Structure Validation of G‐Rich RNAs in Noncoding Regions of the Human Genome

Abstract We present the rapid biophysical characterization of six previously reported putative G‐quadruplex‐forming RNAs from the 5′‐untranslated region (5′‐UTR) of silvestrol‐sensitive transcripts for investigation of their secondary structures. By NMR and CD spectroscopic analysis, we found that only a single sequence—[AGG]2[CGG]2C—folds into a single well‐defined G‐quadruplex structure. Sequences with longer poly‐G strands form unspecific aggregates, whereas CGG‐repeat‐containing sequences exhibit a temperature‐dependent equilibrium between a hairpin and a G‐quadruplex structure. The applied experimental strategy is fast and provides robust readout for G‐quadruplex‐forming capacities of RNA oligomers.


Introduction
G-quadruplexes are noncanonicali nter-or intramolecular structural motifs formed by G-rich DNA or RNA sequences. Their fundamental buildingb locks are planar tetrads composed of four guaninen ucleobases (G-tetrad) that engage in Hoogsteen-type base pairing (Scheme 1A). [1,2] Typically,t wo to four [3] of these tetrads are stacked on top of each other to form the structural motif, though larger assemblies have been reported as well (Scheme1B). [4,5] G-quadruplex folding requires monovalent cations,s uch as K + or Na + ,b ecause the positive charge stabilizes the partially negatively charged O6 atoms of the guanine bases. [6] Although DNA G-quadruplexes show strong structural polymorphism due to loop variety,l eading to various possible folding motifs, [7] the presence of the 2'-OHg roups in RNA G-quadruplexes favors aC 3 '-endo sugar pucker and an allparallel folding topology (Scheme 1B). [8] The strong stacking interactions between the tetrads lead to ar emarkable thermodynamic stability, [9] with the stabilityo fR NA G-quadruplexes exceeding that of DNA G-quadruplexes. [10] DNA G-quadruplexes can be detected in vivo [11] and their fundamental role in telomere maintenance as well as in gene regulation is generally acknowledged. [12] Bioinformatics and in vitro studies [11,13] suggested the presence of G-rich putative quadruplex-forming sites in untranslated regions (UTRs) [14] of mRNAs, as well as in the transcripts of human telomersk nowna st elomeric repeatcontaining RNA. RNA G-quadruplexes might be involved in processes such as polyadenylation, [15] modulation of translational efficiency, [16,17] and splicing. [18] However,t he existence of RNA G-quadruplexes in vivo is am atter of current scientific debate. [19] In 2016, in-cell mapping experiments suggested that G-rich cellular transcriptsc apable of formingG -quadruplex structures after refolding in vitro are largely unfolded in eukaryotic cells. [20] Later studies that combined RNA G-quadruplex-specific precipitation with sequencing were able to detect transientf ormation of RNA G-quadruplex structures in human cells. [21] Collectively,t hese findings suggest the existence of RNA G-quadruplexes in vivo as dynamic structures, their folding/unfolding dynamics governed by the cellular machinery (possibly helicases).
By using high-throughput probingm ethods, deep sequencing, and bioinformatics, severals tudies have identified G-rich motifst hat, it was speculated, might form G-quadruplexes. [22][23][24][25] Moreover, investigationo ft he interaction between RNA Gquadruplexes and protein counterparts often involves the use of short G-rich oligomers [26] that are assumed to form G-quadruplexes in vitro, but without appropriate experimentale vidence in support of this assumption. High abundanceso fg uanosiner esidues in RNA transcripts can indicate G-quadruplex formation,b ut structure, stability, and other biophysical properties remainu nclear withoutd etailed biophysical characterization of these G-rich motifs. Althought he RNA G-quadruplex topologies are limited in terms of strand orientation by the strongp reference of RNA to form all-anti all-parallel G-quadruplexes, dimerization or multimerization can lead to an umber of different general topologies that can alter biological function immensely (Scheme 1). While focusing primarily on the biological contexta nd putting great emphasis on the role of G-quadruplexes in regulatory systems, we wish to remarkh ere We presentt he rapid biophysical characterization of six previously reported putative G-quadruplex-formingR NAs from the 5'-untranslated region (5'-UTR) of silvestrol-sensitive transcripts for investigation of their secondary structures. By NMR and CD spectroscopic analysis, we found that only as ingle sequence-[AGG] 2 [CGG] 2 C-folds into as ingle well-defined G-quadruplex structure. Sequencesw ith longer poly-G strands form unspecific aggregates,w hereas CGG-repeat-containing sequences exhibit at emperature-dependent equilibrium between ah airpin and aG -quadruplex structure. The applied experimental strategy is fast and provides robust readout for G-quadruplex-forming capacities of RNA oligomers. that many cell biology studies do not involve any (or involve only sparse) biophysical characterization of G-quadruplex structures. [27][28][29][30][31] Herein, we show that such advanced biophysical structural characterization can be time-and cost-effectivea nd should be mandatory for any high-profiles tudy.I ts hould be performed in order to establish whether putative G-quadruplex-forming sequences are indeed actuallyformingG-quadruplex structures under the tested in vitro experimental conditions.
After an illustration of the general protocol for ap reliminary biophysical screening, the analysis of six putative G-quadruplex-forming RNAs from the 5'-UTR of silvestrol-sensitive transcripts, previously reportedi nN ature, is presented in detail. [32] As imple biophysical protocol for validation of the formation of G-quadruplex structures We propose as tepwises creening method involving CD and NMR spectroscopic studies and incorporating increasingly sophisticated spectroscopic methods in each step, ultimately providing as trong dataset for the folding potential of RNA sequences predicted to form G-quadruplex structures.
In af irst step, CD spectroscopy is used to determine the thermodynamic stabilities of thep utative G-quadruplex structures. Additionally,g eneral information about the folding topology of the G-quadruplex can be inferred from the CD signature ( Figure 1). [33] As trong response to addition of am onovalentc ation (typically K + )i sa ni ndicator for G-quadruplex formation; however,t he formation of ad ifferents tructure cannot be excluded unambiguously.  AlthoughC Ds pectra are helpful for determining G-quadruplex strand topology,t he questiono fw hether or not aGquadruplexi sf ormed cannotb ea nswered conclusively because the spectra of other structures such as G-wires [34] can look comparable or identicalt ot hose of G-quadruplexes. 1D 1 HNMR spectroscopy can help in identifying aG -quadruplex unambiguously,b ecause reporter signals of imino atoms involved in G-tetrads resonate at ac haracteristic frequency of around1 1ppm. Other than determining the number of guanosine residues involvedi nt he G-quadruplex, 1D NMR data lack furthers tructurali nformation, but 2D NMR spectroscopy and, in particular, 2D 1 H, 1 HNOESY can serve as tools for more advanceds tructural characterization in at hird step. [35] Essentially, 2D 1 H, 1 HNOESYspectra contain all distances below athreshold of % 6 between the protons of the oligonucleotide. Although these data mighte ven be sufficient for the calculation of a basic 3D structure,t he assignment of all protons can be tedious and time-consuming and requires data from other 2D NMR methods. Even withouta ni n-deptha nalysis, however, the general structural features of aG -quadruplex can be determined by analysiso f2 D 1 H, 1 HN OESY spectra.F urther, NMRspectroscopic methods such as DOSY [36] or heterocorrelated NMR can yield additional structural information.
By following these increasingly complex steps, it is possible to achieve the biophysical and structuralc haracterization of any small G-quadruplex-forming oligonucleotide. It shouldb e noted that CD and NMR measurementsr equire0 .05-0.5 mm samples in av olume of approx. 0.5 mL, aq uantity that can easily be ordereda nd delivered within only af ew working days at comparatively low cost.
Acase study-silvestrol-sensitive G-rich transcripts from human 5'-UTR Recent studies have suggested the involvement of G-quadruplexes in the functiono ft he anticancer therapeutic silvestrol, [37] which inhibits the initiation factor eIF4A in human T-ALL-infected cell lines. [38] Wendel et al. found an accumulation of G-rich sequences in the 5'-UTR of human mRNA that experienced downregulation of the translational efficiency under the influenceo fs ilvestrol. [32] Several G-rich sequential motifs that showedaCD profile indicative of G-quadruplex formation were identified.
In this study,w eh ave characterized the tendency of these six short putatively G-quadruplex-forming RNAs from the 5'-UTR of silvestrol-sensitive mRNA transcripts to undergo G-quadruplex formation.B yu sing the protocol described above, including NMR spectroscopy,w ew ere able to monitord irectly the secondary structures actually formed and to assess the influence of varying conditions, such as K + ,c oncentration, and temperature. Five of the G-rich RNA sequences, screenedi n this work (Table 1), are among the most silvestrol-sensitive RNA transcripts, as determined in studies by Wendele tal. [32] In addition, two flankingU residues were added to EP300 to assess their effect on the overall structure and to impede Gquadruplex stacking, [39] resulting in the sequence UEP300U.
CD spectroscopic examination of TGFB1, MTA2, and MAP-KAPK2s howed ap rofile indicative of an all-parallel G-quadruplex structure [40,41] with am aximum at 262 nm and am inimum at 240 nm at 20 mm,2 0mm,a nd 10 mm KCl, respectively ( Figure 1, top left in each panel). No spectralc hanges were visible after addition of more K + ions to the system ( Figure S1 in the Supporting Information). Even without K + ions, am aximum in ellipticity at 262 nm was already observable.G -tetrad formation requires monovalentc ations, so this is atypical for a G-quadruplex and hints at the formationo fadifferent secondary structure. Them elting points could be determined by CD meltingc urve analysis ( Figure 1, top right in each panel)a nd were determined as 54, 43, and 65 8Cf or TGFB1, MTA2, and MAPKAPK2, respectively.
The type of secondary structure was further investigatedb y using NMR spectroscopy.T he 1D 1 Hs pectra of all three sequences showed only av ery broad signal in the imino region between 10 and1 2ppm ( Figure 1, bottom panels). This region is typical for Hoogsteen-paired residues as observed in G-quadruplexes. [42] Although G-quadruplexes, as compact structures, show distinct peaks in the imino region in 1D 1 HNMR spectra, we assumet hat these G-rich RNA sequences form ah igherorder polymorphic structure. The NMR data suggestthe formation of high-order unspecific aggregates, because large structures lead to broadening of peaks in NMR spectra,w hile the peak positions differ throughoutt he numerous different possible lengths of such structures. [43] These aggregates might interact through GG N1-carbonyl symmetric base pairs, which can be formed even in the absence of monovalent cations, asr ecently shown by Plavec et al. [44] Addition of KCl did not lead to any observable change in the NMR spectra,w hereas the CD spectra,atl east in the case of TGFB1, showedt hat arearrangement takes place. The strong positive signal at 262 nm is also in agreement with the formation of G-wires, [5] as observed by Protozanova and Macgregor. [34] For largera ggregates of those G-wires,t he detection of ab road signal in 1D 1 HNMR spectra is expected. [45] In native PAGE experiments the bands of TGFB1 and MAPKAPK2a re strongly broadened, thus supporting the proposed foldings cheme of highly polymorphic structures ( Figure S2). MTA2 shows, apart from the same broad bands, a small defined band with an intensity that diminishes upon refolding ( Figure S2). ADAM10 showedn os ignificant circulard ichroism signal withoutt he addition of KCl (Figure 2, top left). This contrasts with the CD data for TGFB1, MTA2, and MAPKAPK2 (Figure 1), because those sequences showed strong signals at around 260 nm even without addition of KCl. When,h owever,K Cl was added to as ample of ADAM10, ap ositive ellipticity at 260 nm could be observed. The end point of the titration was reached at 30 mm KCl, and no further increase in ellipticity could be measured. CD melting analysis was thus carriedo ut at aK Cl concentration of 30 mm.T he meltingc urve (Figure 2, bottom left) shows ac lear sigmoidal profile, and the meltingp oint could be determined as 44 8C.
G-quadruplexf ormation could be confirmed by 1D 1 HNMR data (Figure 2, right). In the absence of KCl, no signal apart from as mallb roadened bulge was observed in the imino region. This hints at the formation of unspecific aggregates, analogously to the cases of TGFB1,M TA2, and MAPKAPK2. After KCl had been added, however,t hree sharp separated resonances and severalo verlapped signals at 11.2-11.3 ppm could be detected. Eight signals hint at as tack of two G-tetrads, as can be expected from the sequence [AGG] 2 [CGG] 2 C.
The formation of ah ighly symmetric dimer,t hough, cannot be ruled out by 1D 1 HNMR spectroscopy alone. [46] We acquired 1D 1 Ha nd 2D 1 H, 1 HNOESY spectra at ah igher concentration (700 mm)a nd discovered two additional imino protons ignals, resonating at 10.3 and 9.3 ppm ( Figure 3A), that were only faintly visible at 50 mm.Apossibleh ypothesis that could explain this strong chemical shift perturbation might be ring currente ffects due to the stacking of terminal or loop residues on the G-tetrad. We investigated the general topology of the G-quadruplex by analysiso ft he 1 H, 1 HNOESY spectrum (Fig-Figure 2. To pl eft:CDspectrum of ADAM10 in the presence of 0( a)and 30 mm (c)K Cl;30mm KCl marked the endpoint of the titration (Figure S1). The sample contained1 0mm RNA and 10 mm BisTris·HCl buffer (pH 6.8). Bottom left:CDmelting curve of ADAM10 at 262 nm in the presence of 30 mm KCl. The melting pointwas derivedf rom sigmoidal fitting. The sample contained 10 mm RNA in 10 mm potassium phosphate buffer (pH 6.8). Right:Imino region of the 1D 1 HNMR spectrum of ADAM10 in the presenceo f0and 100mm KCl. Samples contained100 mm RNA, 25 mm Bis-Tris·HCl buffer (pH 6.8), and 10 %D 2 OinH 2 O. Figure 3. Further investigation of ADAM10 by 2D NMR spectroscopy. A) Imino region of the 1D 1 Hs pectrum. B) Imino-aromatic region,highlighting strand interactions. C) Four-tetrad G-quadruplexl ayout as suggested from 2D data. D) CD melting curve at an RNA concentration of 300 mm. E) Full iminor egion of the 2D 1 H, 1 HNOESY spectrum, highlighting two peakss howing three NOESY crosspeaks. F) DOSY of ADAM10s howing zoomso fthe 1,4-dioxane reference peak and the aromatic RNA peaks.
ChemBioChem 2020, 21,1656 -1663 www.chembiochem.org 2020 The Authors. Publishedb yWiley-VCH Verlag GmbH &Co. KGaA, Weinheim ure 3B,E ). On inspection of the imino proton crosspeaks, we found that crosspeaks from imino protons ignals to three other signals are visible ( Figure 3E,s olid and dashed lines);this would not be expected from at wo-tetrad G-quadruplex but could be explained by these residues lying between two tetrads, as in three-or four-tetrad G-quadruplexes. In this case two imino proton crosspeaks to the adjacent tetrads and a third crosspeak to the neighboring base in the same tetrad would explain the observation. Additionally,t he imino to aromatic crosspeak region shows that it is possible to follow cross-signals over more than two H1-H8 layers( Figure 3B).
From these observations, the topology of the ADAM10 Gquadruplexw ould be expected to include four G-tetrads, so a bimolecular G-quadruplex should be formed. We were ablet o confirm the formationo fs uch ah igher-order structure by DOSY spectroscopy (Figure 3F), in which we determined ah ydrodynamic radius of 13.7 .Acomparable value of 13.9 was observedi nt he four-tetrad G-quadruplex of the modified DNA GG-Az1-GG. [46] The DOSY peaks of ADAM10 and of the reference substance 1,4-dioxane are shown in Figure 3F.A dditional peaks with ah igher diffusion coefficient are observed, corresponding to an even higher-order structure with ah ydrodynamic radius of 18.8 that is partially populated.
To providea dditional verificationo ft he formation of dimerization we conducted CD melting experiments at ah igher concentration (300 mm,F igure 3D). We observed ar ise in the melting temperature from 44 to 71 8C, in comparison with the lower concentration (10 mm)i nvestigatedb efore (Figure 2). From the number of imino peaks observed, symmetry can be proposed. Combined with the requirement of an all-anti allparallel G-quadruplex,t his leads to two possible topologies ( Figure 3C). These differ in the stacking tetrads,w hich feature the 5'-o r3 '-terminal residues. [47] The exact determination of the topology includingt he G-residue polarity would require a full assignment of the NMR signals.O nly in rare cases can this be achieved from NOESY data alone;m ost commonly it involves isotopic labeling of single nucleotides in various NMR samples [48] or uniform labeling of one NMR sample. These techniques require considerable preparative effort,s ot heir application is beyond the scope of at opological screening,a sc arried out here.
In circulard ichroism spectra of EP300 ( Figure 4A,t op left) only moderate changes in ellipticity were visible upon addition of KCl. Whereas the peak at 260 nm only showeds light variation in intensity,asmall positivep eak at 240 nm and an egative peak at 290 nm appeared at high KCl concentrations. Subsequently,c irculard ichroism at 260 nm was measured over a temperature range between 5and 95 8C, revealing biphasic behavior with two different transition points:2 5a nd 44 8C. Subsequent analysis of the NMR imino proton region on titration with KCl, as well as of its temperature dependence, led to the same finding. At 0 8Cs ignals with equal intensity could be observedi nt he regions characteristico fH oogsteen base pairs (10-12 ppm) and of Watson-Crick base pairs (12-14 ppm). Upon heating, the low-field signals diminished, vanishing completely above 30 8C. Simultaneously,t he intensity of the highfield imino signals increased. This observation indicates at her-mal equilibrium between ad uplex or hairpin and aG -quadruplex, with the G-quadruplex showing ah ighert hermal stability. The two states showed different electrophoretic mobilities in native PAGE,r esulting in two slightly separated bands at 4 8C but in no such observation being made at 40 8C ( Figure 4,  lanes 1-3). The electrophoretic mobility of the duplex or hair- pin structure is comparable with that observed for the G-quadruplex structure.
The same set of experiments was conducted on UEP300U, with comparable results. Interestingly,a ddition of two flanking Uresidues to the sequences enhances the thermal stabilityo f the low-temperature structure (possibly duplex or hairpin) relative to that of the G-quadruplex.T hus, imino signals of Watson-Crick bound residuesw ere observed even at 40 8C, whereas in the case of EP300 these signals had completely vanisheda tt his temperature. Analysis of the CD meltingc urve supports this observation, with the melting point of the lowtemperature structure being shiftedt o3 08Cf or UEP300U. The CD profiles of UEP300U showedamore pronounced potassium dependency then those of EP300. The spectral changes upon addition of small amounts of KCl were subtle and led to a spectrumc haracteristic of an all-anti all-parallel quadruplex, whereas at higher concentrationsapositive signala t2 40 nm and an egative signal at 290 nm were observed, accompanied by ad ecrease in the positive signal at 260 nm. Such spectra have previously also been observed in the cases of Z-DNA [49] and Z-RNA. [50] These helicesc an form from purine-pyrimidine repeat sequences. [51] Hairpinf ormation with two [CG] base pairs and aG Gm ismatchh as been observed previously, [52] so formation of ad uplex mimicking the CD characteristics of Z-RNA at high salt concentrations could explain the unusualC D spectralcharacteristics of EP300 and UEP300U. In contrast, Rypniewski et al. observed an A-type helical structure of CGG repeatsi nc rystallographic studies. [53] As in the case of EP300, the two structures of UEP300U at lowt emperature can also be observed in native PAGE (Figure 4, lanes 4-6).

Discussion
Herein, we report at hree-step protocol for the characterization of putative G-quadruplex-forming RNA oligomers.T he protocol involves CD and NMR screening, including CD meltingc urve analysisand 1D and 2D NMR.
We have demonstrated the application of the protocol by using NMR and circular dichroism studies to probe the structural preferences of six different G-rich RNAs-from the 5'-UTR of human mRNA-thata re involved in the modulation of eIF4A suppression by silvestrol. It was found that three of the sequences, despite being G-rich, did not form distinct G-quadruplex structures,b ut produced aggregates in an unspecific manner.T he [XGG] 4 repeats equences ADAM10, EP300, and UEP300U, however,d id fold into G-quadruplexes, with EP300 and UEP300U existing in thermal equilibrium with hairpin structures.
Circulard ichroism studies of four of the putative G-quadruplex-forming sequences showedh igh spectral similarities, especially under the influence of KCl. CD spectra of TGFB1, MTA2, MAPKAPK2, and ADAM10each showed astrong positive band at 260 nm after addition of KCl, with differences being merely visible beforet he addition, when the RNA was assumed to be unfolded. BeforeK Cl addition, MTA2, MAPKAPK2, andto some extent-TGFB1 already showed ap ositive peak. By using NMR spectroscopy and PAGE we were only able to con-firm G-quadruplex folding in the case of ADAM10, which forms af our-layer quadruplex.T his quadruplex shows al ow thermal stabilityw ith am eltingp oint of only 44 8Ca tac oncentration of 10 mm.A t3 00 mm this rises to 71 8C, likelyd ue to dimer formation,w hich leads to af our-tetrad structure. These have already been observed to be thermodynamically more stable than two-tetrad G-quadruplexes. [54] The bimolecular natureo f this quadruplex is only evident through 2D NMR investigation but could be confirmed by DOSY NMR spectroscopy.N MR spectra revealed that TGFB1, MTA2,a nd MAPKAPK2 form unspecific aggregates of high molecularity with or without addition of KCl. The absence of am onovalent cation hinders Gtetrad formation, so interaction through GG N1-carbonyl symmetric base pairs is assumed in this case. Under the influence of potassium, ac hangei nt he CD spectrum of TGFB1 suggests the formation of G-wires, whereas the behavior of MTA2 and MAPKAPK2r emains unclear.B ecause NMR spectroscopy fails to resolve structures of such as ize, those structural preferences cannotb ec onfirmed beyond doubt, but G-quadruplex formation can be ruled out. The structural polymorphism is supported by native PAGE, with the gel bands being broadened almostb eyondd etection. The small defined band of MTA2, which is observed alongside the broad main band, diminishes upon refolding, thus indicating as mall thermodynamic barrier to aggregate formation.
The structure of CGG-repeat-containingR NA is as ubjecto f current scientific debate. [55,56] Contributing to this, we collected CD and NMR spectroscopicd ata fort he sequences [CGG] 4 (EP300) and U[CGG] 4 U( UEP300U).C GG repeats are known to form helical structures with an on-Watson-Crick GG base pair. [52] We were able to identify Watson-Crickb ases in the NMR spectra of EP300 and UEP300U accordingly.A dditionally a high-temperature structure exists in at emperature-dependent structurale quilibrium. It is assumed to be aG -quadruplex because the involved imino protons resonate in the region typical of Hoogsteen-bound Gresidues.A lthough the temperature-dependent transitionb etween two structural states was determined by CD melting studies, the structural characteristics of each state could be deconvoluted only by NMR. We were ablet os how that high temperature favorst he formation of the G-quadruplex overt he duplex. The transition temperature from hairpin to G-quadruplex was higher in the case of the uridine-flanked sequence UEP300U. This might be because of the capping effect of dangling ends, which stabilizes RNA duplexes. [57] In addition, flanking sequences can negatively affect the stabilityo fG -quadruplex structures; [58] this adds to the higherr elative stabilityo ft he hairpin. Even though Gquadruplexes are stabilized by monovalent cations, the duplex seems to be the preferred conformation at high salt concentrations, if room temperature data as observed from CD experiments are considered. Despite CGG repeatsf orming an A-helix in crystals and Z-RNA in general forming at very high salt concentrations, [59] CD spectra of EP300 and UEP300U hint at duplexes that exhibit characteristics of Z-RNA. The reason fort his behavior remains unclear.H owever,s uch structurale quilibria are of great biological importance and can be tuned by cellular key factors such as cation levels [60] or tRNA concentration. [61] The electrophoretic mobility of the duplex structurei ss lightly lower than the mobility of the G-quadruplex in both RNAs. Because the electrophoreticm obility in native PAGE depends not only on size, but also on compactnesso fthe structure, the molecularity cannot directly be deduced from this data point. NMR spectra of the duplex showa tl east five imino signals, thus hintinga tabimolecular duplex.

Conclusions
We were able to elucidate the conformational space of six small G-rich mRNA fragments, from human 5'-UTR of mRNA transcripts that are sensitivet os ilvestrol, by applyingathreestep screening protocol involving CD and NMR spectroscopy. Three of the six oligonucleotides do not fold into G-quadruplexes as expected, but insteada ggregate unspecifically. ADAM10( [AGG] 2 [CGG] 2 C) forms af our-tetrad all-anti all-parallel G-quadruplex. EP300 and UEP300U ([CGG] 4 andU [CGG] 4 U) each fold either into ad uplex or into aG -quadruplex depending on the conditions, in particulart emperature and salt concentration.G-quadruplex structures are believed to be involved in the therapeutic mechanisms of the anticancer drug silvestrol, so understanding of the structural characteristics of the investigated G-rich sequences is of great importance for discriminating between potentialG -quadruplexes and other G-rich sequences. By applying the three steps of our screening protocol in order,wewere able to show that CD spectroscopy,a lthough offering ar apid and cost-effective method for obtainingp reliminary information on the nature of as econdary structure, cannotb eu sed alone to assesst he actual conformation of a G-rich oligonucleotide. NMR spectroscopy is necessary to unravel the information obtained by CD spectroscopy and to shed light on the actual conformation(s) presenti ns olution. The protocol could be useful for obtaining all data points necessary for characterizationo ft he structuralc haracteristics of such oligonucleotides. Circular dichroism:A ll CD experiments were carried out with a Jasco J-810 CD spectrometer (Jasco, GmbH) and use of quartz optical cuvettes with 0.1 cm path length (0.01 cm path length was used for ADAM10 at 300 mm concentration). For CD titration experiments with KCl, RNA samples with ac oncentration of 10 mm in potassium-free BisTris·HCl buffer (pH 6.8, 10 mm)w ere prepared. The titration range was 0t o7 0mm KCl. Resulting CD spectra were baseline-corrected and corrected for sample dilution. The data were smoothed by application of aS avitzky-Golay filter [62] (10 points).

Experimental Section
For CD melting curves the samples contained the RNA of interest (10 mm,300 mm in one additional experiment for ADAM10), potassi-um phosphate buffer (pH 6.8, 10 mm), and potassium chloride according to the previously determined end point of the titration. The peak ellipticity was monitored over at emperature range from 5to958Cw ith ascan rate of 1 8Cmin À1 .
Native PAGE:A queous stock solutions of RNA were diluted to concentrations of 10-30 mm in TBE buffer (1 )w ith addition of glycerol (30 %) and KCl (50 mm). Polyacrylamide gels (15 %) containing TBE (1 )a nd KCl (50 mm)w ere cast and run with the samples at 0.5 to 0.8 Wf or 4hat 4o r4 08C. The running buffer contained TBE (1 )a nd KCl (50 mm). Gel visualization was achieved by GelRed staining and subsequent imaging under UV light.