spectrum of this ground-state complex, as compared to that of the individual components (Crisalli & Kool, 2011;Johansson et al., 2002;Marras et al., 2002). This is different from the FRET mechanism, when UVvisible absorbance of the complex is expected to be an additive of the spectral contributions of the individual molecules. Also, FRET does not require very close proximity of two fluorophores and may occur over relatively long distances from 10 to 100 Å, which is approximately the distance between 3 and 30 nucleotides in a double-stranded DNA molecule (Bichenkova et al., 2011;Marras et al., 2006).
In order to investigate whether Cy3 and FAM fluorophores form such a ground-state complex upon hybridization of C3 with the complementary target FAM-RNA 1 , we recorded the separate UV-visible spectra of the free conjugate C3, unbound FAM-RNA 1 , as well as their equimolar mixture in Tris-buffered electrolyte ( Figure 5(E)). The UV-visible spectrum of the C3:FAM-RNA 1 complex (red) cannot be reconstructed by summing the absorption spectra of the free C3 (green) and unbound FAM-RNA 1 (black), and thus is not merely additive (cf. red and blue curves in Figure 5(E), where blue curve corresponds to the 'additive' spectrum obtained by summing of the absorption spectra of the FAM-RNA 1 and C3). These observations suggest a certain degree of intermolecular interactions between Cy3 and FAM at their ground state, which led to the formation of a hetero-dimer between two fluorophores with its own, distinctive electronic properties, such as being non-fluorescent and having unique UV-visible characteristics. Therefore, this significant, mutual quenching in fluorescence seen for both fluorophores on hybridization in buffer conditions ( Figure  5(A,B)) is likely to be attributed to 'contact quenching', indicative of a very close location of FAM and Cy3 within the C3:FAM-RNA 1 complex in buffered electrolyte, such as to allow for direct proton-coupled electron transfer at their ground states. This hypothesis is consistent with the earlier proposed model of the interaction between the conjugate and complementary RNA target in the presence of counter cations (see Figure 4(A)). A pictorial diagram showing possible mode of interactions between the Cy3-labelled conjugate C3 and the complementary target FAM-RNA 1 in Tris-buffered electrolyte is given in Figure 6(A), that emphasizes the key role of the oligonucleotide component in the presence of a high level of counter cations, which are necessary to prevent repulsion between the negatively-charged complementary oligonucleotide strands.
In contrast, the conjugate C3 showed negligible interactions with the non-complementary RNA sequence under identical conditions (i.e. Tris buffer in Figure 5(C)). Presumably, the presence of excess counter cations effectively suppressed any possible electrostatic interaction between the conjugate and the noncomplementary RNA target, where sequence-specific Watson-Crick hydrogen bonds could not be formed between the noncomplementary oligonucleotide sequences (Figure 6(C)).
The mode of interaction of the conjugate C3 toward both complementary (FAM-RNA 1 ) and non-complementary (FAM-RNA 2 ) RNA seemed to completely change in water, in the absence of electrolyte. Indeed, the level of Cy3 fluorescence quenching on binding with RNA dropped from 87% in buffer ( Figure 5(A,B)) to only 8% for the C3:FAM-RNA 1 complex in water (Table 1), which presumably indicates the higher degree of separation between Cy3 and FAM under these conditions. The interaction between the non-complementary target FAM-RNA 2 and the C3 conjugate in water also showed the reduced level of response (34% fluorescence decrease) from the Cy3 tag ( Figure 5(D)). This behavior corresponds well with the model proposed earlier for C2 and C1 in water (Figure 4(B,D)), when the detected interactions with RNA are mediated mainly by the peptide component via electrostatic interactions, so that RNA and the oligonucleotide component become rather distant (Figure 6(B,D)). According to this model, separation between the FAM and Cy3 within the C3:FAM-RNA 1 complex becomes significant, thus reducing or eliminating completely the contribution from the 'contact quenching' for these two fluorophores. and after the addition of conjugate C3 (red). (B) Fluorescence spectra of the Cy3-labeled conjugate C3 alone (blue) and after the addition of FAM-RNA 1 target (red). (C-D) Fluorescence response on interaction of C3 with non-complementary target FAM-RNA 2 in Tris buffer (C) and water (D). Fluorescence spectra of the Cy3-labeled conjugate C3 alone (blue) and after the addition of FAM-RNA 2 sequence (red). The spectra were recorded following excitation either of FAM at 496 nm (A) or Cy3 at 552 nm (B-D). The concentration of each component in the test cuvette was 2.5 mM, and slit width of 2.5 nm was used for both excitation and emission. Emission maxima at around 520 nm and 560 nm correspond to the fluorescence bands of FAM and Cy3, respectively. (E) UV-visible spectra of the unbound FAM-RNA 1 (black), free conjugate C3 (green) and their mixture (red). Blue curve shows the 'additive' spectrum obtained by summing of the absorption spectra of the FAM-RNA 1 and C3. All spectra were recorded under identical conditions in Tris buffer at 5 C. The concentrations of the components in the test cuvette were 2.5 lM.

Melting temperature profiles
Additional evidence in support of our hypothesized interactions between the conjugates C1, C2 and C3 with the complementary target RNA 1 in buffered electrolyte and water was provided by melting temperature measurements ( Figure  7(A-I)). As expected, the T m profile of the complex formed between conjugate C2 and FAM-RNA 1 (Figure 7(A)) revealed a sigmoidal curve, typical for double-stranded nucleic acids, with T m value of 29 C and 23% hyperchromicity. This characteristic temperature-induced transition seems to be attributed to the dissociation of the duplex formed by the conjugate and its complementary RNA target. Although the melting curve obtained for the C1: FAM-RNA 1 complex in buffered electrolyte (Figure 7(B)) was not very well defined, due to a lower stability of the shorter duplex formed between the 6mer conjugate C1 and the complementary region of the FAM-RNA 1 , there was a clear indication of sigmoidal-like transition with the T m around 17 C and 15.5% hyperchromicity. It was not possible to obtain a low-temperature plateau for this short complex, as this region was below 0 C. However, this data indicated that at 5 C C1 and FAM-RNA 1 seem to exist predominantly in a duplex form, which is in agreement with the fluorescence data. The peptide seemed to stabilize complexes formed between conjugates and RNA, presumably by nonspecific electrostatic interactions. This was supported by the slight increase in T m value seen for the C2:FAM-RNA 1 as compared to that of the 'control' duplex ON 2 -FAM-RNA 1 (T m ¼24 C) (cf. Figure 7(A and E)).
Interestingly, the melting temperature profiles observed in water for the C2 and C1 conjugates with both the complementary FAM-RNA 1 target (Figure 7(C,D), respectively) as well as with non-complementary FAM-RNA 2 target (Figure 7(G,H), respectively) revealed a co-operative, sigmoidal-like transition, which is more typical for thermal denaturation of a double-stranded duplex, although the temperature intervals for the transitions in water were in general considerably broader than those seen in buffered electrolyte. The observed S-like character of the melting curves seen for the C2:FAM-RNA 1 , C2:FAM-RNA 2 , C1:FAM-RNA 1 and C1:FAM-RNA 2 complexes in water suggests an involvement of numerous nonspecific contacts between adjacent nucleotides and amino acid fragments, which thereby showed cooperative behavior. The higher thermal stability of the C1:FAM-RNA 2 and C1:FAM-RNA 1 complexes in water (T m ¼30 C and T m ¼33 C, respectively) than those seen for the C2:FAM-RNA 2 and C2:FAM-RNA 1 (T m ¼24 C and T m ¼27 C, respectively) under identical conditions correlates well with the smaller net-negative charge of the conjugate C1 (i.e. À2), as compared to that seen for the conjugate C2 (i.e. À5), and the corresponding reduction of repulsive forces with the negatively-charged FAM-RNA 1 or FAM-RNA 2 . The lower level of hyperchromicity of these transitions (cf. 16.5% and 20.9% hyperchromicity for C2:FAM-RNA 2 and C2:FAM-RNA 1 , respectively, in water against 23% hyperchromicity for C2:FAM-RNA 1 in Tris-buffered electrolyte) probably indicates that the degree of p-p stacking interactions within the C2:FAM-RNA 2 and C2:FAM-RNA 1 complexes formed in water is considerabley lower than that within the regular, doublestranded duplex (e.g. C2:FAM-RNA 1 in buffered electrolyte), formed by the fully-complementary oligonucleotides.
The dually-labelled C3:FAM-RNA 1 formed between the Cy3labelled conjugate C3 and the complementary RNA target showed a 4 C increase in T m value in buffered electrolyte, as compared with its unlabeled counterpart, C2:FAM-RNA 1 (cf. Figure 7(F) and Figure 7(A)), presumably due to extra molecular interactions (i.e. between the Cy3 and FAM dyes). This observation supports the proposed earlier hypothesis on the formation of the ground-state complex between the closely located fluorophores upon hybridization of C3 with the complementary target FAM-RNA 1 (Figure 6(A)), which led to a considerable decrease of the fluorescence ( Figure 5(A,B)), presumably, due to 'contact quenching', and to a change in the UV-visible spectrum ( Figure 5(E)) for C3:FAM-RNA 1 complex in buffered electrolyte. The melting temperature profiles of the C3:FAM-RNA 2 complex between the Cy3-labelled conjugate C3 and the non-complementary RNA target in water (Figure 7(I)) showed T m value of 27 C. Again, the duallylabelled system C3:FAM-RNA 2 showed a small, but reproducible increase in T m in water, as compared with its monolabelled counterpart, C2:FAM-RNA 2 (cf. Figure 7(I) and Figure  7(G)). For convenience, Table 1  We also recorded T m profiles for the free conjugates C1 and C2, as well as for the isolated FAM-RNA 1 and FAM-RNA 2 in the absence or presence of the peptide, which were measured either in water or in Tris buffer (see Supplementary Figure S15). In contrast to the co-operative behavior of the C1:FAM-RNA 1 , C1:FAM-RNA 2 , C2:FAM-RNA 1 and C2:FAM-RNA 2 complexes during their thermal denaturation (see Figure 7), the isolated components C1, C2, FAM-RNA 1 and FAM-RNA 2 did not show the sigmoidal character of their T m profiles, thus indicating the absence of any folding or stable secondary structures. The only slight exception was seen for FAM-RNA 1 in Tris buffer, which showed a minor low-temperature transition at 5 C-15 C interval with mild (6.7%) hyperchromicity, which could signify the presence of short, unstable duplex-like regions. In Tris buffer, the addition of the unconjugated peptide to RNA 1 or RNA 2 had no effect on the shape of the T m profiles (see Supplementary Figure S15, C and E), thus confirming the absence of any stable RNA-peptide interactions in the presence of counter cations at sufficient concentrations. However, we could not fully exclude such possibilities when these experiments were carried out in deionized water (Supplementary Figure S15, D and F), although the impact of the peptide on the shape of the RNA T m profiles was only minor and not well-defined.

NMR evidence of conjugate folded structure
The ability of the positively-charged amphipathic peptide to interact with the negatively-charged sugar-phosphate backbone of nucleic acids in the absence of counter cations, demonstrated in this research, suggested the high probability of strong intra-molecular interactions between the peptide and oligonucleotide components within the same conjugate molecule. In order to investigate this possibility, we recorded a 1 H-NOESY spectrum (500 MHz) of C2 in D 2 O Table 1. Summary of hybridization monitored by fluorescence and T m profiles for six studied systems, C1:FAM-RNA 1 , C1:FAM-RNA 2 , C2:FAM-RNA 1 , C2:FAM-RNA 2 , C3:FAM-RNA 1 and C3:FAM-RNA 2 . Experiments were carried out in Tris buffer (50 mM Tris,200 mM KCl,0.5 mM EDTA,pH 7.0) or in water at 5 C.

Conjugate-RNA system
Relative change of fluorescence intensity on hybridization (%) a T m ( C) (Hyperchromicity, %) b Â 100%, where F(RNA) refers to the fluorescence intensity of the test cuvette containing the target sequence alone (i.e. FAM-RNA 1 or FAM-RNA 2 ) and F(Complex) refers to the fluorescence intensity of the test cuvette after the addition of the conjugate. Experimental errors in estimation of relative change of fluorescence intensity on hybridization usually ranged between 1.4 and 2%. b T m (%; in bold) refers to the melting temperature ( C) of the hybridized complex, measured from the UV-monitored melting temperature profiles at 260 nm. Figures in brackets (in italics) represent the level of hyperchromicity (%) measured from T m profile. ND -Not detectable due to absence of sigmoidal T m profile. Table 2. Through-space 1 H-1 H NOE interactions detected from inter-nucleotide and peptide-nucleotide interactions, as well as from intra-and inter-peptide interactions observed in the 1 H NOESY spectrum (Figure 8) of the conjugate C2.
In summary, we provide NMR evidence of a stable, hairpin-like folded conformation of the conjugates, which seems to dominate in water and can be stabilized by p-p stacking interactions between nucleotide aromatic bases, as well as by inter-residue contacts between oligonucleotide and peptide fragments.

Dimer formation
The amphipathic nature of the peptide component within POCs, and the fact that each conjugate contains both positively-charged (i.e. peptide) and negatively-charged (i.e.  Table 2. Nomenclature and numbering of hydrogen atoms used for assignment of the peptide and oligonucleotide 1 H NMR signals is shown in Figure S1 and Figure S2, respectively (see Supplementary Material). (B) Expanded region of the 1 H NOESY spectrum of the conjugate H 2 N-Gly-[Arg-Leu] 4 -5 0 p 1 T 2 T 3 T 4 T 5 C 6 -A 7 A 8 T 9 C 3 0 (D 2 O, 5 C, 500 MHz, 2.5 mM, 0.5 s mixing time), showing NOE interactions between the aromatic (H6/H8) and the H1 0 protons of the oligonucleotide part of the conjugate C2. The route for sequential assignment was based on the 1 H NOE connectivity: $ (H6/H8) n $ (H1 0 ) n $ (H6/H8) nþ1 $ (H1 0 ) nþ1 , which is indicated by a red line in the spectrum. This network shows through-space contacts between the H1 0 sugar ring protons of the residue n with its own aromatic H6/H8 protons as well as with the H6/H8 protons of the next (n þ 1) residue. Cytidine H5-H6 interactions for the 5 C and 9 C residues are also indicated. oligonucleotide) components, suggested that they may potentially form inter-molecular assemblies. We investigated this possibility by analysis of changes in hydrodynamic radius of one of the conjugates (C1) by measuring its molecular diffusion in aqueous solution using Diffusion Ordered Spectroscopy (Figure 9). The diffusion constant D of a molecule can be correlated to its shape and size by the Stokes-Einstein equation (Valentini et al., 2004): where D is the diffusion coefficient, k is Boltzmann constant, T is the temperature, g is the viscosity of the solution and r s is the hydrodynamic radius of the particle.
The diffusion coefficient D of the conjugate was measured by 1 H DOSY Provencher, 1982aProvencher, , 1982bValentini et al., 2004) from the ratio between the intensity of the attenuated NMR signal I and the initial signal I 0 (as described in the 'Material & Methods' section) using the Stejskal-Tanner equation (Kerssebaum & Salnikov, 2006): where D is the diffusion coefficient, c is the gyromagnetic ratio, g is the gradient field pulse strength, d is the gradient field pulse length and D is the diffusion time.
The diffusion coefficients of various molecular species, which often exist in a fast exchange within the complex sample, may have close values that are difficult to resolve due to a continuous distribution of particle sizes. To mitigate this matter, we used an inverse Laplace transform (also known as CONTIN algorithm) (Provencher, 1982a;Valentini et al., 2004) in an attempt to provide the distribution of the diffusion coefficients for C1 and, if possible, to resolve the individual diffusion coefficients in the mixture of different molecular species with close diffusion coefficients. C1 conjugate molecules were present in solution with diffusion coefficients ranging between 2.37e À10 and 1.29e À10 m 2 /s (Figure 9), suggesting predominant molecular species with close molecular radii ranging from 1.0 nm to 2.0 nm, which corresponds to particle diameters ranging from 20 Å to 40 Å.
The estimated overall length of the folded structure of the C1 conjugate, evaluated using molecular graphics ranged between 21 Å and 23 Å, which shows reasonable agreement with the lowest size limit (20 Å) of the molecular species detected by DOSY, which could be attributed to the monomeric forms of the C1 conjugate. The other particles detected seem to correspond to lower diffusion coefficients (up to 1.29e À10 m 2 /s) and larger sizes (up to 40 Å), which might be attributed to higher-level of aggregates (e.g. dimeric forms). DOSY provided possible evidence that C1 molecules in aqueous may self-assemble into particles ranging in size between 20 Å and 40 Å, presumably stabilized by electrostatic interactions between the positively-charged peptide of one conjugate molecule and the negativelycharged oligonucleotide of a neighboring molecule. This selfassembly of the conjugates into supra-molecular structures might be a key factor contributing to catalysis. A number of previous studies highlighted the non-linear enhancement of the cleavage activity for this type of POCs, triggered by concentration increase, when possible formation of catalytic multiplexes would be expected (Kovalev et al., 2008;Patutina et al., 2017Patutina et al., , 2019Staroseletz, Williams et al., 2017).

Conclusion
This research provides key information on the interactions of functional catalytically-active bioconjugates with RNA, essential for development of novel therapeutic agents with controlled chemical and biological properties, capable of recognizing and destroying disease-relevant RNAs in a highly selective manner to enhance potency and reduce potential toxicity in humans. By looking at the individual roles of the recognition and biocatalytic moieties of the conjugates toward RNA, we provide insight into the behavior of such bioconjugates against complementary and non-complementary targets and illuminate our understanding of how these bioconjugates interact with RNA sequences and catalyze their cleavage.
Very different behavior of the peptidyl and oligonucleotide components of conjugates is here revealed as selective interaction under buffered electrolyte, representative of intracellular conditions, whereas nonselective interactions arose under non-physiological conditions in water. We demonstrated here that the oligonucleotide component of the peptidyl-oligonucleotide conjugates dominate interactions between conjugates and complementary targets in the presence of physiological levels of counter cations (e.g. K þ ), sufficient to prevent repulsion between the complementary nucleic acid strands. The important consequence of this is that in physiologically-relevant conditions, when counter cations screen the negatively-charged sugar-phosphate backbones of the RNA target, the positively-charged catalytic peptide shows poor or negligible nonspecific interactions with RNA, thus minimizing or avoiding nonspecific RNA cleavage. Indeed, the peptide component only became the key player in these interactions, when counter cations were insufficient for charge screening. This is a significant finding implying that stable nonspecific complexes between such peptidyl-oligonucleotide conjugates and RNA sequences can only be formed under non-physiologically conditions, in the absence of physiological concentrations of counter cations (e.g. in de-ionized water), when nonspecific cleavage by the complexed peptide is also insignificant.
Based on these findings, it is reasonable now to assume that the 'point of contact' between conjugates and RNA can be switched between the oligonucleotide and peptide components by simple manipulation with a few key parameters, including (i) the length and thermodynamic stability of the complementary region between the oligonucleotide recognition motif and RNA target, (ii) the availability of counter cations and (iii) the net charge of the conjugate molecules. Strong complementary binding of conjugates through the oligonucleotide recognition component may provide precise sequence-selectivity for RNA targeting, but discourage the ability of the conjugate to leave RNA after each catalytic event, thus compromising the reaction catalytic turnover. In contrast, the transient, nonspecific contacts via the peptide component offer an opportunity for conjugates to 'cleave and leave' RNA; however, this mode of binding suffers from poor target selectivity, which may ultimately lead to high systemic toxicity. Finding an appropriate balance between the sequence selectivity and sufficient level of catalytic turnover will rely on fine tuning and matching between the above parameters.
Our recent work (Patutina et al., 2017(Patutina et al., , 2019Staroseletz, Williams et al., 2017;Williams et al., 2015) and earlier studies from other groups (Mironova et al., 2002;Mironova, Boutorine et al. 2004;Mironova et al., 2006Mironova et al., , 2007Pyshnyi et al., 1997) demonstrated that the biological activity of this class of bioconjugates can only be achieved through chemical conjugation of the catalytic peptide with the oligonucleotide component. This suggested that the oligonucleotide component may induce an 'active' conformation of the peptide to make it catalytically active by some unclear molecular interactions between the oligonucleotide and peptide components. We provide here the first direct structural evidence of the conformation necessary for sequence-selective catalysis, and the first NMR evidence of a stable, folded, hairpin-like structure of conjugates in water. Prediction and rationalization of the binding properties and hydrolytic activity of such conjugates is now strongly facilitated to allow a robust 3D structural determination of the conjugates alone and within the hybridized complexes with RNA sequences by a combination of high-resolution multi-dimensional NMR and molecular modeling. NMR spectroscopy can be used not only to illuminate the direct contacts between the different structural elements of the conjugates, but also to evaluate inter-nuclear distances and torsion angles, which can then serve as constraints in restrained molecular dynamic calculations using the full relaxation matrix approach. This research, which is currently in progress, will provide more detailed and clear understanding of the structure-activity correlations for this class of chemical ribonucleases to facilitate future rational design of the new structural variants. Harmesh Aojula for his invaluable assistance in performing peptide synthesis, especially in completing the final global deprotection and cleavage of the peptides from the resin using HF.