A new macrocyclic terbium(III) complex for use in RNA footprinting experiments

Abstract

Reaction of terbium triflate with a heptadentate ligand derivative of cyclen, L1 = 2-[7-ethyl-4,10-bis(isopropylcarbamoylmethyl)-1,4,7,10-tetraazacyclododec-1-yl]-N-isopropylacetamide, produced a new synthetic ribonuclease, [Tb(L1)(OTf)(OH₂)](OTf)₂·MeCN (C1). X-ray crystal structure analysis indicates that the terbium(III) centre in C1 is 9-coordinate, with a capped square-antiprism geometry. Whilst the terbium(III) center is tightly bound by the L1 ligand, two of the coordination sites are occupied by labile water and triflate ligands. In water, the triflate ligand is likely to be displaced, forming [Tb(L1)(OH₂)₂]³⁺, which is able to effectively promote RNA cleavage. This complex greatly accelerates the rate of intramolecular transesterification of an activated model RNA phosphodiester, uridine-3′-p-nitrophenylphosphate (UpNP), with k_obs = 5.5(1) × 10^-2 s^-1 at 21°C and pH 7.5, corresponding to an apparent second-order rate constant of 277(5) M^-1s^-1. By contrast, the analogous complex of an octadentate derivative of cyclen featuring only a single labile coordination site, [Tb(L2)(OH₂)](OTf)₃ (C2), where L2 = 2-[4,7,10-tris(isopropylcarbamoylmethyl)-1,4,7,10-tetraazacyclododec-1-yl]-N-isopropyl-acetamide, is inactive. [Tb(L1)(OH₂)₂]³⁺ is also capable of hydrolyzing short transcripts of the HIV-1 transactivation response (TAR) element, HIV-1 dimerization initiation site (DIS) and ribosomal A-site, as well as formyl methionine transfer RNA (tRNA^fMet), albeit at a considerably slower rate than UpNP transesterification (k_obs = 2.78(8) × 10^-5 M^-1s^-1 for TAR cleavage at 37°C, pH 6.5, corresponding to an apparent second-order rate constant of 0.56(2) M^-1s^-1). Cleavage is concentrated at the single-stranded “bulge” regions of these RNA motifs. Exploiting this selectivity, [Tb(L1)(OH₂)₂³⁺ was successfully employed in footprinting experiments, in which binding of the Tat peptide and neomycin B to the bulge region of the TAR stem-loop was confirmed.

Introduction

Phosphate ester bonds are an integral component of many important biological molecules, most notably DNA and RNA, and many fundamental biological processes are underpinned by reactions involving their formation or cleavage, e.g., energy storage and transfer, and cellular signaling and regulation.¹ Nature uses a complex series of enzymes (protein-based as well as ribozymes) to mediate phosphate ester condensation and hydrolysis reactions.² Many of these enzymes contain metal centers that are crucial to their structural integrity, or which play pivotal mechanistic roles, e.g., assisting with substrate binding and activation, provision of highly reactive metal-bound nucleophiles, and stabilization of transitions states and leaving groups.^2-8 By way of example, the group I intron ribozyme features multiple magnesium centers that help to neutralize the negative charge of the phosphodiester backbone and promote structure stabilization, but which also serve to coordinate the phosphodiester backbone of the exon RNA strand and activate it towards scission.^3,9

Inspired by the occurrence of metal-containing nucleases, ribonucleases, phosphatases and ribozymes in nature, many research groups have sought to develop low-molecular weight metal complexes that are able to cleave biologically-important phosphate esters.^10-19 This research has been further stimulated by the burgeoning growth in gene sequencing and genomics, and the realization that small, hydrolytically-active metal complexes (and their conjugates with various targeting agents) may potentially find utility as robust, versatile replacements for restriction enzymes in molecular biology research, and as nucleic acid-targeting therapeutics. A number of metal complexes have already found application as so-called “conformational probes” or “footprinting reagents”, used to map the structure of nucleic acids and their adducts with ligands, proteins or other nucleic acids in solution.^20-22

Complexes of lanthanide ions have proved to be amongst the most effective synthetic hydrolases reported to date. The hard Lewis acidity of the trivalent and tetravalent lanthanide ions,^23,24 together with their high coordination numbers, fast ligand exchange rates, and absence of accessible redox chemistry, make them ideally suited to mimic the activity of the alkaline earth metal ions, Ca²⁺ and Mg²⁺, employed by many hydrolases to bind and activate phosphate esters for cleavage. The “free” (hydrated) lanthanide ions themselves are capable of hydrolyzing a wide range of phosphate esters, including the sugar-phosphate backbone of DNA,^25-27 but are toxic and have a tendency to precipitate out of solution as hydroxides under mildly basic conditions (pH ≥ 9). Consequently, the last two decades have seen a wide variety of multidentate ligands developed to encapsulate lanthanide(III) ions in such a way as to tune and target their reactivity for biochemical applications.^28-30 Importantly, work by Hüsken and co-workers has shown that by conjugating various lanthanide complexes to an antisense oligonucleotide they were able to develop an antisense gene shear.^31,32

Amongst the various supporting ligands that have been reported, those based on the cyclen macrocycle (1,4,7,10-tetrazacyclododecane) have featured prominently, due in large part to their ease of synthesis. The majority of these ligands are tetra-substituted derivatives featuring carboxylate, amide or alcohol pendants,^33-35 which present a total of eight coordination donors, four provided by the macrocylic amines and four by the pendant groups. Whilst these ligands can be viewed as ideal as they form complexes which are thermodynamically stable and which exhibit greater kinetic inertness, only complexes of certain larger-sized lanthanide ions (featuring higher coordination numbers) are active as phosphate ester cleavage agents. This is because insufficient sites are left available around the smaller-sized lanthanides to allow for substrate binding and simultaneous provision of a metal-bound nucleophile.

With the above in mind, we have designed a new cyclen ligand derivative, L1 (2-[7-ethyl-4,10-bis(isopropylcarbamoylmethyl)-1,4,7,10-tetraazacyclododec-1-yl]-N-isopropylacetamide), with the specific objective of producing a highly active, lanthanide-based synthetic phosphate ester hydrolase. L1 is distinguished by its three, as opposed to four, amide pendants, thus “freeing up” an additional coordination site to participate in hydrolytic reactions. Another notable feature is the presence of isopropyl groups on the amide pendants. Previous work conducted with similar tetra-substituted cyclen ligands has demonstrated that, upon lanthanide complexation, such alkyl groups produce a hydrophobic cavity that reduces the pK_a of inner-sphere coordinated water ligands to values as low as 7.^19,36 Given that metal-bound hydroxo ligands are superior nucleophiles/bases to metal-bound waters, it was envisaged that L1 might therefore afford lanthanide complexes exhibiting improved hydrolytic activity under near physiological conditions (pH ~ 7.4). Herein, we report the synthesis of the new ligand L1, and the crystal structure of its terbium(III) complex, [Tb(L1)(OTf)(OH₂)](OTf)₂·MeCN (C1) (Figure 1), and examine the utility of C1 as a synthetic ribonuclease (the triflate ligand readily dissociates in aqueous solution, forming [Tb(L1)(OH₂)₂]³⁺. Initially, we investigated the ability of the complex to accelerate cleavage of the activated phosphodiester model compound, bis(p-nitrophenyl)phosphate (BNPP), and the RNA model, uridine-3′-p-nitrophenylphosphate (UpNP). For comparative purposes, we also prepared an analogue of C1 featuring an additional amide pendant (C2), but found it to be unreactive towards these substrates. The ability of [Tb(L1)(OH₂)₂]³⁺ to promote cleavage of unactivated substrates was then probed using a series of short RNA olionucleotides—a construct of the HIV-1 transactivation response (TAR) element, a construct of the HIV-1 dimerization initiation site (DIS), and a model of the prokaryotic ribosomal A-site (A-Site), as well as the more complex, folded structure of formyl methionine transfer RNA (tRNA^fMet) (Figure 2). We found that [Tb(L1)(OH₂)₂]³⁺ preferentially cleaves the exposed regions of these RNA motifs. This ultimately led us to explore the application of this complex as a new footprinting agent, through an examination of the changes in the metal complex-induced TAR cleavage pattern brought about by addition of the well-known TAR binders, the Tat peptide and neomycin B.

Figure 1.

Terbium(III) complexes investigated in this study.

Figure 2.

RNA oligonucleotides employed in this study.

Experimental

Materials

All chemicals were of reagent grade quality or better, obtained from commercial suppliers and used without further purification. HPLC-grade CH₃CN and distilled H₂O were used throughout, and DMF was dried over 4 Å sieves. 1-Ethyl-1,4,7,10-tetraazacyclododecane was prepared according to a literature method.³⁷ Uridine-3′-p-nitrophenylphosphate was prepared according to a previously published procedure.²⁵ Custom RNA oligonucleotides were procured from Dharmacon and were deprotected using the procedure supplied by the company, then purified by PAGE gels and desalted on a Waters Corporation C-18 Sep-Pak column. Unlabelled tRNA^fMet was kindly donated by Prof. Simpson Joseph (Department of Chemistry and Biochemistry, University of California, San Diego). The Tat peptide (YGRKKRRNRRRP) was available from a previous study.³⁸

Instrumentation

Microanalytical (C, H and N) analyses were performed by Campbell Microanalytical Service, Otago, New Zealand. ¹H NMR Spectra were recorded in CDCl₃ on a Varian 300 MHz spectrometer. The chemical shifts, δ, are reported in ppm (parts per million). The abbreviations for the peak multiplicities are as follows: s (singlet), d (doublet), t (triplet), and m (multiplet). Electrospray Mass Spectrograms were recorded on Micromass Platform II Quadrupole Mass Spectrometer fitted with an electrospray source. A cone voltage of 20 V was used for the ligands and ligand precursor, whilst 70 V was used for the complexes. UV-Vis spectra were recorded using Varian Cary 3 or 5G spectrophotometers and quartz cuvettes.

Syntheses

2-Chloro-N-isopropylacetamide

A solution of chloroacetylchloride (5.48 mL, 68.9 mmol) in acetone (30 mL) was added in a dropwise fashion to a stirred, ice-cooled mixture of isopropylamine (5.86 mL, 68.2 mmol) and Na₂CO₃·H₂O (8.56 g, 69.0 mmol) in H₂O (30 mL) and acetone (30 mL). The mixture was then filtered and the solvent removed under reduced pressure. The remaining residue was dissolved in CH₂Cl₂ (80 mL) and washed with 1 M HCl (2 × 40 mL), followed by 2 M NaHCO₃ (2 × 40 mL). After drying over MgSO₄, the CH₂Cl₂ was removed under reduced pressure to yield the product as a white, crystalline solid. Yield: 8.01 g, 77.5%. ¹H NMR: δ 1.26 (d, J = 6.6 Hz, 6H, CH(CH₃)₂), 4.07 (s, 2H, ClCH₂), 4.15 (septet, J = 6.6 Hz, 1H, CH(CH₃)₂), 6.41 (broad s, 1H, NH). ESI-MS (m/z): 136.1 [M+H]⁺(100%).

2-[7-Ethyl-4,10-bis(isopropylcarbamoylmethyl)-1,4,7,10-tetraazacyclododec-1-yl]-N-isopropylacetamide (L1)

A stirred mixture of 2-chloro-N-isopropylacetamide (0.416 g, 2.74 mmol), 1-ethyl-1,4,7,10-tetraazacyclododecane (0.168 g, 0.839 mmol), CsCO₃ (0.967 g, 2.97 mmol) and KI (0.492 g, 2.96 mmol) in DMF (50 mL) was heated at 70°C for 72 h. After cooling to RT, the mixture was filtered to remove inorganic salts. Removal of the solvent under reduced pressure yielded an oily brown residue, which was taken up in H₂O (30 mL) and then extracted into CHCl₃ (3 × 30 mL). The combined organic layers were dried over anhydrous MgSO₄ and the solvent removed under reduced pressure to yield a yellow oil, which was dissolved in CH₃CN (3 mL). Vapor diffusion of Et₂O into this solution yielded the product in the form of fine, white crystals. Yield: 0.171 g, 40.9 %. ¹H NMR: δ 1.02 (t, J = 7.0 Hz, 3H, NCH₂CH₃), 1.16 (d, J = 6.6 Hz, 6H, CH(CH₃)₂), 1.20 (d, J = 6.6 Hz, 12H, CH(CH₃)₂), 2.03-2.65 (broad set of unresolved signals, 18H, cyclen CH₂ and NCH₂CH₃), 2.95-3.39 (m, 6H, NCH₂CO), 3.36 (s, 2H, NCH₂CO), 4.00 (septet, J = 6.6 Hz, 1H, CH(CH₃)₂), 4.11 (septet, 2H, CH(CH₃)₂), 6.61 (broad s, 2H, NH), 6.78 (d, J = 7.2 Hz, 1H, NH). ESI-MS (m/z): 250.1 [M+2H]²⁺ (100 %), 498.9 [M+H]⁺ (22 %), 520.4 [M+Na]⁺ (5 %).

2-[4,7,10-tris(Isopropylcarbamoylmethyl)-1,4,7,10-tetraazacyclododec-1-yl]-N-isopropylacetamide (L2)

A stirred mixture of 2-chloro-N-isopropylacetamide (1.70 g, 11.2 mmol), 1,4,7,10-tetraazacyclododecane (0.434 g, 2.52 mmol), Cs₂CO₃ (4.064 g, 12.5 mmol) and KI (2.09 g, 12.5 mmol) in DMF (50 mL) was heated at 70°C for 48 h. After cooling to RT, the mixture was filtered to remove inorganic salts. Removal of the solvent under reduced pressure yielded an oily brown residue, which was taken up in H₂O (30 mL) and then extracted into CHCl₃ (3 × 30 mL). The combined organic layers were dried over anhydrous MgSO₄ and the solvent removed under reduced pressure to yield a yellow oil, which was dissolved in CH₃CN (3 mL). Vapor diffusion of Et₂O into this solution yielded the product in the form of fine, white crystals. Yield: 0.394 g, 27.5%. ¹H NMR: δ 1.17 (d, J = 6.6 Hz, 24H, CH(CH₃)₂), 2.25-2.80 (broad set of unresolved signals, 16H, cyclen CH₂), 3.09 (broad s, 8H, CH₂CO), 3.99 (heptet, J = 6.6 Hz, 4H, CH(CH₃)₂), 6.81 (d, J = 7.2 Hz, 4H, NH). ESI-MS (m/z): 285.6 [M+2H]²⁺ (10 0%), 569.8 [M+H]⁺ (26 %), 591.8 [M+Na]⁺ (4 %).

[Tb(L1)(OTf)(OH₂)](OTf)₂·MeCN (C1)

Tb(OTf)₃ (0.727 g, 1.20 mmol) and L1 (0.585 g, 1.18 mmol) were dissolved in CH₃CN (50 mL) and refluxed for 72 h. The solvent was then removed under reduced pressure to yield a brown solid, which was redissolved in CH₃CN (3 mL). Vapor diffusion of Et₂O into this solution yielded very pale yellow crystals of the product, which were suitable for X-ray crystallography. Yield: 0.906 g, 66.0 %. Micro: Calc for C₃₀H₅₆F₉N₈O₁₃S₃Tb: C 31.0 H 4.8 N 9.6%, Found C 31.0 H 4.9 N 9.7 %. ESI-MS (m/z): 218.8 [Tb+L1]³⁺ (39 %), 654.7 [Tb+L1-2H]⁺ (35 %), 804.6 [Tb+L1+OTf-H]⁺ (100 %), 954.3 [Tb+L1+2OTf]⁺ (4 %).

[Tb(L2)(OH₂)](OTf)₃ (C2)

Tb(OTf)₃ (0.270 g, 0.445 mmol) and L2 (0.250 g, 0.440 mmol) were dissolved in CH₃CN (50 mL) and refluxed for 24 h. The solvent was then removed under reduced pressure to yield a brown solid, which was redissolved in CH₃CN (3 mL). Vapor diffusion of Et₂O into this solution yielded very pale yellow crystals of the product. Yield: 0.174 g, 33.1 %. Micro: Calc for C₃₁H₅₈F₉N₈O₁₄S₃Tb: C 31.2 H 4.9 N 9.6 %, Found C 31.5 H 5.3 N 9.4 %. ESI-MS (m/z): 242.7 [Tb+L2]³⁺ (100%), 725.7 [Tb+L2-2H]⁺ (83%), 875.6 [Tb+L2+OTf-H]⁺ (42 %), 1025.3 [Tb+L2+2OTf]⁺ (11 %).

BNPP cleavage kinetics

A solution (total reaction volume = 2.5 mL) containing 200 μM Tb(III) complex, 50 mM MOPS buffer (pH 7.5 @ 21°C) and 100 mM NaClO₄ was equilibrated at 50°C for 20 min in a quartz cuvette (1 cm). After this time, 10 μL of a 5 mM bis(p-nitrophenyl)phosphate (BNPP) solution was added (yielding [BNPP]₀ = 20 μM), and the release of p-nitrophenoxide (NP^-) ion monitored by measuring the absorbance at 400 nm every 0.5 seconds for 45 min, using a UV-Vis spectrophotometer. The cleavage of BNPP (release of NP^-) was modeled as a first-order process, by fitting the absorbance versus time data to the equation, Abs = A + Be^-k_obst, where A and B are constants.

UpNP cleavage kinetics

A solution (total reaction volume = 2.5 mL) containing 200 μM Tb(III) complex, 50 mM MOPS (pH 7.5 @ 21°C), and 150 mM NaClO₄ was equilibrated at the appropriate temperature for 20 min in a quartz cuvette (1 cm). After this time, 10 μL of 5 mM UpNP was added (yielding [UpNP]₀ = 20 μM), and the release of NP^- measured and modeled as described above for the BNPP cleavage experiments.

³²P 5′-end-labeling of RNA oligonucleotides

Dephosphorylated RNA (300 pmol) was reacted with Phage T4 polynucleotide kinase (80 U) in kinase buffer (70 mM Tris.HCl (pH 7.6 @ 25°C), 10 mM MgCl₂, 5 mM DTT) containing 40 μCi of γ-³²P-ATP (4500 C_i mmol^-1) and 1 mg mL^-1 bovine serum albumin. The total volume of the reaction was 30 μL. The solution was heated at 37°C for 1.5 h, quenched with 20 μL of loading buffer and loaded directly onto a 20% denaturing PAGE gel. The gel was imaged by autoradiography, allowing location of the product band, which was cut and extracted from the gel using 0.5 M NH₄OAc (2 × 3 mL at 55°C for 30 min). The resulting solution was desalted using a Water C-18 Sep-Pak column. The product was then resuspended in 100 μL of H₂O, yielding a 200k cpm μL^-1 solution of end-labelled RNA.

Cleavage of RNA oligonucleotides

The ³²P end-labelled RNA oligonucleotides were refolded by slow thermal annealing, achieved by heating a 10 μM solution at 80°C for 2 min, and then cooling the solution to 4°C over a period of 1 h. The annealed RNA solutions were used to prepare a solution containing1 μM RNA, 50 mM MOPS buffer (pH 6.5 @ 21°C), 100 mM NaClO₄, and a varying concentration of C1 (which forms [Tb(L1)(OH₂)₂]³⁺ upon dissolution), to a total reaction volume of 50 μL. Control reactions containing RNA in the reaction buffer, or 2.5 μL of DMSO and the complex, were also prepared and incubated along with the other reaction mixtures. The resulting solutions were incubated at 37°C. 5 μL aliquots were periodically taken from the reaction mixture (up to 48 h), quenched with 5 μL of loading buffer (7 M urea in 10 mM Tris.HCl (pH 8.0), 100 mM EDTA and 0.05% w/v bromophenol blue), and cooled to -20°C. The solutions were stored at this temperature until all samples had been collected. Aliquots of these stored solutions were then subjected to electrophoretic separation on a 20% denaturing PAGE gel. The developed gel was visualized by phosphorimaging and analyzed using the ImageQuant™ software package (Amersham Pharmacia Biotech, 2000).

Cleavage of ^fMettRNA

A 10 μM solution of ^fMettRNA (spiked with ³²P end-labeled ^fMettRNA) was prepared in folding buffer (10 mM MgCl₂, 20 mM MOPS buffer (pH 7.0 @ 37°C) and 100 mM NaClO₄). This solution was incubated at 42°C for 30 min, to fold the tRNA, and then used directly to prepare a 50 μL solution containing 1 μM RNA, 50 or 100 μM C1, 50 mM MOPS buffer (pH 6.5 @ 21°C), 1 mM MgCl₂ and 100 mM NaClO₄, which was incubated at 37°C. 10 μL aliquots were periodically taken (up to 8 h), quenched with 10 μL of loading buffer, and cooled immediately to -20°C. After all samples had been collected, they were subjected to electrophoretic separation on a 10% denaturing PAGE gel. The developed gel was visualized by phosphorimaging and analyzed using the ImageQuant™ software package.

Footprinting of TAR-Tat and TAR-neomycin interaction by C1

A 50 μL solution containing 1 μM TAR (spiked with ³²P end-labeled TAR), 100 mM NaClO₄, 50 mM MOPS buffer (pH 6.5 @ 21°C) and either the Tat peptide (5-50 μM) or neomycin B (10-30 μM) was equilibrated at 37°C for 20 min. C1 was then added to yield a 50 μM final concentration of [Tb(L1)(OH₂)₂]³⁺. The solution was incubated at 37°C, with 10μL aliquots being removed after 2-6 h, quenched with 10μL loading buffer and cooled immediately to -20°C. The aliquots were subjected to electrophoretic separation on a 20% denaturing PAGE gel, visualized by phosphorimaging, and analyzed using the ImageQuant™ software package.

X-ray crystallography

Intensity data were collected on a colorless crystal of C1 (0.31 × 0.21 × 0.10 mm) at 123 K using a Bruker Apex II CCD fitted with graphite-monochromated Mo-$\mathrm{K}\underset{.}{\alpha }$ radiation (0. 71073 Å). The data were collected to a maximum 2θ value of 55° and processed using the Bruker Apex II software package. Crystal parameters and details of the data collection are summarized in Table S1 (See Supplementary Material).

The crystal structure of C1 was solved using the SHELX-97 software package,^39,40 and expanded using standard Fourier transform routines. All hydrogen atoms were placed in idealized positions, except the water hydrogens, which were located on the Fourier difference map. There was positional disorder for one of the isopropyl groups (containing C18, C19 and C20). The two disordered positions had the occupancies refined against each other.

Results and Discussion

Synthesis of ligands and terbium(III) complexes

The two ligands employed in this study, L1 and L2, were readily accessed through Finkelstein reactions between 2-chloro-N-isopropylacetamide and 1-ethylcylen or cyclen (Scheme 1). Formation was confirmed by ¹H NMR spectroscopy and electrospray ionization (ESI) mass spectrometry. Reaction of the ligands with Tb(OTf)₃ in refluxing acetonitrile yielded the corresponding terbium(III) complexes, which were crystallized as [Tb(L1)(OTf)(OH₂)](OTf)₂·MeCN (C1) and [Tb(L2)(OH₂)](OTf)₃ (C2), following diffusion of diethyl ether into the reaction mixtures. ESI mass spectra showed peaks corresponding to [Tb+L]³⁺, [Tb+L-2H]⁺, [Tb+L+OTf-H]⁺ and [Tb+L+2OTf]⁺ (L = L1, L2), confirming formation of the complexes.

Scheme 1.

Synthesis of ligands L1 and L2.

X-ray structure determination

The precise molecular structure of C1 was established by X-ray crystallography. The asymmetric unit of the triclinic ($P\stackrel{‒}{1}$) unit cell of C1 contains the complex cation [Tb(L1)(OTf)(OH₂)]²⁺ (Figure 3), two non-coordinated triflate anions and a co-crystallized acetonitrile molecule. The terbium(III) center is nine-coordinate, residing in a distorted, capped square-antiprism coordination environment (Figure 4), with the four nitrogen donors of the cyclen macrocycle occupying the basal positions of the square antiprism, and three oxygens from the amide pendant arms and an O-coordinated triflate anion forming the “roof” of the antiprism. A coordinated water provides the apical cap. Importantly, the crystal structure confirms that whilst the terbium center in C1 is tightly bound by L1, two of the coordination sites are occupied by labile ligands.

Figure 3.

Thermal ellipsoid plots of the complex cation unit and the asymmetric unit of C1 (ellipsoids drawn at 50% probability level; hydrogen atoms and the disorder about the isopropyl group (attached to N6) have been omitted for clarity; hydrogen-bonds are indicated by dashed lines).

Figure 4.

Thermal ellipsoid plots showing the distorted, capped square antiprism coordination environment about the terbium(III) center in C1 (least-squares planes through four oxygen donors and through four nitrogen donors shown in red and blue, respectively).

The Tb-N bonds at the base of the square vary by 0.1 Å, c.f. Tb(1)-N(1)/Tb(1)-N(3) (2.715(1) and 2.705(2) Å, respectively) and Tb(1)-N(2)/Tb(1)-N(4) (2.616(2) and 2.634(2) Å, respectively). This difference is also reflected in the N-Tb-N angles, which are distorted from the ideal square angle of 90° (Table S2, see Supplementary Material), due to the constraints imposed by the cyclen macrocycle. The coordinated amide oxygens all share a Tb-O distance of ca. 2.35 Å, with the weaker coordinating triflate having a Tb-O length of 2.421(1) Å. Due to the affinity of terbium for oxygen donors, the Tb-O bond lengths are shorter on average than the Tb-N bonds by ca. 0.3 Å. Again, due to the geometric constraints of the five-membered rings formed by the coordinating amide oxygens, the N-Tb-O_amide angles are all below the ideal 90°. The capping coordinated water has a Tb-O distance of 2.413(1) Å, and is symmetrically positioned atop the square antiprism, as evidenced by the fact that all of the O(1)-Tb(1)-O_{Amide/Triflate} angles are around 70°. The coordination geometry, bond distances and angles found in C1 are very similar to those found for other cyclen-based terbium(III) complexes with pendants featuring oxygen donors.^33,41-43

Cleavage of model phosphate diesters

The capacity of the two new terbium(III) complexes to cleave phosphate ester linkages was initially probed using the well-studied, activated phosphate diesters, bis(p-nitrophenyl)phosphate and (BNPP) and uridine-3′-p-nitrophenylphosphate (UpNP), which on cleavage liberate p-nitrophenoxide (NP^-), a chromophore that adsorbs at 400 nm, allowing the reaction progress to be readily monitored by UV-Visible spectrophotometry. The major difference between these substrates is that UpNP incorporates a ribose ring with a 2′-OH group, and is thus more representative of the structure of RNA. Akin to RNA cleavage, release of NP^- from UpNP is able to proceed via an intramolecular transesterification mechanism, involving internal nucleophilic attack by the 2′-OH group (Scheme 2). Hydrolytic cleavage of BNPP occurs through a mechanism more closely aligned to that involved in DNA hydrolysis, for which an external nucleophile is required.

Scheme 2.

Mechanism for cleavage of UpNP by general base-assisted intramolecular transesterification.

Test reactions conducted at pH 7.5, under first-order conditions (large excess of complex relative to substrate), revealed that neither C1 nor C2 were reactive towards BNPP, even at elevated temperature (50°C).⁴⁴ C2 was also found to be completely ineffectual in promoting the cleavage of UpNP above background (hydroxide ion-catalyzed) levels. In stark contrast, C1 was found to greatly accelerate the rate of UpNP cleavage; 200 μM complex yielded an observed rate constant, k_obs, of 5.5(1) × 10^-2 s^-1 at 21°C, compared to 8.25(9) × 10^-4 s^-1 measured in the absence of complex. This corresponds to an apparent second-order rate constant of 277(5) M^-1s^-1 (a plot of k_obs against C1 concentration was linear up to [complex] = 800 μM, indicating a first-order dependence of the reaction rate on C1). Thus, under these conditions, 1 M of complex is expected to accelerate the rate of UpNP cleavage by a factor of 3.4 × 10⁵ above the background rate of transesterification.

In the case of C2, the lack of reactivity towards both BNPP and UpNP can be rationalized in terms of the availability of only one labile coordination site on the terbium(III) center; substrate binding leaves no sites available for a water or hydroxo ligand that may act as a nucleophile or general base catalyst. The coordination sphere in C1 is better suited to accommodate both monodentate substrate binding and provision of a water/hydroxide nucleophile. The inability of this complex to promote BNPP cleavage, however, suggests that the steric bulk of the L1 ligand and BNPP prevent the C1-BNPP adduct from adopting a suitable conformation for direct nucleophilic attack to occur (Figure 5). Chang et. al., who conducted a systematic study of the effects of various parameters on the hydrolytic activity of macrocyclic lanthanide complexes, reported that a europium(III) complex with only two inner-sphere coordinated waters was also relatively inactive as BNPP cleavage agent.⁴⁵ UpNP cleavage by C1 therefore most likely proceeds through a mechanism involving formation of a C1-UpNP adduct, followed by internal nucleophilic attack by the 2′-OH group on the electropositive P atom. The disparate reactivity of C1 and C2 towards UpNP indicates that the terbium(III) center in C1 is able to facilitate this transesterification step by virtue of its additional labile coordination site, conceivably through provision of a hydroxide ligand that functions as a general base catalyst in the mechanism (Figure 5). Our observations are consistent with the findings of Morrow,¹⁹ who reported that [LaTCMC]³⁺, with two inner-sphere coordinated waters, promotes hydrolysis of RNA oligomers, while two complexes with only one coordinated water, [EuTCMC]³⁺ and [DyTCMC]³⁺, are inactive (TCMC = 1,4,7,10-tetrakis(carbomylmethyl)-1,4,7,10-tetraazacyclododecane; this ligand is capable of octadentate coordination).

Figure 5.

Proposed rationale for the disparate reactivity of C1 towards BNPP and UpNP. UpNP features a potential internal nucleophile (2′-OH), whose reactivity may be enhanced through deprotonation by a Tb(III)-bound hydroxo group. Direct nucleophilic attack of a Tb(III)-bound hydroxo/water group on BNPP appears to be unfavorable.

Cleavage of RNA

Encouraged by the finding that C1 is able to greatly enhance the rate of cleavage of the RNA system model, UpNP, we next examined whether this complex could cleave the sugar-phosphate backbone of RNA. In the first instance, C1 was reacted with a short ³²P 5′-end-labeled construct of the HIV TAR element (Figure 2) under pseudo first-order conditions, and the reaction mixture analyzed by polyacrylamide gel electrophoresis (PAGE) at increasing time intervals (Figure 6). Cleavage activity was evident by the appearance of several bands of lower molecular weight than the original TAR construct. Concomitant PAGE analysis of the degradation products obtained by RNase T1 digestion of the TAR construct (lane 2 of PAGE gel), located the initial scission points to be almost exclusively within the triplet base “bulge” (U²³C²⁴U²⁵) of the stem-loop structure of the construct, which NMR studies have previously shown to coincide with the position where the sugar-phosphate backbone is most exposed (Figure 6).⁴⁶ Over time, the initial cleavage fragments were themselves cleaved to produce smaller secondary fragments. This was indicated by the fact that the intensities of the initial set of cleavage bands began to decrease after ca. 15 h (Figure 7) and coincide with the increased appearance of a set of lower molecular weight bands in the PAGE gel (Figure 6). The decrease in intensity of the band corresponding to the intact TAR construct was successfully modeled as a first-order process (Figure 7), yielding a k_obs value of 2.78(8) × 10^-5 s^-1, corresponding to an apparent second-order rate constant of 0.56(2) M^-1s^-1 (under the assumption of a first-order dependence of the cleavage on the concentration of C1). This rate of reaction is substantially slower than that found for the transesterification of UpNP (vide supra), which is not surprising, given that the phosphate diester linkage in UpNP is highly activated towards cleavage because of the strongly electron-withdrawing nitrophenyl group, combined with the fact that the leaving group (NP^- ion) is highly resonance-stabilized.

Figure 6.

Phosphorimage of denaturing PAGE gel showing time dependence of cleavage of TAR construct by C1 (lanes 3-9; incubation times indicated above lanes), together with cleavage products from alkaline hydrolysis or RNase T1 digestion (under denaturing conditions) of TAR (lanes 1 and 2, marked “AH” and “T1 CEU”, respectively). Conditions: 1 μM TAR, 50 μM C1, 50 mM MOPS (pH 6.5 @ 21°C), 100 mM NaClO₄, 37°C. Also shown is the NMR structure of the TAR stem-loop, highlighting the “bulge” region where C1-induced strand cleavage is initially focussed (PDB reference: 1ANR,⁴⁶ graphic produced using UCSF Chimera⁴⁷).

Figure 7.

Plot of normalized band intensity-versus- time for the intact TAR construct as well as the major initial cleavage fragment produced via scission of the phosphate ester linkage adjacent to C24 after incubation with C1.

Similar cleavage experiments performed with short constructs of the HIV DIS and ribosomal A-site (Figure 2) revealed that C1 is also able to cleave the more exposed `solvent-exposed' sections of the sugar-phosphate backbone of these RNA oligonucleotides (see Figure S1 in Supplementary Material). This is evident by cleavage located about the unpaired adenine doublet in the A-Site construct and the C²⁸⁰A²⁷⁹C²⁷⁸ bulge in the HIV DIS extended duplex (Figure S1). In all cases, the extent of cleavage was found to be essentially the same in the presence or absence of dimethylsulfoxide (DMSO), a known radical scavenger, indicating that the cleavage reaction does not proceed through an oxidative or reductive mechanism.⁴⁸ Furthermore, the cleavage bands were found to correspond to those obtained via alkaline hydrolysis of the various RNA constructs, strongly supporting a transesterification/hydrolytic mechanism of cleavage (Figures 6 and S1). The sites of cleavage were unaffected, and the degree of cleavage was found to be similar, over the pH range 6.0-8.5 (see Figure S2 in Supplementary Material). It was also found that using a more strongly coordinating salt (NaCl, instead of NaClO₄) to control ionic strength led to minimal impact on cleavage efficiency. Taken together, these results indicate that C1 is able to function as a robust synthetic ribonuclease across a reasonably broad spectrum of conditions commonly encountered in RNA manipulation/characterization studies.

The RNA cleavage properties of C1 were further probed by examining its interaction with a more complex RNA motif, tRNA^fMet (Figure 2), a 77-nucleotide RNA sequence that is folded into an intricate three-dimensional structure featuring a number of solvent-exposed loops. PAGE analysis of incubated mixtures of ³²P 5′-end-labeled tRNA^fMet with C1 again clearly revealed that C1 is able to greatly accelerate RNA cleavage above the background rate (Figure 8), and that cleavage is restricted to specific sites within the RNA structure. The rate of cleavage was found to roughly double upon increasing the concentration of C1 from 50 to 100 μM, confirming that C1 is responsible for the observed activity. By mapping the cleavage product bands to co-migrating RNase T1 digestion bands, the cleavage sites were established to be within the more exposed, single-stranded regions of tRNA^fMet, namely the DHU arm, the anticodon arm, and the extra arm (Figure 8). Correlation of these cleavage regions with the crystal structure of tRNA^fMet as solved by Tisné and co-workers⁴⁹ clearly reveals that the strand scission induced by C1 is located in regions where the phosphate ester backbone is more `exposed' (See Figure 8) to attack by the bulky Tb³⁺ complex.

Figure 8.

Phosphorimage of denaturing PAGE gel showing time dependence and concentration dependence of cleavage of ^fMettRNA by C1 (lanes 3-11), together with cleavage products from alkaline hydrolysis or RNase T1 digestion (under denaturing conditions) of TAR (lanes 1 and 2, marked “AH” and “T1 CEU”, respectively). Lanes 2-4: 50 μM C1 with ^fMettRNA (2, 4, 8 h incubation time); lanes 5-7: 100 μM C1 with ^fMettRNA (2, 4, 8 h incubation time); lanes 8-10: ^fMettRNA only control (2, 4, 8 h incubation time); incubation times indicated above lanes. Conditions: 1 μM ^fMettRNA, 50 mM MOPS (pH 6.5 @ 21°C), 100 mM NaClO₄, 1 mM MgCl₂, 37°C. Shown also is the crystal structure of tRNA^fMet (as determined by Tisné et al.⁴⁹ (PDB: 3cw5), graphic prepared using UCSF Chimera⁴⁷) highlighting (in red) the areas of focused strand scission induced by C1.

Footprinting experiments

Having established that C1 is able to selectively cleave solvent-exposed, single-stranded (or strained sections) of a folded RNA structure, we next explored whether C1 could be employed as a new footprinting reagent, useful for examining the interaction between ligands and their RNA targets. RNA-ligand footprinting is often carried out using RNase enzymes, however, this can sometimes be a complex and time consuming task (requiring case-by-case optimization), and the conditions required for enzyme activity may be incompatable with the conditions under which ligand binding occurs. Thus, robust synthetic reagents that are able to function as artificial ribonucleases under near physiological conditions represent valuable additions to the arsenal of tools used to examine RNA structure and RNA-ligand interactions. The Fe(II)-EDTA/H₂O₂ system, which is utilised to generate Fenton based hydroxide radicals, is often used for DNA and RNA footprinting,²² but its lack of sequence specificity can be problematic.⁵⁰

We chose the TAR construct as a model RNA target, since a great deal is known about its interactions with a range of different ligands. Figure 9 shows the cleavage “footprints” observed upon incubating C1 with the TAR construct, both in isolation and in the presence of two known TAR-binders, the Tat peptide and neomycin B. The Tat peptide comprises the portion of the Tat protein that is involved in binding of this protein to the TAR element, an interaction that that is vital to the HIV replication cycle. It is known that the Tat peptide binds to the triplet base “bulge” region of TAR.⁵¹ Moreover, NMR studies carried out by Crothers et al.⁵¹ suggest that upon Tat binding, the phosphate ester groups associated with U²³ and C²⁴ become more exposed relative to that of U²⁵. Consistent with these observation/predictions, we found a change in the cleavage footprint upon addition of Tat peptide to the C1/TAR mixture: cleavage adjacent to U²⁵ was suppressed, whilst cleavage adjacent to U²³ was enhanced.

Figure 9.

Phosphorimage of denaturing PAGE gel showing “footprints” for cleavage of TAR by C1, both in the absence and presence of Tat peptide or neomycin B. Lanes 1 and 9: alkaline hydrolysis ladder; lane 2: RNase T1 digestion of TAR (under denaturing conditions); lanes 3 and 4: C1 with TAR (2, 6 h incubation time); lanes 4 and 5: C1 with TAR (6 hour reaction time); lanes 5 and 6: C1 with TAR and Tat (2, 6 h incubation time); incubation times indicated above lanes. Conditions: 1 μM TAR, 50 μM C1, 50 mM MOPS (pH 6.5 @ 21°C), 100 mM NaClO₄, 37°C.

It has been postulated that the naturally-occurring aminoglycoside, neomycin B, also binds the triplet base “bulge” region of TAR,⁵² but that this is a more dynamic on/off binding interaction than for Tat. Accordingly, we found that the cleavage patterns obtained both in the absence and presence of neomycin B were very similar, but that the extent of cleavage was about four-fold lower when neomycin B was present, as indicated by the fact that the intensities of the cleavage bands were about four-fold less intense than when neomycin B was absent. C1 is thus able to detect the interaction of neomycin B with TAR, by virtue of the bound state of neomycin B providing TAR with some protection against C1-induced cleavage.

Conclusion

A new terbium(III) complex, C1, which is able to rapidly accelerate the cleavage of UpNP, as well as the solvent-exposed, single-stranded regions of folded RNA motifs, has been developed. The complex does not require co-reactants to cause RNA scission, is able to function over a range of pHs, and exhibits cleavage activity that is sensitive to the presence of RNA-binding ligands. C1 is thus a suitable candidate for use in footprinting assays directed at improving our understanding of RNA structure and RNA-ligands interactions.