Binding Between G-quadruplexes at the Homodimer Interface of the Corn RNA Aptamer Strongly Activates Thioflavin T Fluorescence

SUMMARY

Thioflavin T (ThT) is widely used for the detection of amyloids. Many unrelated DNAs and RNAs that contain G-quadruplex motifs also bind ThT, and strongly activate its fluorescence. To elucidate the structural basis of ThT binding to G-quadruplexes and its fluorescence turn-on, we determined its co-crystal structure with the homodimeric RNA 'Corn', which contains two G-quadruplexes. We find that two ThT molecules bind in the dimer interface, constrained by a G-quartet from each protomer into a maximally fluorescent planar conformation. The unliganded Corn homodimer crystal structure reveals a collapsed fluorophore-binding site. In solution, Corn must fluctuate between this and an open, binding-competent conformation. A co-crystal structure with another benzothiazole derivate, thiazole orange (TO) also shows binding at the Corn homodimer interface. As the bound ThT and TO make no interactions with the RNA backbone, their Corn co-crystal structures likely explain their fluorescence activation upon sequence-independent DNA and RNA G-quadruplex binding.

Graphical Abstract

eTOC Blurb

Thioflavin T (ThT) derivatives are fluorescent probes for amyloid and G-quadruplexes, and therapeutics for neurodegenerative diseases. Sjekloca shows that ThT binds between the G-quartets of the homodimeric RNA G-quadruplex-containing aptamer Corn and interacts exclusively with the exposed guanine bases.

INTRODUCTION

Corn is a 28 nucleotide (nt) in vitro selected RNA aptamer that binds and turns on by over 1,000-fold the fluorescence of DFHO (Figure 1A, 3,5-difluoro-4-hydroxybenzylidene-imidazolinone-2-oxyme, 1), a small molecule analog of the intrinsic fluorophore of red fluorescent protein (RFP). DFHO is cell-permeable and non-cytotoxic, and exhibits minimal non-specific binding to cellular nucleic acids. By soaking cells in DFHO, Corn was successfully employed as a genetically encoded fluorescent tag to visualize biological RNAs in live cells (Song et al., 2017). Crystallographic structure determination revealed that Corn homodimerizes and binds one molecule of DFHO at its interprotomer interface (Warner et al., 2017). Each of the protomers folds as a stem-loop, with a four-tiered mixed-sequence quadruplex distal to the stem; the two most distal tiers are both canonical G-quartets. In the complex, DFHO is held into a planar, maximally fluorescent conformation by stacking between two G-quartets, one from each protomer. Notably, the homodimer interface lacks any base pairs, and the bound DFHO is surrounded by six unpaired adenine residues (three from each protomer) that locally break symmetry. Biochemical analyses demonstrated that Corn activates DFHO fluorescence only as a dimer, and that the dimer is very stable (K_d < 1 nM) even in the absence of the small molecule, suggesting some degree of pre-organization of the DFHO binding site (Song et al., 2017; Warner et al., 2017).

Figure 1. Corn Aptamer RNA Activates the Fluorescence of Various Fluorophores.

(A) Chemical structure of fluorophores used in this study.

(B) Fluorescence spectra of DFHO (1), ThT (2), and TO (3), in complex with Corn RNA, normalized to the Corn-DFHO emission maximum.

Thioflavin T [Figure 1A, ThT, 4-(3,6-dimethyl-1,3-benzothiazol-3-ium-2-yl)-N,N-dimethylaniline, 2] is a benzothiazole dye widely employed for detection and quantitation of amyloids, upon binding which it exhibits an increase in fluorescence (Biancalana and Koide, 2010; Naiki et al., 1989; Vassar and Culling, 1959). Recently, it was reported that ThT binds to G-quadruplex DNAs and RNAs with micromolar dissociation constants and with little sequence preference. Association with G-quadruplexes gives rise to strong fluorescence enhancement, which is generally much lower or absent when ThT is mixed with single-stranded, duplex or triplex DNA or RNA (Gabelica et al., 2013; Li et al., 2016; Mohanty et al., 2013; Renaud de la Faverie et al., 2014; Warner et al., 2017; Xu et al., 2016). In solution, the benzothiazole and aniline rings of ThT lie predominantly at an angle to each other, which leads to non-radiative decay after photoexcitation. Binding to amyloid, or increase of the viscosity of the medium, enhances fluorescence by populating the conformation in which the two rings are coplanar (Biancalana and Koide, 2010; Stsiapura et al., 2008). Molecular dynamics simulations suggested that ThT would bind preferentially to the exposed guanine bases of the human telomeric G-quadruplex DNA, stacking against which may restrain it to a planar conformation, consistent with increased fluorescence (Mohanty et al., 2013).

The presence of two G-quadruplexes in the structure of Corn-DFHO, and the likely preorganization of its fluorophore binding site, led us to examine whether this aptamer RNA would bind ThT. We found that Corn binds ThT with K_d ~ 3 μM, and despite the difference in structure between DFHO and ThT, strongly (> 10,000 fold) enhances fluorescence of the latter (Warner et al., 2017). In order to elucidate how Corn binds ThT and activates its fluorescence, and to provide a structural framework for understanding how other G-quadruplex nucleic acids preferentially turn on the fluorescence of ThT, we have now determined the crystal structure of the Corn-ThT complex at 2.9 Å resolution. The structure reveals two molecules of ThT occupying the interprotomer space, and the presence of alternate conformations in the crystal structure is consistent with sequence-independent binding to the planar binding site. We also determined crystal structures of Corn bound to thiazole orange {Figure 1A, TO, 1-methyl-4-[(3-methyl-2(3H)-benzothiozolydiene)methyl]quinolinium, 3}, a second benzothiazole derivative that exhibits preferential binding to G-quadruplex nucleic acids, and of the unliganded Corn dimer, at 2.9 A and 2.8 Å resolution, respectively. The former shows one molecule of TO bound at the Corn homodimer interface in two alternate conformations. The latter reveals a collapse of the ligand binding site, such that the G-quadruplexes from the two protomers directly stack on each other. The dimer must sample open and closed conformations in solution, and crystallization has likely selected the latter. In contrast to the locally asymmetric Corn-DFHO complex, the complexes of the dimeric aptamer with ThT and TO, as well as the unliganded aptamer are strictly symmetric. Thus, binding to the asymmetric ligand DFHO induces local symmetry breaking. Because ThT and TO bind DNA and RNA with limited selectivity and therefore produce high background, their complexes with Corn are not suitable for in vivo imaging. However, these complexes are 3 to 5 times brighter than the cognate Corn-DFHO complex, and thus would be useful for applications where brightness rather than selectivity is required.

RESULTS

ThT Activates Corn Fluorescence More Strongly than the Cognate Ligand

At the same concentration of Corn aptamer (0.5 μM), and in the presence of excess fluorophore, the RNA-ThT complex fluoresces over three times brighter than the cognate complex between the RNA and DFHO (Figure 1B). Brightness is a function of extinction coefficient (7,250 M⁻¹ cm⁻¹ and 14,760 M⁻¹ cm⁻¹ at their absorption maxima, for DFHO and ThT, respectively) and quantum yield. To calculate the latter, we first established by CD spectroscopy that ThT does not alter the secondary structure of Corn, even at concentrations over 10 times above the apparent K_d (Figure 2A), and that its dimeric state is not altered upon binding ThT, by analytical ultracentrifugation (AUC, Figure 2B). Isothermal titration calorimetry (ITC) analysis yields saturable thermograms that are best fitted by a random two-site binding model with dissociation constants of 2 μM and 11 μM (Figure 2C). Assuming a 2:2 stoichiometry between the Corn and ThT [also consistent with previous Job plot analysis (Warner et al., 2017), and our co-crystal structure, see below], this indicates a quantum yield (Φ) for ThT bound to the RNA of 0.41, which is considerably higher than that reported previously (Song et al., 2017) for DFHO bound to Corn (Φ = 0.25).

Figure 2. Corn Aptamer Binds ThT with a 2:2 Stoichiometry.

(A) Circular dichroism for Corn RNA alone or with ThT at two different concentrations (on monomeric basis).

(B) c(S) distributions derived from velocity analytical ultracentrifugation of Corn RNA, alone and in varying RNA concentrations, in the presence of excess ThT.

(C) Representative isothermal titration calorimetry thermogram, and non-linear least-squares fit and residuals. Average of three experiments and standard error of the mean (error bars). The fit was to a model with random binding to two non-equivalent sites (s1 and s2). The estimated incompetent fraction was 30%. ΔG_s1 = − 7.9 kcal mol⁻¹, ΔH_s1= − 5.6 kcal mol⁻¹, ΔS_s1= 8.0 cal mol⁻¹ K⁻¹; ΔG_s2 = 1.3 kcal mol⁻¹, ΔH_s2= − 2.6 kcal mol⁻¹, ΔS_s2= − 13.2 cal mol⁻¹ K⁻¹. Experiments were performed at 295 K.

(D) Fluorescence lifetime of Corn-ThT. Cyan dashes, fluorescence activation of ThT (2) by Corn. Black dashes IRF, instrument response function. Yellow line, single exponential fit to the data. This fit, corrected for the IRF, gives a fluorescence lifetime of 4.35 ns, closely matching DFHO-Corn lifetime (Warner et al., 2017).

The unexpectedly high quantum yield of the Corn-bound ThT led us to examine whether this RNA aptamer can turn on the fluorescence of thiazole orange, another benzothiazole derivative for which a preferential binding to DNA triplexes and quadruplexes has been reported (Lubitz et al., 2010). Upon incubation with the aptamer, TO fluoresces brightly (Figure 1B). To calculate its quantum yield, we established by titration an apparent K_d of 1.4 μM and an RNA:TO stoichiometry of 2:1 (Figure S1A,B; the poor solubility of TO in aqueous buffers with K⁺ precluded ITC analysis). Furthermore, AUC experiments confirmed that this fluorophore does not alter the dimeric state of the Corn aptamer RNA, even at high concentrations (Figure S1C). Assuming 2:1 stoichiometry between the Corn RNA aptamer and TO, the quantum yield for the Corn-bound thiazole orange is 0.53, even higher than that of ThT. Consistent with the high quantum yields of the Corn-ThT and Corn-TO complexes, their fluorescence lifetimes are comparable to or longer than, respectively, that of the Corn-DFHO complex (Figure 2D and Figure S1D; Warner et al., 2017). These experiments demonstrate that Corn can strongly turn on the fluorescence of two chemically related non-specific fluorophores, producing quantum yields over 2-fold higher than what it achieves with the RFP-derived fluorophore DFHO against which it was selected in vitro.

ThT is Bound at the Corn Homodimer Interface in a Planar Conformation

The Corn-ThT complex yielded rhombohedral co-crystals with one Corn protomer in the crystallographic asymmetric unit (ASU) (Experimental Section, Table 1). Structure determination revealed that a crystallographic 2-fold axis bisects the ligand binding pocket of the crystallographic Corn dimer (Figure 3A,B), and residual electron density (after building and refining the RNA) suggested that ThT occupies this pocket. The electron density features were consistent with the two bound ThT molecules arranged in an antiparallel (“head-to-tail”) orientation in the same plane. At the current resolution limit (2.9 μ), the position of the methyl groups of the benzothiazole of the bound ThT is ambiguous. On steric grounds, we modeled the two heterocycles with the sulfur atoms facing each other, yielding a R_free factor of 27.1 %. Reversing the orientation of the benzothiazole rings, such that the methyl groups face each other, did not result in an improved crystallographic residual (R_free = 27.8 %). To estimate the crystallographic occupancy of the bound ThT, we performed parallel refinements in which the occupancy of each ThT was fixed at 0.25, 0.5 and 1.0, and compared the refined B-factors of the ligand atoms with those of the surrounding RNA. This analysis was consistent with an occupancy of 0.5 for each bound ThT. Therefore, in the co-crystal structure, there is one molecule of ThT per RNA protomer (two head-to tail molecules of ThT at an occupancy of 0.5 each), and because of the crystallographic 2-fold axis, the interprotomer interface is occupied by a total of two molecules of ThT. Moreover, because the crystallographic 2-fold axis does not coincide with the non-crystallographic 2-fold rotation relating the two bound ThT molecules, this results in two distinct alternate poses of the bound fluorophore molecules (Figure 3C,D). Since the two bound ThT molecules are in van der Waals contact with each other, sequential binding would result in the first and second ThT molecules encountering different structural environments, consistent with the ITC results. Consistent with the importance of having a G-quadruplex on each side of the bound ThT to achieve high fluorescence intensity, control experiments in which ThT fluorescence was induced by binding to Baby Spinach, in which the fluorophore binding site is flanked by one G-quadruplex and a base triple, resulted in reduced fluorescence intensity (Figure S2A).

Table 1.

Summary of Crystallographic Statistics

	Corn-ThT	Apo Corn	Corn(A14U)- ThT	Corn-TO
Data collection
Space group	H32	H32	H32	H32
Cell dimensions
a, b, c (Å)	131.4,131.4, 40.7	93.0, 93.0, 66.5	130.0,130.0, 40.7	128.5,128.5, 40.8
α, β, γ (°)	90, 90, 120	90, 90, 120	90, 90, 120	90, 90, 120
Resolution (Å)a	38.3-2.9 (3.6-2.9)	46.5-2.8 (3.2-2.8)	38.2-2.9 (3.0-2.9)	64.2-2.8 (3.2-2.9)
Rmeige (%)	7.7 (229)	18.4 (43.8)	11.5 (174.7)	9.2 (313.3)
CC1/2	0.98 (0.87)	0.88 (0.70)	1 (0.419)	1 (0.700)
<I>/<σ(I)>	36.4 (2.5)	11.1 (4.8)	15 (1.8)	14.77 (3.0)
Completeness (%)	97.9 (96.2)	89.2 (93.4)	100 (100)	100 (100)
Redundancy	40 (41.2)	1.7 (1.7)	18.9 (18.9)	18.8 (17.9)
Refinement
Resolution (Å)	38.2-2.9 (3.6-2.9)	46.5-2.9 (3.2-2.9)	38.2-3.1 (3.5-3.1)	38.3-2.9 (3.2-2.9)
No. reflections	3004 (1323)	4068 (1208)	4661 (1361)	5507 (1178)
Rwork / Rfree (%)	23.8 / 26.7	25.1 / 30.0	26.0 / 32.2	22.4 / 28.2
No. atoms
RNA	785	762	783	785
Fluorophore	40	N/A	40	22
Buffer/ion	1	36	1	2
Mean B-factors (Å2)
RNA	82.8	52.8	93.1	121.7
Fluorophore	83.1	N/A	55.5	74.2
Buffer/ion	124.1	87.7	79.8	128.0
R.m.s deviations
Bond lengths (Å)	0.006	0.001	0.006	0.006
Bond angles (°)	1.4	0.3	1.4	1.2
<Coordinate precision> (Å)b	0.22	0.39	0.32	0.43
PDB ID	6E81	6E80	6E82	6E84

^aValues in parentheses are for highest-resolution shell.

^bMean coordinate precision estimated by PHENIX. One crystal was used for each of the four datasets.

Figure 3. Crystal Structure of Corn in Complex with ThT.

(A) Cartoon representation of the structure of the dimeric Corn RNA recognizing two molecules of ThT. Single-letter nucleotide names are in upper and lower case for the two protomers. Shaded boxes highlight the locations of T1, T2, T3 and T4. T1 and T2 are canonical G-quartets. T3 and T4 are non-canonical. (B) 90° rotation. (C, D) Views along the G-quadruplex axis showing the two crystallographic poses of the bound ThT molecules. Green mesh depicts |F_o|-|F_c| electron density contoured at 3 s.d. above mean peak height before any ligand was built. The T1 G-quartet immediately below the bound fluorophores, and surrounding interfacial adenosines from one protomer are shown. In all panels, the location of the crystallographic 2-fold axis bisecting the Corn dimer is indicated by a line with a lune.

When considering just the structure of a protomer, the structure of the Corn RNA in complex with ThT is very similar to that in its cognate complex with DFHO (Warner et al., 2017). The RNA is a stem-loop with the loop folded back on itself, producing a quadruplex in this region. The quadruplex comprises two canonical G-quartets adjacent to the interprotomer interface (T1 and T2), and two non-canonical quartets comprised of doubled non-canonical A•U pairs and doubled G•C pairs (T3 and T4, respectively). A density feature consistent with a bound K⁺ ion is present between T1 and T2. Although the K⁺ and water molecules that coordinate T3 in the Corn-DFHO co-crystal structure are not resolved in the current Corn-ThT complex electron density, the 16 residues of the quadruplex (including T3) superimpose closely between the two structures (r.m.s.d. 0.38 Å), implying that the second K⁺ ion and inter-base waters are also present in the ThT complex.

The cognate DFHO and non-cognate ThT complexes of the Corn aptamer differ substantially in their interprotomer dimerization interface. As described previously, binding of the asymmetric DFHO in a planar conformation (which endows it with a plane of mirror symmetry) by the chiral dimer of Corn RNA necessarily results in a quasisymmetric dimer interface, in which the three interfacial adenosines, A11, A14 and A24 from the two protomers adopt distinctly different conformations (Warner et al., 2017). Thus, for instance, only one of the two A14 nucleobases contacts the oxime function of DFHO, and one of the two A11 hydrogen bonds to an imidazolinone nitrogen while the other is extruded from the binding pocket. In contrast, the crystallographic 2-fold symmetry relating the two protomers of the dimeric RNA in the ThT complex arises from the three interfacial adenosines exhibiting precisely the same conformations in the two RNA chains. In this complex, A11, A14 and A24 make no contacts with the bound fluorophores and are arranged so that the nucleobase of A11 of one protomer stacks on that of A24 from the other, and the two pairs of stacked adenosines are situated on the same side of dimerization interface. The riboses and nucleobases of A14 of the two protomers lack interactions with adjacent residues and they are extruded from dimerization interface (Figure 3A,B). This produces an opening that may allow ThT ingress and egress from its binding site.

The Fluorophore-binding Pocket Collapses in Unliganded Corn

The fluorophore-free Corn aptamer dimer also yielded rhombohedral crystals with one RNA protomer per asymmetric unit (Experimental Section, Table 1). Except at the dimerization interface, the structure of the unliganded Corn RNA superimposes closely on those of the Corn-DFHO and Corn-ThT complexes (r.m.s.d. for all non-hydrogen RNA atoms omitting the three interfacial adenines of 1.63 ± 0.14 Å and 1.33 Å, respectively), and all the pairing interactions in the G-quadruplexes are conserved with the structures of the DFHO and ThT complexes. In the unliganded structure, electron density for the axial K⁺ ions of the quadruplex is weak, but precise conservation of the location of all nucleobases suggests that the ions and waters are bound in positions equivalent to those of the DFHO complex structure.

The most notable difference between the structures of fluorophore-bound and unbound aptamers becomes apparent when the dimerization interface of Corn dimers is compared. In the crystal structure of the unliganded Corn, the fluorophore binding site has disappeared, as the T1 G-quartets of the symmetry-related Corn protomers stack directly on each other (Figure 4A). In this rearranged structure, the nucleobases of the interfacial adenosines adopt conformations similar to those in the ThT complex, but distinctly different from those in the DFHO complex (Figure 4B,C,D). The rotational orientation of the two protomers along the 4-fold axis of the G-quadruplexes is similar in the fluorophore-free and ThT complexes, but different from that of the DFHO complex. Thus, in the unliganded and ThT-bound structures, G12 of one protomer stacks under G22 of the other protomer, and vice versa. In the DFHO complex, G12 of one protomer stacks under G15 of the other protomer, but the second does not stack on the opposite quadruplex. Overall, relative to the unliganded complex, the second protomer of the DFHO-bound complex has undergone a 60° rotation along its quadruplex axis, and this axis has also shifted by 3 Å (Figure 4D).

Figure 4. Crystal Structure of the Unliganded Corn RNA Dimer.

(A) Cartoon representation of the unliganded dimeric Corn RNA. Color-code: grey-RNA backbone and nucleotides that form Watson-Crick pairs; dark grey-guanines organized in all G quartets T1 and T2; cyan- mixed quartet T3; green-mixed quartet T4; pink-extra helical and extra quartet nucleotides. Shaded boxes highlight the locations of T1, T2, T3 and T4.

(B) Axial view of the two T1 G-quartets and the interfacial adenosines from the unliganded structure.

(C) Axial view of the two T1 G-quartets and the interfacial adenosines from the ThT-bound structure. For clarity, the bound ThT has been omitted.

(D) Axial view of the two T1 G-quartets from the DFHO-bound structure. For clarity, the bound DFHO has been omitted.

The Asymmetric Non-specific Fluorophore Thiazole Orange Binds Symmetric Corn

Since the structures of unliganded and ThT-bound Corn dimers determined in this work exhibit crystallographic (strict) symmetry, whereas the previously reported structure of Corn bound to its cognate, asymmetric ligand DFHO is quasisymmetric, we examined whether the asymmetry of the latter is a consequence of the asymmetry of the ligand by solving the structure of the RNA bound to a second non-specific asymmetric ligand, TO (3). Unexpectedly, the rhombohedral crystals of this complex also have a single RNA chain in the ASU. Structure determination (Methods, Table 1) revealed, consistent with the stoichiometry observed in solution, a single molecule of TO bound at the dimer interface (Figure 5). Because the bound TO is traversed by a crystallographic 2-fold axis, the ligand exists in two distinct poses in the interface (Figure 5B,C). Each asymmetric unit contains a single molecule of TO at an occupancy of 0.5 (adding to a full molecule of TO in the binding site of the crystallographic dimer).

Figure 5. Crystal Structure of Corn in Complex with TO.

(A) Cartoon representation of the structure of the dimeric Corn RNA recognizing one molecule of TO. For clarity, single-atom nucleotide names are in upper and lower case for the two protomers. Shaded boxes highlight the planes of the four tetrads T1, T2, T3 and T4 and the bound fluorophore. (B, C) Ball-and-stick representation of the ligand binding site looking down the G-quadruplex axis showing two alternate binding poses for TO. Green mesh depicts |F_o|-|F_c| electron density contoured at 3 s.d. above mean peak height before any ligand was built. In all panels line with lune indicates location of the crystallographic 2-fold rotation axis.

The overall structure of the TO complex of Corn superimposes closely on the ThT-bound RNA complex (r.m.s.d. = 0.85 Å for all non-hydrogen RNA atom pairs). The similarity extends to the relative orientation of the two protomers, which is indistinguishable from that of the ThT complex. Notably, the arrangement of the interfacial adenines A11, A14 and A24 is also very similar between the two structures, despite the fact that the single TO fluorophore, which is bound in a near planar conformation, leaves a portion of the ligand binding site unoccupied (Figure 5B,C). The modest geometric fit between TO (monomer) and ThT (head-to-tail dimer) and the fluorophore binding pocket of Corn, compared to that of DFHO is also reflected in the shape complementarity statistic (Lawrence and Colman, 1993), which is 0.72, 0.74 and 0.83 for the TO, ThT and DFHO co-crystal structures, respectively.

Interfacial Adenines Modulate Corn Dimerization

Previously, point mutants of the interfacial adenines of Corn were evaluated for their ability to turn on DFHO fluorescence (Warner et al., 2017). In that study, it was found that all point mutants led to diminished fluorescence turn-on (ranging from none to 25% of wild-type). The least affected mutant was A11U, which preserved 25% of the brightness of the parental sequence, while A14U exhibited no detectable fluorescence. To complement that study, and in light of the differences in the arrangement of the interfacial adenines between the DFHO- and ThT-bound Corn structures, we tested the point mutants for their ability to turn on the fluorescence of ThT (Figure 6A). We find that, consistent with the lack of interaction of the interfacial mutants with the non-cognate fluorophore (Figure 3C,D) the point mutants substantially retain their ability to turn on the fluorescence of ThT (Figure 6A).

Figure 6. Interfacial Adenine Mutants Modulate ThT Fluorescence Turn-on and Dimerization.

(A) Normalized fluorescence, relative to wild-type Corn, of point mutants of the interfacial adenosines in complex with ThT.

(B) Normalized fluorescence of the Corn mutants in complex with DFHO (Warner et al., 2017) Error bars in (A) and (B) denote standard errors of the mean from triplicate experiments.

(C)CD spectra of wild-type, A14C, A14U, and (A11U, A14U, A24U) Corn RNAs.

(D) AUC of Corn(A14U).

(E) Native polyacrylamide gel electrophoretic analysis (buffer contained 20 mM KCl and 5 mM MgCl₂) of Corn point mutants in the presence of excess ThT, visualized by ethidium bromide staining. RNAs were loaded at 19 μM concentration.

(F) Ball-and-stick representations of T1, T2, interfacial adenines and G8 of ThT-bound Corn(A14U), highlighting the interaction between G13 and G8, and the location of U14 near T2. For clarity, the bound ThT has been omitted.

(G) Ball-and-stick representation of T1, T2, interfacial adenines and G8 of TO-bound wild type Corn showing interaction between A14 and G8. For clarity, the bound TO has been omitted.

The A14G and A14U mutants exhibit the largest decrease in enhancement of ThT fluorescence (Figure 6A). Comparison of the CD spectra of wild-type and the two transversion point mutations of residue 14 show that the presence of a pyrimidine at this position does not alter the overall secondary structure of the RNA. Even the (A11U, A14U, A24U) triple mutant exhibits a CD spectrum that differs little from that of wild-type (Figure 6C). Analysis of the oligomerization state of the point mutants over the concentration range 280 nM to 100 μM (monomer basis) by analytical ultracentrifugation, and at 19 μM by native polyacrylamide gel electrophoresis (Figure 6D,E; Figure S3) shows that while the A11G, A11U, A24C, A24G and A24U mutants remain predominantly dimers over this concentration range, A11C exists as a mixture of monomers and dimers, and A14U is predominantly monomeric.

We further analyzed A14U, because it is predominantly monomeric and previously found to be drastically impaired in its ability to activate DFHO fluorescence (Figure 6B) (Warner et al., 2017). We determined its crystal structure in complex with ThT (Methods and Table 1), and found that its structure is overall identical to that of the wild-type Corn-ThT complex (r.m.s.d. = 0.51 Å for all non-hydrogen RNA atom pairs, excluding residue 14), and also binds to two molecules of ThT at a crystallographic dimer interface (Figure S4). Electron density for A14 is best defined in the Corn-TO co-crystal structure. Comparison of that structure with the A14U mutant co-crystal structure highlights the loss in the point mutant of a potentially stabilizing intra-protomer interaction between the sugar edge of A14 and the Watson-Crick face of G8 (Figure 6F,G). This interaction may stabilize the TO and ThT complexes of the wild-type Corn aptamer, and thus facilitate fluorescence turn-on. Analytical ultracentrifugation of Corn A14U at 100 μM concentration (at which crystallization was carried out) shows a heterogeneous mixture of monomer, dimer, and trimer indicating that the crystallographic structure represents only one of several species coexisting in the sample. If the unliganded Corn dimer transiently samples a conformation stabilized by the interaction between A14 and G8 seen in the wild-type co-crystal structures, the mutation A14U may decrease the dimer population.

DISCUSSION

The co-crystal structures of the Corn aptamer bound to ThT and TO provide the first experimental description of the modes of nucleic acid binding of these G-quadruplex preferent fluorophores. Both ThT and TO bind in near-planar conformations that maximize their fluorescence, and stack extensively with the exposed guanine bases of the G-quartets. Our structures show that, perpendicular to the planes of their heterocycles, the fluorophores make no hydrogen bonding interactions with the surrounding RNA, and only minimal van der Waals contacts (Figures 3,5). This is consistent with the limited hydrogen bonding potential of the fluorophores, and their reported low sequence specificity. The co-crystal structures suggest that preferential fluorescence turn-on upon binding to G-quadruplexes over duplexes or triplexes by these benzothiazole derivatives reflects both, the larger solvent-exposed non-polar area at the unstacked end of a G-quadruplex, as well as the increased planarity of G-quartets compared to base pairs or base triples, which often exhibit propeller twisting or buckling. The planar, RNA backbone-independent mode of RNA Corn binding observed for ThT and TO contrasts with that reported previously for a variety of G-quadruplex-targeted small molecules as well as the G-quadruplex-specific protein DHX36 (Chen et al., 2018; Haider et al., 2011). Those ligands interact with both, the exposed guanine nucleobases and the phosphodiester backbone. Interaction with the backbone endows some of those small molecules and the helicase protein with a preference for G-quadruplexes with a particular connectivity (e.g., parallel vs. antiparallel). The backbone-independent binding of ThT and TO suggests that the variation in their fluorescence enhancement observed for different G-quadruplexes (Gabelica et al., 2013; Li et al., 2016; Mohanty et al., 2013; Renaud de la Faverie et al., 2014; Warner et al., 2017; Xu et al., 2016) reflects instead the efficiency of DNA or RNA folding, the planarity of the bound fluorophores in the photoexcited state, and the stoichiometry of binding.

The previously determined Corn-DFHO co-crystal structure revealed that the ligand-binding site of the RNA is locally quasisymmetric (Jones and Ferré-D'Amaré, 2015) with the interfacial adenosines of each protomer participating in distinctly different RNA-fluorophore and RNA-RNA interactions (Warner et al., 2017). In contrast, the Corn-ThT and Corn-TO co-crystal structures show strictly symmetric Corn dimers, suggesting that the symmetry of the RNA binding site is broken only upon binding to DFHO. Consistent with induction of asymmetry by the cognate fluorophore DFHO, the crystal structure of the ligand-free dimeric Corn RNA is also strictly symmetric (Figure 4). In this crystal structure, the ligand binding site has collapsed, the G-quadruplexes from the two protomers stacking directly against each other. Since biochemical analyses show that Corn RNA dimers are pre-formed, it is likely that crystallization has captured a closed conformation, and that open conformations that are conducive to binding, either symmetrically to ThT and TO, or asymmetrically to DFHO, exist in solution. In this regard, it is noteworthy that the distribution of unbound Corn RNA between monomers and dimers is sensitive to mutation of the interfacial adenosines. The co-crystal structure of the A14U mutant in complex with ThT shows that the mutation disrupts an intramolecular RNA interaction that may stabilize the wild-type RNA. Thus, it is possible that A11, A14 and A24, in addition to playing a role in RNA dimerization and ligand binding, also contribute to the stability of the aptamer and, indirectly, to its oligomerization. It has been noted that mutants of Corn that are obligate heterodimers could form the basis of RNA analogues of split GFP (Warner et al., 2017). Our experiments suggest that mutation of the interfacial adenosines, as well as distal residues with which they may interact in the fluorophore-free or bound states may modulate the tendency of the RNA to homodimerize.

Although the structural basis of preferential binding and fluorescence activation of ThT by protein amyloids remains unknown, several crystal structures of ThT bound to globular proteins have been reported (Figure S5). A dimeric β2-microglobulin, proposed to be an early aggregate precursor to amyloid, was found to bind either one or four molecules of ThT at an interface between two β-barrel domains (Halabelian et al., 2015). The mixed αβ proteins acetylcholynesterase and butyrylcholinesterase were found to bind either one or two molecules of ThT in a cavity formed at the juncture of several α-helices (Harel et al., 2008; Rosenberry et al., 2017). In all these cases, the ThT binding sites are lined by aromatic amino acid side chains. In contrast to the mode of ThT binding to G-quadruplexes that we have uncovered, these protein binding sites do not accommodate more than one coplanar ThT molecule. Indeed, when multiple ThT molecules are bound, they stack on each other (Halabelian et al., 2015; Harel et al., 2008). Moreover, the protein-bound ThT molecules show variable degrees of rotation between their benzothiazole and aniline rings, unlike the near-planar conformation observed in Corn bound ThT.

Among the fluorescence turn-on aptamers structurally characterized to date, Corn is unique in that the active RNA species is a dimer. Unlike Spinach and Mango, whose fluorophore binding sites constrain their respective cognate small molecule by sandwiching between a G-quadruplex and either a base triple or unpaired nucleotide 'flaps', respectively, the Corn dimer binding site is flanked on either side by a G-quartet, one from each protomer (Huang et al., 2014; Trachman et ThT. al., 2018; Trachman et al., 2017; Warner et al., 2014; Warner et al., 2017). Comparison of the fluorescence turn-on of ThT by Spinach and Corn (Figure S2) shows that the binding site flanked by two G-quadruplexes is better able to constrain the photoexcited fluorophore. Corn RNA was selected to bind DFHO and turn on its fluorescence. Despite its limited brightness, DFHO is attractive for cellular studies because its non-specific turn-on by RNAs other than Corn is minimal. In contrast, ThT and TO, which yield much brighter fluorescence when bound to Corn also bind and are turned on by a variety of nucleic acids (as well as proteins; Figure S6). While this limits the applications of Corn-ThT and Corn-TO in cellular studies, the high quantum yields of these complexes and their intense fluorescence makes them useful in contexts where selective binding is not required. The Corn-ThT co-crystal structure also shows that a G-quartet has sufficient non-polar area to accommodate two molecules of ThT, and can thus form the basis for the design of new, larger fluorophores that better occupy the non-polar exposed face of G-quadruplex nucleic acids.

STAR*METHODS

METHOD DETAILS

Chemicals

DFHO (1) RNA 1 (Key Resources Table) was prepared as previously described (Song et al., 2017) ThT (2) (Key Resources Table) and TO (3) (Key Resources Table) were dissolved freshly for each experiment in 100 mM KCl, 10 mM MgCl₂, 50 mM Tris-HCl pH 7.5, or where indicated, in 100 mM KCl, 10 mM MgCl₂, 50 mM Hepes-KOH pH 7.5, and filtered (0.22 μm cutoff). Concentrations were calculated from the dry weight of the compounds.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals
DFHO	Chemical synthesis Ref1
Thioflavin T	Abcam	120751
Thiazole Orange	Biotium	40077
Thiazole Orange	Sigma Aldrich	390062
Rhodamine 123	Enzo Life Sciences	52307
RNA Constructs
RNA 1(Corn wild type) GGCGCGAGGAAGGAGGUCUGAGGAGGUCACUGCGCC	In vitrotranscription	N/A
RNA2 (Corn A11C) GGCGCGAGGACGGAGGUCUGAGGAGGUCACUGCGCC	In vitrotranscription	N/A
RNA3 (Corn A11G) GGCGCGAGGAGGGAGGUCUGAGGAGGUCACUGCGCC	In vitrotranscription	N/A
RNA4 (Corn A11U) GGCGCGAGGAUGGAGGUCUGAGGAGGUCACUGCGCC	In vitrotranscription	N/A
RNA 5 (Corn A14C) GGCGCGAGGAAGGCGGUCUGAGGAGGUCACUGCGCC	In vitrotranscription	N/A
RNA 6 (Corn A14G) GGCGCGAGGAAGGGGGUCUGAGGAGGUCACUGCGCC	In vitrotranscription	N/A
RNA 7 (Corn A14U) GGCGCGAGGAAGGUGGUCUGAGGAGGUCACUGCGCC	In vitro transcription	N/A
RNA 8 (Corn A24C) GGCGCGAGGAAGGAGGUCUGAGGCGGUCACUGCGCC	In vitrotranscription	N/A
RNA 9 (Corn A24G) GGCGCGAGGAAGGAGGUCUGAGGGGGUCACUGCGCC	In vitrotranscription	N/A
RNA 10 (Corn A24U) GGCGCGAGGAAGGAGGUCUGAGGUGGUCACUGCGCC	In vitrotranscription	N/A
RNA 11 (Corn A11U, A14U, A24U) GGCGCGAGGAUGGUGGUCUGAGGUGGUCACUGCGCC	In vitrotranscription	N/A
RNA 12 (Baby Spinach) GGUGAAGGACGGGUCCAGUAGUUCGCUACUGUUGAGUAGAGUGUGAGCUCC	Dharmacon	N/A
DNA competitor (T7 RNA polymerase promoter) GAATTCTAATACGACTCACTATAG	IDT	N/A
Homo sapiens tRNA Lys3	In vitro transcription	N/A
Deposited Structures
Crystal Structure of unliganded Corn	PDB	6E80
Crystal Structure of Corn bound to Thioflavin T	PDB	6E81
Crystal Structure of Corn A14U bound to Thioflavin T	PDB	6E82
Crystal Structure of Corn bound to Thiazole Orange	PDB	6E84
PDB used for comparison
Crystal Structure of Corn bound to DFHO	PDB	5BJP Warner et al., 2017
Software and Algorithms
NITPIC	Keller et al., 2012	http://biophysics.swmed.edu/MBR/software.html
SEDPHAT	Zhao et al., 2002	http://www.analyticalultracentrifugation.com/sedphat/
HKL2000	Otwinowski and Minor, 1997	http://www.hkl-xray.com/
XDS	Kabsch, 2010	http://xds.mpimf-heidelberg.mpg.de/
DIALS	Winter et al., 2018	https://dials.github.io/
Phenix	Adams et al., 2010	https://www.phenix-online.org/
Coot	Emsley and Cowtan, 2004	https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/
Erraser	Chou et al., 2013	https://www.phenix-online.org/documentation/reference/erraser.html
PyMol	DeLano, 2002	https://pymol.org/2/

RNA Preparation

The 36 nt Corn construct previously used for crystallization (Warner et al., 2017) (RNA 1, Key Resources Table), point mutants (A11C; A11G; A11U; A14C; A14G; A14U; A24C; A24G; A24U- respectively: RNA 2, RNA 3, RNA 4, RNA 5, RNA 6, RNA 7, RNA 8, RNA 9, RNA 10, Key Resources Table), and the triple mutant (A11U, A14U, A24U- RNA 11, Key Resources Table) were prepared by in vitro transcription. DNA templates generated by PCR were transcribed in vitro by T7 RNA polymerase as described (Xiao et al., 2008). After 4h, 10 mM CaCl₂ and RNase-free DNase RQ1 (Promega) were added and the reaction incubated for 30 min. RNAs were purified by electrophoresis (15% 29:1 acrylamide:polyacrylamide, 1 × TBE, 8 M urea gels), electroeluted, and washed with 1 M LiCl and water and finally exchanged into 100 mM KCl, 50 mM Tris-HCl pH 7.5 using 3 kDa cutoff centrifugal concentrators (EMD Millipore). For refolding, RNAs were heated for 2 min at 95 °C, placed 2 min on ice, then incubated at 65 °C for 5 min after addition of MgCl₂ to 10 mM, and cooled from 65 °C to 25 °C in 15 min. Refolded RNAs were stored at concentration 50 μM (monomer basis) at 4 °C and diluted or concentrated by ultrafiltration as appropriate for different experiments. RNAs were quantitated assuming an extinction coefficient of 354,600 M⁻¹ cm⁻¹ at 260 nm.

Fluorescence Spectroscopy

Fluorescence spectra were recorded with an Easy Life-LS PTI fluorimeter at 20°C. Refolded RNA (0.5 μM, monomer basis) in 100 mM KCl, 10 mM MgCl₂, 50 mM Tris-HCl pH 7.5, were incubated with ThT at concentrations ranging from 0.25 μM to 1 mM for 75 min, or for different times (0 – 96 h) at a fixed ThT concentration of 2.5 μM, at 21 °C. Excitation was at 450 nm. For competition of wild type Corn (0.5 μM) for DFHO, ThT and TO (each at 2.5 μM) fluorophores were added two at a time to the folded RNA and incubated for 30 min at 21 ° C. Excitation was at 468 nm, 450 nm, and 510 nm for DFHO, ThT, and TO, respectively. For evaluating the effect of molecular crowding and viscosity, Corn (0.5 μM) was mixed with PEG400 (0 – 15 %, v/v) or glycerol (0 – 30%, v/v) and incubated 10 min with fluorophores (2.5 μM) before fluorescence measurements. Corn affinity for TO was determined by varying the RNA concentration between 125 nM and 8.0 μM while keeping the TO concentration fixed at 2.5 μM. Excitation was at 510 nm and emission was monitored over 515-550 nm. For Corn-TO stoichiometry determination using the Job (Job, 1928) plot, the RNA and the fluorophore were incubated together in 50 mM Hepes-KOH, 100 mM KCl, 10 mM MgCl₂, pH 7.5, at 20°C. The total concentration of the two was fixed at 30 μM, while each was varied in 2.5 μM steps over the 0-30 μM range. Excitation was at 510 nm and emission was monitored over 515 – 550 nm. For evaluating specificity of Corn and ThT interaction in presence of other biological macromolecules, ThT was incubated with different amounts of recombinant bovine serum albumin (0.5-50 μM), a 24 nt single-stranded DNA oligonucleotide containing bacteriophage T7 RNA polymerase promoter (0.5-50 μM), or recombinant tRNA^Lys3 (0.5-10 μM) without any Corn or in presence of 0.5 μM Corn; fluorescence was measured after 30 min incubation at 21 ° C.

Fluorescence life time measurements were performed at 21 ° C; integration time was 1 s nm⁻¹ and each measurement was averaged seven times. A 10-mm path length quartz cuvette was used. ThT-Corn lifetime measurement was performed on EasyLife-LS PTI system equipped with an EL445 diode (Horiba), 445/20-25 and 494/34-25 filter sets (AVR Optics); 7 μM Corn was incubated 10 min with 5 μM ThT. TO-Corn lifetime was measured on same instrument equipped with EL510 diode and 535/6 filter; 30 μM Corn was incubated 10 min with 20 μM TO. The instrument response function (IRF) was determined using Ludox (Sigma). The data shown are average of three independent experiments. Data were fitted to one, two and three exponentials using software provided by the manufacturer.

Quantum Yield Determination

To determine quantum yields of Corn complexes, integrals of their emission spectra were compared with that of the reference fluorophore rhodamine 123 (Key Resources Table) which has a quantum yield of 0.9 (Kubin and Fletcher, 1982). Samples were freshly prepared in 40 mM HEPES-KOH, 100 mM KCl, 5 mM MgCl₂. 100 μM Corn was mixed with 10 μM fluorophore (ThT or TO); rhodamine concentration was 10 μM.

Circular Dichroism (CD)

Measurements were performed using a JASCO J715 spectrometer connected to a Peltier temperature controller. Spectra were acquired over the wavelength range 200-340 nm using a 1 mm path length quartz cuvette. The scanning speed was 100 nm min⁻¹ and response time was 1 s. Fluorophores, ThT at 5-50 μM, DFHO at 10 μM, were added to prefolded RNA (10 μM) in 150 mM KCl, 10 mM MgCl₂, 1 mM Tris-HCl pH 7.5, or in 10 mM KCl, 1 mM MgCl₂, 1 mM Tris-HCl pH 7.5. Thermal unfolding was performed between 25 – 95 °C, at 1 °C min⁻¹, and signal was monitored at 269 nm. Heated samples were then refolded by cooling at 1 °C min⁻¹ back to 25 °C.

Analytical Ultracentrifugation (AUC)

RNAs were in 100 mM KCl, 10 mM MgCl₂, 50 mM Tris-HCl pH 7.5 (AUC buffer) at 2.8 μM (OD ~1.0), 1.4 μM (OD ~ 0.5) and 0.28 μM (OD ~ 0.1). Absorbance was measured at 260 nm at 20 °C; run speeds were 60,000 RPM. For the Corn complexes with DFHO, ThT and TO complexes, RNA at the same concentrations was mixed with either 5 μM DFHO or 30 μM ThT and TO, and absorbance was measured at 260 for DFHO and TO, or 280 nm for ThT. The reference cell contained buffer with no fluorophore. Data were analyzed using SEDFIT (Schuck, 2000) (Key Resources Table), assuming a partial specific volume of 0.53 cm³ g⁻¹ and hydration value of 0.59 (Ramesh et al., 2011).

Isothermal Titration Calorimetry (ITC)

Experiments were performed in a Microcal 200 calorimeter (GE Life Sciences) with samples in 100 mM KCl, 10 mM MgCl₂, 50 mM HEPES-KOH pH 7.5. ThT (1.5 mM) was injected in 25 steps (first injection 0.4 μl followed by 24 1.5 μl injections), with 3 min interval between steps. The cell contained 95 μM refolded Corn aptamer (monomer basis) (RNA 1, Key Resources Table). Thermograms were integrated using NITPIC (Keller et al., 2012) (Key Resources Table) and the resulting titration curves were fitted globally using SEDPHAT (Zhao et al., 2015) (Key Resources Table) to estimate stoichiometry and affinity. Experiments were in triplicate. Control experiments consisted of the same number of injections of 1.5 mM ThT in 100 mM KCl, 10 mM MgCl₂, 50 mM HEPES-KOH pH 7.5 into the same buffer. ITC could not be employed for analysis of TO - Corn interaction because the poor solubility of TO.

Van't Hoff Analysis

Fluorescence of Corn (in 0.125 - 8.0 mM) in the presence of TO (2.5 mM) was monitored at six different temperatures: 10, 15, 20, 25, 30, and 37 °C, and K_d'^S were derived for each temperature. The logarithm of the K_d'^S were plotted against reciprocal absolute temperature. ΔH and ΔS were determined from linear least-squares fit. The logarithms of K_d'^S at 10 °C, 15 °C, 20 °C, 25 °C, 30 °C and 37 °C from fluorescence titrations (0.73 ± 0.18 μM, 1.32± 0.13 μM, 1.36 ± 0.01 μM, 0.71±0.07 μM, 0.6 ± 0.01 μM, and 0.61 ± 0.01 μM, means of three replicates ± s.d., respectively), were plotted against the reciprocal absolute temperature. Scatter in the data, possibly a result of severely limited solubility of TO (that precluded ITC analysis) led to poor regression, indicative of an unreliable fit (r = 0.31). The derived thermodynamic parameters are: ΔG = 0.07 kcal mol⁻¹; ΔH = −3.9 kcal mol⁻¹, ΔS =−13.6 cal mol⁻¹K⁻¹.

Crystallization and Diffraction Data Collection

Fluorophore-free Corn (RNA 1, Key Resources Table) was crystallized by vapor diffusion. Hanging drops prepared by mixing 200 nl of refolded Corn (100 μM, monomer basis) and 100 nl of a reservoir solution comprised of 20% (w/v) PEG 4000, 0.2 M ammonium acetate pH 6.7, 0.1 M sodium citrate pH 5.6, were equilibrated against 100 μl of reservoir at 21 °C. Rectangular plate-shaped crystals grew over two days to maximum dimensions of 100 × 50 × 20 μm³. 2 μl of 1 mM iridium (III) hexamine chloride dissolved in reservoir solution supplemented with 20% (v/v) PEG 400 were added to the drop. After 45 min, crystals were transferred to reservoir solution supplemented with 20% (v/v) PEG 400, immediately mounted in nylon loops and flash-frozen by plunging into liquid nitrogen. Co-crystals of Corn-ThT, Corn(A14U)-ThT, and Corn-TO were grown in the same manner, using reservoir solutions comprised of 20% (w/v) PEG 4000, 5% (v/v) PEG 400, 0.2 M ammonium acetate pH 6.7, 0.1 M sodium citrate pH 5.6, and 0.5 mM ThT, or 20% (w/v) PEG 4000, 5% (v/v) PEG 400, 10% (v/v) glycerol, 0.2 M ammonium acetate pH 6.7, 0.1 M sodium citrate pH 5.6, and 0.5 mM TO. ThT and TO co-crystals were grown at 15 °C and 21 °C, respectively. These crystals, which grew to similar dimensions as those of fluorophore-free Corn, were briefly soaked in the reservoir solution used for fluorophore-free Corn crystallization supplemented with 20% (v/v) PEG 400, mounted in nylon loops and flash-frozen by plunging into liquid nitrogen. Diffraction data were collected at 100 K in rotation mode at beamlines 5.0.1 and 5.0.2 of the Advanced Light Source, Lawrence Berkeley National Laboratory (ALS) using 1.0 Å or 1.1 Å X-radiation and reduced with HKL2000 (Otwinowski and Minor, 1997) (fluorophore-free Corn), XDS (Kabsch, 2010) (Corn- ThT), or DIALS (Winter et al., 2018) (Corn(A14U)-ThT and Corn-TO) (Key Resources Table).

Structure Determination

The Corn-ThT co-crystal structure was determined by molecular replacement (McCoy et al., 2007) using chain A of the Corn-DFHO crystal structure (PDB ID:5BJP) as the search model. The top solution had TFZ and LLG scores of 15.4 and 309, respectively. Iterative rounds of simulated annealing, energy minimization and individual isotropic B-factor refinement (Adams et al., 2010) interspersed with manual model building (Emsley and Cowtan, 2004), produced a model with R_free ~ 0.27. At this stage, two molecules of ThT, each with an occupancy of 0.5 were added to the model, RNA geometry was optimized (Chou et al., 2013), and further refinement carried out to yield the current model. The structure of fluorophore-free Corn was determined by molecular replacement using a search model comprised of the four tetrads and residue U19 (a total of 17 nt) of the Corn-ThT structure. The best solution had TFZ and LLG scores of 8.3 and 101, respectively. The structure of the Corn(A14U)-ThT complex was determined by molecular replacement using the Corn-ThT structure (chain A) after omitting A14 and ThT. The top solution had TFZ and LLG scores of 17.4 and 501, respectively. The structure of the Corn-TO complex was determined by molecular replacement using the Corn-ThT structure (chain A) from which ThT was omitted. The top solution has TFZ and LLG scores of 17.7 and 570, respectively. The fluorophore-free Corn, Corn(A14U)-ThT, and Corn-TO structures were refined as the Corn-ThT structure. Metal ions were identified based on electron density feature shape, B-factor analysis, and coordination geometry. Structural figures were generated with PyMol (DeLano, 2002) (Key Resources Table).

Fluorescence Lifetime Measurements

Measurements were performed at 20 °C on an EasyLife- LS (Photon Technology International) system equipped with an EL445 diode (Horiba), 445/20-25 and 494/34-25 filters (AVR Optics) for ThT, or an EL510 diode (Horiba) and 535/6 filters for TO. Integration time was 1 s per point and each measurement was averaged 7 times. Corn (7 μM) (RNA 1, Key Resources Table) was refolded and then incubated with 5 μM fluorophore for 1 h at room temperature. The instrument response function (IRF) was determined using Ludox (Sigma). The data shown are the average of three independent experiments. Data were fitted to one, two and three exponentials using Origin 2017 (OriginLab).

Native Polyacrylamide Gel Electrophoresis

Native running buffer was comprised of 16.5 mM Tris, 33 mM HEPES, 0.05 mM EDTA, 20 mM KCl, 5 mM MgCl₂. Samples (RNA 1-11, Key Resources Table) were prepared mixing 15 μL of 25 μM RNA and 5 μL of native loading buffer (50% glycerol, 5 × native running buffer) and loaded on a 15% 19:1 acrylamide:bisacrylamide gel (16 cm wide, 20 cm long, 0.05 cm thick) for 3 hr at 20 W, at 4 °C; native running buffer was exchanged with fresh one every 1.5 hour. For analysis of samples that contained ThT, running buffer was supplemented with 250 μM ThT.

QUANTIFICATION and STATISTICAL ANALYSIS

DFHO (1), ThT (2), TO (3) (Key Resources Table) were dissolved freshly for each experiment in 100 mM KCl, 10 mM MgCl₂, 50 mM Tris-HCl pH 7.5, or where indicated, in 100 mM KCl, 10 mM MgCl₂, 50 mM Hepes-KOH pH 7.5, and filtered (0.22 μm cutoff). Concentrations were calculated from the dry weight of the compounds. Corn RNAs were quantitated assuming an extinction coefficient of 354,600 M⁻¹ cm⁻¹ at 260 nm. Recombinant tRNA^Lys3 produced by T7 polymerase in vitro transcription was quantitated assuming an extinction coefficient 604,000 M⁻¹ cm⁻¹ at 260 nm. Known amount of synthetic T7 RNA polymerase promoter DNA (IDT) was dissolved in 100 mM KCl, 10 mM MgCl₂, 50 mM Tris-HCl pH 7.5 to give 100 μM stock solution. Known amount of BSA was diluted in 100 mM KCl, 10 mM MgCl₂, 50 mM Tris-HCl pH 7.5 to give 100 μM stock solution assuming an extinction coefficient of 43,824 M⁻¹ cm⁻¹at 280 nm. All fluorescence-based experiments were performed in triplicate of which reported values are the mean with standard deviation; triplicate measurements for each experiment were performed during the same session without switching off the xenon lamp of the fluorimeter.

DATA and SOFTWARE AVAILABILITY

The atomic coordinates od Corn (PDB ID 6E80), Corn-ThT (PDB ID 6E81), Corn(A14U)-ThT (PDB ID 6E82), Corn-TO (6E84) are deposited at Worldwide Protein Data Bank.

CONTACT and REAGENT RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact: Adrian R. Ferré-D’Amaré

EXPERIMENTAL MODEL and SUBJECT DETAILS:

No experimental models and subjects were used in this study.

SIGNIFICANCE.

Because of their solubility, biocompatibility and good cell permeability, ThT and its derivatives are intensively studied as fluorescent probes for amyloid and G-quadruplexes, and as therapeutics for different neurodegenerative diseases. The structural basis for ThT binding to biological macromolecules and cellular substructures are not well understood. Hence, for imaging, diagnostic, and therapeutic applications of benzothiazole derivatives it is important to evaluate how they may interact with proteins, nucleic acids and physiological or pathological protein-nucleic acids complexes, both in vitro and in vivo. Our co-crystal structures of Corn bound to ThT and TO reveal that these dyes bind between G-quartets of the two adjacent G-quadruplexes where they interact exclusively with the exposed guanine bases, and make no interactions with the nucleic acid backbone. Our crystallographic analysis reveals no evidence of other structural modes of association with G-quadruplexes, such as binding in the grooves or by intercalation between the quartets of a quadruplex. Thus, our results suggest that these benzothiazole dyes will bind exclusively to the exposed guanine bases of other, monomeric G-quadruplexes, and this mode of binding mode gives rise to fluorescence turn-on.

HIGHLIGHTS.

• Two molecules of ThT bind Corn in a planar conformation that maximizes fluorescence.

• ThT binds in the dimerization interface of the Corn RNA aptamer.

• ThT makes no direct contact with Corn RNA backbone atoms.

• ThT and thiazole orange bind only to the exposed guanine bases of G-quadruplexes.