Article pubs.acs.org/bc
Binding and NMR Structural Studies on Indoloquinoline− Oligonucleotide Conjugates Targeting Duplex DNA Andrea Eick, Fanny Riechert-Krause, and Klaus Weisz* Institute of Biochemistry, Ernst-Moritz-Arndt University Greifswald, Felix-Hausdorff-Strasse 4, D-17487 Greifswald, Germany S Supporting Information *
ABSTRACT: An 11-phenyl-indolo[3,2-b]quinoline (PIQ) was tethered through an aminoalkyl linker to the 5′-end of four pyrimidine oligonucleotides with T/C scrambled sequences at their two 5′-terminal positions. Binding to different double-helical DNA targets formed parallel triple helices with a PIQ-mediated stabilization that strongly depends on pH and the terminal base triad at the 5′-triplex−duplex junction. The most effective stabilization was observed with a TAT triplet at the 5′-junction under low pH conditions, pointing to a protonated ligand with a high triplex binding affinity and unfavorable charge repulsions in the case of a terminal C+GC triplet at the junction. The latter preference of the PIQ ligand for TAT over CGC is alleviated yet still preserved at higher pH. Intercalation of PIQ at the 5′-triplex−duplex junction as suggested by the triplex melting experiments was confirmed by homonuclear and heteronuclear NMR structural studies on a specifically isotope-labeled triplex. The NMR analysis revealed two coexisting species that only differ by a 180° rotation of the indoloquinoline within the intercalation pocket. NOEderived molecular models indicate extensive stacking interactions of the indoloquinoline moiety with the TAT base triplet and CG base pair at the junction and a phenyl substituent that is positioned in the major groove and oriented almost perpendicular to the plane of the indoloquinoline.
■
INTRODUCTION The sequence-selective targeting of double-helical DNA bears enormous potential for various biotechnological applications, as well as in the area of DNA therapeutics and diagnostics.1−3 For an effective recognition of longer stretches of DNA associated with a reliable readout of the corresponding base pair sequence, synthetic polyamides4 as well as engineered zinc finger proteins5 and single-stranded oligonucleotides6 have been shown in the past to be the most promising tools when used under appropriate conditions. Thus, oligonucleotides can bind to double-helical DNA in its major groove based on a welldefined base−base recognition code to form triple-helical structures.7 However, because formation of triple-stranded DNA is based on Hoogsteen hydrogen bond interactions between bases of the triplex-forming oligonucleotide (TFO) and available hydrogen bond donor and acceptor groups of purine bases in the double-helical DNA target, effective triplex formation is mostly limited to the presence of oligopyrimidine·oligopurine sequences. Also, in the pyrimidine-type triplex motif a pyrimidine third strand oriented parallel to the oligopurine strand of the cognate duplex forms isomorphous T·AT and C+·GC base triplets through the formation of two Hoogsteen hydrogen bonds only after protonation of third strand cytosines, resulting in a considerable pH dependence of triplex stability and less stable triplexes under physiological pH conditions. On the other hand, many small cytotoxic and antitumor drugs bind to DNA through intercalation or groove binding with considerable affinity but often lack sufficient sequence discrimination due to their limited contact area with the duplex that hampers a readout of longer sequences. © 2012 American Chemical Society
Consequently, clinical use of such drugs is compromised by their unspecific binding often associated with severe side effects. Given the various limitations associated with the triplex methodology and with the duplex binding of low-molecularweight ligands, a combination of both approaches is expected to provide for synergistic effects and for a more favorable recognition in terms of affinity and specificity. In fact, TFO conjugates have been employed in the past to improve the sequence-specific molecular recognition of duplex targets under physiological conditions.8,9 Whereas triplex formation mostly benefits from additional stabilization through interactions of the covalently attached ligand with the duplex or triplex, a ligand employed for its chemical reactivity may considerably benefit from the high sequence specificity of the TFO. Thus, the TFO can be used as a delivery system that guides the agent to a particular target site to exert its cross-linking, strand-cleaving, or base-modifying effects.10,11 Several natural and synthetic members of indoloquinolines have been reported to exhibit a large spectrum of biological activities. We have recently shown that a synthetic 11-phenylsubstituted indolo[3,2-b]quinoline (PIQ) exhibits significant cytotoxic activity.12 Interestingly, this polycyclic compound does not seem to bind duplex DNA with high affinity but is able to considerably stabilize triple-helical structures. The thermodynamics of triplex formation when binding a duplex target was recently determined for a corresponding indoloquinoline−TFO Received: October 31, 2011 Revised: March 28, 2012 Published: May 2, 2012 1127
dx.doi.org/10.1021/bc200582u | Bioconjugate Chem. 2012, 23, 1127−1137
Bioconjugate Chemistry
Article
W × 8 100−200 mesh) and applied on a target plate. The main peak is equivalent to m/z M−1. UV−vis Spectroscopy. All UV measurements were performed on a Cary 100 spectrophotometer equipped with a temperature control unit (Varian, Darmstadt, Germany). Melting curves were recorded with 1 data point/°C at 260 nm in a temperature interval from 1 to 90 °C. To prevent water condensation at temperatures below 25 °C on the 1-cm cuvettes, the sample chamber was constantly flushed with nitrogen gas. The lyophilized unmodified or conjugated TFO (1.7 μM) and target duplex (1.45 μM) were dissolved in cacodylate buffer pH 6.0 (0.1 M NaCl, 0.02 M cacodylate, 1 mM spermine) and annealed prior to each melting experiment by heating to 90 °C followed by slow cooling to room temperature. For pH-dependent melting experiments, the cacodylate buffer was adjusted with HCl or NaOH within a pH range 5.0 to 6.5. Measurements included one cooling (0.5 °C/min) followed by a heating period (0.5 °C/min). The melting temperature was determined by the maximum of the first derivative plot of the final heating curve after appropriate smoothing. Fluorescence Measurements. Fluorescence data were acquired with a Jasco FP-6500 spectrofluorometer (Jasco, Tokyo, Japan) with an emission and excitation bandwidth of 5 nm, a response time of 1 s, and 1 data point/nm. Measurements were performed with single-strand concentrations from 1.45 μM up to 3 μM in cacodylate buffer (0.1 M NaCl, 0.02 M sodium cacodylate, 1 mM spermine) at pH values from 5.0 to 6.0. Fluorescence melting experiments were carried out within a temperature range from 20 to 90 °C, using an excitation wavelength of 350 nm to monitor changes in the PIQ fluorescence upon DNA melting. Denaturation temperatures were extracted from the first derivative of the sigmoidal curve obtained from the temperature-dependent fluorescence at 466 nm for the indoloquinoline emission. All data were blank and volume-corrected. NMR Experiments. Samples for the NMR measurements were prepared by the stepwise addition of the TFO or TFO− PIQ conjugate to the target duplex up to a 1:1 molar ratio in an aqueous solution with 100 mM NaCl (TFO) or 84 mM NaCl (TFO−PIQ), 1 mM NaN3, pH 5.0. One-dimensional 1H NMR spectra acquired after each titration step confirmed the complete formation of triple-helical structures through the disappearance of resolved Watson−Crick imino protons of the duplex. Triplex concentrations of the final solutions varied between 0.42 and 0.76 mM. If necessary, the pH of the unbuffered solutions was readjusted after titrations or lyophilizations by the addition of HCl. All NMR spectra were acquired on a Bruker Avance 600 MHz spectrometer equipped with an inverse 1H/13C/15N/31P quadruple resonance cryoprobehead and z-field gradients. Data were processed using Topspin 2.1. Proton chemical shifts were referenced relative to TSP by setting the H2O or HDO signal in 90% H2O/10% D2O or 100% D2O to δH = 4.95 ppm at 10 °C and δH = 4.84 ppm at 20 °C. For the one- and two-dimensional measurements in 90% H2O/10% D2O a Watergate with w5 element or the excitation sculpting sequence was employed for solvent suppression. NOESY experiments in 90% H2O were performed at 283 K with several mixing times between 30 and 300 ms and a spectral width of 15 kHz. 2K × 800 data points with 32, 64, or 128 transients per t1 increment and a recycle delay of 2 s were collected in t2 and t1. Prior to Fourier transformation data were zero-filled to give a 2K × 1K matrix
conjugate with the PIQ drug covalently tethered to 5′-, 3′-, or internal positions of the oligonucleotide with different linkers.13 However, details regarding the sequence selectivity of the drug, its favorable binding site and specific drug−DNA contacts are still vague, yet are a prerequisite for the further structural optimization of the ligand for future applications. Depending on its protonation at the quinoline ring nitrogen, the present PIQ drug may resemble some of the most efficient triplex binding compounds like benzo[e]pyrido[4,3-b]indoles (BePI)14 or naphthylquinolines15 in their sequence-selective binding. The ring chromophores of the latter ligands carry a positive charge at acidic or neutral pH and have been shown to prefer binding to TAT-rich over C+GC-rich triplexes, pointing to charge repulsion between protonated third strand cytosines and the positively charged ligand. Unfortunately, even for the best characterized triplex binders there is a remarkable lack of experimental structural data that clearly define and pinpoint any specific binding interactions. To the best of our knowledge, there is no X-ray or NMR-derived structure of a triplex-bound ligand available. Rather, structural models reported for triplexligand complexes derive from indirect evidence, low-resolution spectroscopic methods, and computational methodologies. In this report, we present studies on the pH dependence, on the sequence selectivity, and on structural details of duplextargeting PIQ conjugates by UV, fluorescence, and NMR spectroscopy. The covalent attachment of the drug at the 5′terminus of the TFO is expected not only to result in a more favorable free energy of binding, but also to promote welldefined binding interactions, amenable to a more detailed thermodynamic and structural analysis. In this respect, the PIQ drug may also serve as an excellent model for the study of triplex−drug interactions in general and of the factors governing their sequence and pH-dependent binding.
■
EXPERIMENTAL PROCEDURES Materials. Unmodified oligonucleotides as well as oligonucleotides modified with 2-aminopurine, with isotope-labeled nucleosides and with 5′-C6 amino linkers were synthesized employing the phosphoramidite method (TIB MOLBIOL, Berlin, Germany). All oligonucleotides carrying modifications were purified by HPLC. Uniformly 13C (98%) and 15N (98%) 2′-deoxyadenosine and thymidine phosphoramidites for the oligonucleotide synthesis were purchased from Cambridge Isotope Laboratories, Inc. (Andover, MA, USA). The NHSfunctionalized indoloquinoline derivative 4,9-dimethyl-11-(4succinimido-carboxyphenyl)-10H-indolo[3,2-b]quinoline16 was used for the coupling reaction with the 5′-amino-modified oligonucleotides as previously described.13 All conjugates were characterized by MALDI-TOF mass spectrometry (see Supporting Information). The concentration of the oligonucleotides was calculated using molar extinction coefficients at 260 nm derived from a nearest-neighbor model.17 For the TFO-PIQ conjugates, the molar extinction coefficient of the ligand (ε260 nm = 26 818 L mol−1 cm−1) was added to the calculated extinction coefficient of the oligonucleotide. Mass Spectrometry. Mass spectral data were recorded on a Bruker Microflex Maldi-TOF with a linear detection method and negative ion mode using Flexcontrol and Flexanalysis software. The measurements were performed with 400 shots. The HPLC-purified conjugates and the matrix solution (saturated 3-hydroxypicolinic acid in water) were desalted via ammonium-activated cation exchange resin beads (Dowex 50 1128
dx.doi.org/10.1021/bc200582u | Bioconjugate Chem. 2012, 23, 1127−1137
Bioconjugate Chemistry
Article
Figure 1. Sequences of triplexes with nonconjugated and PIQ-conjugated TFOs; positions −2, −1, +1, and +2 at the 5′-triplex−duplex junction are highlighted. The PIQ ligand is shown at the top (right) with atom numbering.
additional 5′-linked nucleotides of the third strand were removed, and the N5-protonated PIQ ligand was attached through the C6 amino linker to the TFO at the intercalation site with the model builder of Spartan’08 (Wave function Inc., Irvine, CA, USA). Also, to allow for the correct sequence, base triplets and base pairs were exchanged where necessary. Distortions and excessive strain in the resulting complex were initially removed by a simple energy minimization keeping hydrogen bonds of the Watson−Crick and Hoogsteen base pairs constrained. The PIQ ligand was manually positioned within the intercalation site at the newly created triplex−duplex junction to give drug−DNA interproton distances for all observed NOE crosspeaks of
