Localized chemical reactivity in double-stranded DNA associated with the intercalative binding of benzo[e]pyridoindole and benzo[g]pyridoindole triple-helix-stabilizing ligands

Responses of DNA Mismatch Repair Proteins to a Stable G-Quadruplex Embedded into a DNA Duplex Structure

DNA mismatch repair (MMR) plays a crucial role in the maintenance of genomic stability. The main MMR protein, MutS, was recently shown to recognize the G-quadruplex (G4) DNA structures, which, along with regulatory functions, have a negative impact on genome integrity. Here, we studied the effect of G4 on the DNA-binding activity of MutS from Rhodobacter sphaeroides (methyl-independent MMR) in comparison with MutS from Escherichia coli (methyl-directed MMR) and evaluated the influence of a G4 on the functioning of other proteins involved in the initial steps of MMR. For this purpose, a new DNA construct was designed containing a biologically relevant intramolecular stable G4 structure flanked by double-stranded regions with the set of DNA sites required for MMR initiation. The secondary structure of this model was examined using NMR spectroscopy, chemical probing, fluorescent indicators, circular dichroism, and UV spectroscopy. The results unambiguously showed that the d(GGGT)₄ motif, when embedded in a double-stranded context, adopts a G4 structure of a parallel topology. Despite strong binding affinities of MutS and MutL for a G4, the latter is not recognized by E. coli MMR as a signal for repair, but does not prevent MMR processing when a G4 and G/T mismatch are in close proximity.

The effects of DNA supercoiling on G-quadruplex formation

Guanine-rich DNAs can fold into four-stranded structures that contain stacks of G-quartets. Bioinformatics studies have revealed that G-rich sequences with the potential to adopt these structures are unevenly distributed throughout genomes, and are especially found in gene promoter regions. With the exception of the single-stranded telomeric DNA, all genomic G-rich sequences will always be present along with their C-rich complements, and quadruplex formation will be in competition with the corresponding Watson–Crick duplex. Quadruplex formation must therefore first require local dissociation (melting) of the duplex strands. Since negative supercoiling is known to facilitate the formation of alternative DNA structures, we have investigated G-quadruplex formation within negatively supercoiled DNA plasmids. Plasmids containing multiple copies of (G₃T)_n and (G₃T₄)_n repeats, were probed with dimethylsulphate, potassium permanganate and S1 nuclease. While dimethylsulphate footprinting revealed some evidence for G-quadruplex formation in (G₃T)_n sequences, this was not affected by supercoiling, and permanganate failed to detect exposed thymines in the loop regions. (G₃T₄)_n sequences were not protected from DMS and showed no reaction with permanganate. Similarly, both S1 nuclease and 2D gel electrophoresis of DNA topoisomers did not detect any supercoil-dependent structural transitions. These results suggest that negative supercoiling alone is not sufficient to drive G-quadruplex formation.

Ligand binding mode to duplex and triplex DNA assessed by combining electrospray tandem mass spectrometry and molecular modeling

In this paper, we report the analysis of seven benzopyridoindole and benzopyridoquinoxaline drugs binding to different duplex DNA and triple helical DNA, using an approach combining electrospray ionization mass spectrometry (ESI-MS), tandem mass spectrometry (MS/MS), and molecular modeling. The ligands were ranked according to the collision energy (CE(50)) necessary to dissociate 50% of the complex with the duplex or the triplex in tandem MS. To determine the probable ligand binding site and binding mode, molecular modeling was used to calculate relative ligand binding energies in different binding sites and binding modes. For duplex DNA binding, the ligand-DNA interaction energies are roughly correlated with the experimental CE(50), with the two benzopyridoindole ligands more tightly bound than the benzopyridoquinoxaline ligands. There is, however, no marked AT versus GC base preference in binding, as supported both by the ESI-MS and the calculated ligand binding energies. Product ion spectra of the complexes with triplex DNA show only loss of neutral ligand for the benzopyridoquinoxalines, and loss of the third strand for the benzopyridoindoles, the ligand remaining on the duplex part. This indicates a higher binding energy of the benzopyridoindoles, and also shows that the ligands interact with the triplex via the duplex. The ranking of the ligand interaction energies compared with the CE(50) values obtained by MS/MS on the complexes with the triplex clearly indicates that the ligands intercalate via the minor groove of the Watson-Crick duplex. Regarding triplex versus duplex selectivity, our experiments have demonstrated that the most selective drugs for triplex share the same heteroaromatic core.

Sequence-specific interactions of minor groove binders with restriction fragments of cDNAs for H tau 40 protein and MAP kinase 2. A qualitative and quantitative footprinting study

A series of DNA minor groove binders comprising netropsin, distamycin, the bisquaternary ammonium heterocycles SN 6999 and SN 6570, cis-diammine platinum(II)-bridged bis-netropsin, cis-diammine platinum(II)-bridged bis-distamycin and bis-glycine-linked bis-distamycin were investigated for sequence-specific interactions. The oligonucleotides used were the 154 base pair HindIII-RsaI restriction fragment of cDNA of h tau 40 protein and the 113 base pair NcoI-PvuII restriction fragment of cDNA of MAP kinase 2. Both proteins are believed to be involved in the pathology of Alzheimer's disease. For all these ligands, binding sites were localised at positions 1134-1139 (5'AATCTT3'), 1152-1156 (5'ATATT3') and 1178-1194 (5'TTTCAATCTTTTTATTT3') for the former and 720-726 (5'TATTCTT3'), 751-771 (5'AATTGTATAATAAATTTAAAA3') and 781-785 (5'TATTT3') for the latter. The AT-preference of ligand binding was obvious and footprint titration experiments were applied to estimate binding constants (Ka) for each individual binding site mentioned above. The binding strength decreases in the order netropsin > distamycin > SN 6999 approximately SN 6570>platinum-bridged netropsin or distamycin approximately bis-glycine-bridged distamycin and was found independently of the binding sites examined. GC-base pairs interspersed in short AT-tracts reduced the Ka-values by as much as two orders of magnitudes. The dependence of extended bidentate as well as of monodentate binding of netropsin and distamycin derivatives on the length of AT-stretches has been discussed.

Stabilization of triple helical DNA by a benzopyridoquinoxaline intercalator

Biophysical, footprinting, and chemical probing experiments are described which characterize the triple helix-stabilizing effects of a benzo[f]pyridoquinoxaline derivative BfPQ-4,3 structurally related to the previously reported benzo[f]pyridoindole compound BePI [Mergny et al. (1992) Science 256, 1681-1684]. Two parallel triple helix model systems have been investigated; one in which the third strand matched perfectly a 27 base pair purine-pyrimidine motif in target DNA and another in which the third strand was one nucleotide longer, i.e., a 28-mer. In the latter system, the pairing of the (Y)28 third strand to the (Y.R)27 target induces the formation of a bulge containing at least one unpaired base, which can be evidenced by chemical probing experiments with osmium tetroxide. BPQ, which uinwinds a duplex DNA by 17 degrees as judged by viscometric experiments and otherwise behaves as a typical nonspecific intercalculating drug, promotes the formation of Y.R.Y parallel triple helix containing both T.A.T and C.G.C+ triplets. Both DNase I and MPE.FeII footprinting experiments concur that triplex formation with the target (Y.R)27 sequence can be detected in the presence of BPQ at about 10-fold lower oligonucleotide concentrations than are required to produce an equivalent footprint in the absence of the drug. In addition, BPQ will promote binding to the polypurine-polypyrimidine target sequence by the longer mismatched oligonucleotide, providing significant stabilization of the parallel bulge-containing(Y.R)27,(Y)28 triplex with nearly the same efficiency as the bulge-free (Y.R)27.(Y)28 triplex. Thus in vivo BPQ might enhance the formation of both undesired and desired DNA triplexes. By performing an MPE*FeII probing reaction with a 5'-32 P-labeled oligonucleotide third strand, we have obtained evidence that BPQ is actually bound to the triplex region and may distort in a sequence-specific fashion.

Ligand-induced formation of hoogsteen-paired parallel DNA

INTRODUCTION

Based on molecular modeling studies, a model has been proposed for intercalation of triple-helix-specific ligands (benzopyridoindole (BPI) derivatives) into triple helices, in which the intercalating compounds interact mainly with the Hoogsteen-paired strands of the triple helix. We set out to test this model experimentally using DNA duplexes capable of forming parallel Hoogsteen base-paired structures.

RESULTS

We have investigated the possible formation of a parallel DNA structure involving Hoogsteen hydrogen bonds by thermal denaturation, FTIR spectroscopy and gel-shift experiments. We show that BPI derivatives bind to Hoogsteen base-paired duplexes and stabilize them. The compounds induce a reorganization from a non-perfectly matched antiparallel Watson- Crick duplex into a perfectly matched parallel Hoogsteen-paired duplex.

CONCLUSIONS

These results suggest that preferential intercalation of BPI derivatives in triple helices is due to their ability to interact specifically with the Hoogsteen-paired bases. The results are consistent with a model proposed on the basis of molecular modeling studies using energy minimization, and they open a new field of investigations regarding the biological relevance of Hoogsteen base-pairing.