SELEX selection of high-affinity oligonucleotides for bacteriophage Ff gene 5 protein

The Ff gene 5 protein (g5p) is a cooperative ssDNA-binding protein. SELEX was used to identify DNA sequences favorable for g5p binding at physiological ionic strength (200 mM NaCl) and 37 degrees C. Sequences were selected from a library of 58-mers that contained a central variable segment of 26 nucleotides. DNA sequences selected after eight rounds of SELEX were mostly G-rich, with multiple copies of CPuGGPy, TPuGGGPy, and/or PyPuPuGGGPy motifs. This was unexpected, since g5p has higher binding affinities for polypyrimidine than for polypurine sequences. The most recurrent G-rich sequence, named I-3, was found to have g5p-binding properties that were correlated with a structural transition. At 10 mM NaCl, I-3 existed in a single-stranded form that was saturated by g5p in an all-or-none fashion. At 200 mM NaCl, I-3 existed in a structured form that showed CD spectral features of G-quadruplexes. The g5p binding affinity for this structured form of I-3 was >100-fold higher than for the single-stranded form. Moreover, the structured I-3 was saturated by g5p in two steps, the first of which was the formation of an apparent initiation complex consisting of one I-3 strand and about three g5p dimers. Nuclease S1 footprinting and other experiments showed that g5p molecules in the initiation complex at 200 mM NaCl were bound directly to the G-rich variable segment and that the structure of I-3 was retained after saturation by g5p. Thus, G-rich motifs may form structures favorable for initiation of g5p binding and also provide the actual g5p-binding sites.     

Vaccinia virus telomeres: interaction with the viral I1, I6, and K4 proteins

The 192-kb linear DNA genome of vaccinia virus has covalently closed hairpin termini that are extremely AT rich and contain 12 extrahelical bases. Vaccinia virus telomeres have previously been implicated in the initiation of viral genome replication; therefore, we sought to determine whether the telomeres form specific protein-DNA complexes. Using an electrophoretic mobility shift assay, we found that extracts prepared from virions and from the cytoplasm of infected cells contain telomere binding activity. Four shifted complexes were detected using hairpin probes representing the viral termini, two of which represent an interaction with the "flip" isoform and two with the "flop" isoform. All of the specificity for protein binding lies within the terminal 65-bp hairpin sequence. Viral hairpins lacking extrahelical bases cannot form the shifted complexes, suggesting that DNA structure is crucial for complex formation. Using an affinity purification protocol, we purified the proteins responsible for hairpin-protein complex formation. The vaccinia virus I1 protein was identified as being necessary and sufficient for the formation of the upper doublet of shifted complexes, and the vaccinia virus I6 protein was shown to form the lower doublet of shifted complexes. Competition and challenge experiments confirmed that the previously uncharacterized I6 protein binds tightly and with great specificity to the hairpin form of the viral telomeric sequence. Incubation of viral hairpins with extracts from infected cells also generates a smaller DNA fragment that is likely to reflect specific nicking at the apex of the hairpin; we show that the vaccinia virus K4 protein is necessary and sufficient for this reaction. We hypothesize that these telomere binding proteins may play a role in the initiation of vaccinia virus genome replication and/or genome encapsidation.     

The high binding affinity of phosphorothioate-modified oligomers for Ff gene 5 protein is moderated by the addition of C-5 propyne or 2′-O-methyl modifications

One of the problems that hamper the use of antisense DNAs as effective drugs is the non-specific binding of chemically-modified oligonucleotides to cellular proteins. We previously showed that the affinity of a model ssDNA-binding protein, the Ff gene 5 protein (g5p), was >300-fold higher for phosphorothioate-modified DNA (S-DNA) than for unmodified dA₃₆, consistent with the propensity of S-DNA to bind indiscriminately to proteins. The current work shows that g5p binding is also sensitive to sugar and pyrimidine modifications used in antisense oligomers. Binding affinities of g5p for 10 36mer oligomers were quantitated using solution circular dichroism measurements. The oligomers contained C-5-propyne (prC), 2′-O-methyl (2′-O-Me) or 2′-OH (RNA) groups, alone or combined with the phosphorothioate modification. In agreement with reported increases in antisense activity, the addition of prC or 2′-O-Me modifications substantially reduced the affinity of oligomers for g5p by ~2-fold compared with the same DNA oligomer sequences containing only phosphorothioate linkages. That is, such modifications moderated the propensity of the phosphorothioate group to bind tightly to the g5p. The Ff g5p could be a useful model protein for assessing non-specific binding effects of antisense oligomer modifications.     

Major cold shock proteins, CspA from Escherichia coli and CspB from Bacillus subtilis, interact differently with single-stranded DNA templates

The family of bacterial major cold shock proteins is characterized by a conserved sequence of 65-75 amino acid residues long which form a three-dimensional structure consisting of five beta-sheets arranged into a beta-barrel topology. CspA from Escherichia coli and CspB from Bacillus subtilis are typical representative members of this class of proteins. The exact biological role of these proteins is still unclear; however, they have been implicated to possess ssDNA-binding activity. In this paper, we report the results of a comparative quantitative analysis of ssDNA-binding activity of CspA and CspB. We show that in spite of high homology on the level of primary structure and very similar three-dimensional structures, CspA and CspB have different ssDNA-binding properties. Both proteins preferentially bind polypyrimidine ssDNA templates, but CspB binds to the T-based templates with one order of magnitude higher affinity than to U- or C-based ssDNA, whereas CspA binds T-, U- or C-based ssDNA with comparable affinity. They also show similarities and differences in their binding to ssDNA at high ionic strength. The results of these findings are related to the chemical structure of DNA bases.     

Ff gene 5 protein has a high binding affinity for single-stranded phosphorothioate DNA

The gene 5 protein (g5p) of Ff bacteriophages is a well-studied model ssDNA-binding protein that binds cooperatively to the Ff ssDNA genome and single-stranded polynucleotides. Its affinity, K omega (the intrinsic binding constant times a cooperativity factor), can differ by several orders of magnitude for ssDNAs of different nearest-neighbor base compositions [Mou, T. C., Gray, C. W., and Gray, D. M. (1999) Biophys. J. 76, 1537-1551]. We found that the DNA backbone can also dramatically affect the binding affinity. The K omega for binding phosphorothioate-modified S-d(A)(36) was >300-fold higher than for binding unmodified P-d(A)(36) at 0.2 M NaCl. CD titrations showed that g5p bound phosphorothioate-modified oligomers with the same stoichiometry as unmodified oligomers. The CD spectrum of S-d(A)(36) underwent the same qualitative change upon protein binding as did the spectrum of unmodified DNA, and the phosphorothioate-modified DNA appeared to bind in the normal g5p binding site. Oligomers of d(A)(36) with different proportions of phosphorothioate nucleotides had binding affinities and CD perturbations intermediate to those of the fully modified and unmodified sequences. The influence of phosphorothioation on binding affinity was nearly proportional to the extent of the modification, with a small nearest-neighbor dependence. These and other results using d(ACC)(12) oligomers and mutant proteins indicated that the increased binding affinity of g5p for phosphorothioate DNA was not a polyelectrolyte effect and probably was not an effect due to the altered nucleic acid structure, but was more likely a general effect of the properties of the sulfur in the context of the phosphorothioate group.     

Insertions of palindromic DNA sequences into the J-F intercistronic region of bacteriophage phi X174 interfere with normal phage growth

The effect of hairpin (cruciform) size on the regulation of gene expression was investigated by cloning a series of palindromic sequences into the non-essential J-F intercistronic region of the bacteriophage phi X174 ins6 genome. Genetic stability of the insert sequence and its effect on the growth efficiency of the phage was used as an initial measure of the biological consequence of hairpin insertions. Multimers of increasing size of the BamHI linker sequence C-C-G-G-A-T-C-C-G-G were inserted into the PvuII site of the parental strain ins6. The largest hairpin that could be constructed and maintained in the phi X174 genome had a stem length of 22 base-pairs and a loop size of four nucleotides (linker tetramer). However, this structure proved to be disadvantageous to the phage and was rapidly deleted from its genome. Trimer inserts were more stable, but were eventually deleted also. Monomer and dimer inserts, though genetically stable, decreased the growth efficiency of the phage as judged by competitive growth experiments and measurements of burst size. The physical formation of these hairpins was shown by restriction digests of single-stranded DNA with BamHI and HpaII. We argue that these secondary structures form in vivo, at least in the single-stranded genome and the polycistronic mRNAs, and were responsible for the observed growth defects.     

The Bacillus subtilis phage phi 29 protein p16.7, involved in phi 29 DNA replication, is a membrane-localized single-stranded DNA-binding protein.

The functional role of the phi 29-encoded integral membrane protein p16.7 in phage DNA replication was studied using a soluble variant, p16.7A, lacking the N-terminal membrane-spanning domain. Because of the protein-primed mechanism of DNA replication, the bacteriophage phi 29 replication intermediates contain long stretches of single-stranded DNA (ssDNA). Protein p16.7A was found to be an ssDNA-binding protein. In addition, by direct and functional analysis we show that protein p16.7A binds to the stretches of ssDNA of the phi 29 DNA replication intermediates. Properties of protein p16.7A were compared with those of the phi 29-encoded single-stranded DNA-binding protein p5. The results obtained show that both proteins have different, non-overlapping functions. The likely role of p16.7 in attaching phi 29 DNA replication intermediates to the membrane of the infected cell is discussed. Homologues of gene 16.7 are present in phi 29-related phages, suggesting that the proposed role of p16.7 is conserved in this family of phages.     

The binding affinity of Ff gene 5 protein depends on the nearest-neighbor composition of the ssDNA substrate

The Ff gene 5 protein (g5p) is considered to be a nonspecific single-stranded DNA binding protein, because it binds cooperatively to and saturates the Ff bacteriophage single-stranded DNA genome and other single-stranded polynucleotides. However, the binding affinity Komega (the intrinsic binding constant times a cooperativity factor) differs by over an order of magnitude for binding to single-stranded polynucleotides such as poly[d(A)] and poly[d(C)]. A polynucleotide that is more stacked, like poly[d(A)], binds more weakly than one that is less stacked, like poly[d(C)]. To test the hypothesis that DNA base stacking, a nearest-neighbor property, is involved in the binding affinity of the Ff g5p for different DNA sequences, Komega values were determined as a function of NaCl concentration for binding to six synthetic sequences 48 nucleotides in length: dA48, dC48, d(AAC)16, d(ACC)16, d(AACC)12, and d(AAACC)9A3. The binding affinities of the protein for these sequences were indeed found to be related to the nearest-neighbor compositions of the sequences, rather than to simple base compositions. That is, the g5p binding site, which is spanned by four nucleotides, discriminates among these sequences on the basis of the relative numbers of nearest neighbors (AA, CC, and AC plus CA) in the sequence. The results support the hypothesis that the extent of base stacking/unstacking of the free, nonbound ssDNA plays an important role in the binding affinity of the Ff gene 5 protein.     

Selection of genomic sequences that bind tightly to Ff gene 5 protein: primer-free genomic SELEX

Single-stranded DNA or RNA libraries used in SELEX experiments usually include primer-annealing sequences for PCR amplification. In genomic SELEX, these fixed sequences may form base pairs with the central genomic fragments and interfere with the binding of target molecules to the genomic sequences. In this study, a method has been developed to circumvent these artificial effects. Primer-annealing sequences are removed from the genomic library before selection with the target protein and are then regenerated to allow amplification of the selected genomic fragments. A key step in the regeneration of primer-annealing sequences is to employ thermal cycles of hybridization-extension, using the sequences from unselected pools as templates. The genomic library was derived from the bacteriophage fd, and the gene 5 protein (g5p) from the phage was used as a target protein. After four rounds of primer-free genomic SELEX, most cloned sequences overlapped at a segment within gene 6 of the viral genome. This sequence segment was pyrimidine-rich and contained no stable secondary structures. Compared with a neighboring genomic fragment, a representative sequence from the family of selected sequences had about 23-fold higher g5p-binding affinity. Results from primer-free genomic SELEX were compared with the results from two other genomic SELEX protocols.     

Ultrafast fluorescence decay profiles reveal differential unstacking of 2-aminopurine from neighboring bases in single-stranded DNA-binding protein subsites

Gene 5 protein (g5p) is a dimeric single-stranded DNA-binding protein encoded by Ff strains of Escherichia coli bacteriophages. The 2-fold rotationally symmetric binding sites of a g5p dimer each bind to four nucleotides, and the dimers bind with high cooperativity to saturate antiparallel single-stranded DNA (ssDNA) strands. Ultrafast time-resolved fluorescence spectroscopies were used to investigate the conformational heterogeneity and dynamics of fluorescent 2-aminopurine (2AP) labels sequestered by bound g5p. The 2AP labels were positioned within the noncomplementary antiparallel tail sequences of d(AC)(8) or d(AC)(9) of hairpin constructs so that each fluorescent label could probe a different subsite location within the DNA-binding site of g5p. Circular dichroism and isothermal calorimetric titrations yielded binding stoichiometries of approximately six dimers per oligomer hairpin when tails were of these lengths. Mobility shift assays demonstrated the formation of a single type of g5p-saturated complex. Femtosecond time-resolved fluorescence spectroscopy showed that the 2AP in the free (non-protein-bound) DNAs had similar heterogeneous distributions of conformations. However, there were significant changes, dominated by a large increase in the population of unstacked bases from ~22 to 59-68%, depending on their subsite locations, when the oligomers were saturated with g5p. Anisotropy data indicated that 2AP in the bound state was less flexible than in the free oligomer. A control oligomer was labeled with 2AP in the loop of the hairpin and showed no significant change in its base stacking upon g5p binding. A proposed model summarizes the data.     

Interactions of hairpin oligo-2'-O-methylribonucleotides containing methylphosphonate linkages with HIV TAR RNA

Methylphosphonate-modified oligo-2'-O-methylribonucleotides 15-20 nucleotides (nt) in length were prepared whose sequences are complementary to the 5' and 3' sides of the upper hairpin of HIV trans-acting response element (TAR) RNA. These anti-TAR oligonucleotides (ODNs) form stable hairpins whose melting temperatures (Tm) range from 55 degrees C to 80 degrees C. Despite their rather high thermal stabilities, the hairpin oligo-2'-O-methylribonucleotides formed very stable complexes with TAR RNA, with dissociation constants in the nanomolar concentration range at 37 degrees C. The affinities of the hairpin oligomers for TAR RNA were influenced by the positions of the methylphosphonate linkages. The binding affinity was reduced approximately 17-fold by the presence of two methylphosphonate linkages in the TAR loop complementary region (TLCR) of the oligomer, whereas methylphosphonate linkages outside this region increased binding affinity approximately 3-fold. The configurations of the methylphosphonate linkages in the TLCR also affected binding affinity, with the RpRp isomer showing significantly higher binding than the SpSp isomer. In addition to serving as probes of the interactions between the oligomer and TAR RNA, the presence of the methylphosphonate linkages in combination with the hairpin structure increases the resistance of these oligomers to degradation by exonucleases found in mammalian serum. The combination of high binding affinity and nuclease resistance of the hairpin ODNs containing methylphosphonate linkages suggests their potential utility as antisense compounds.     

The Ff gene 5 single-stranded DNA-binding protein binds to the transiently folded form of an intramolecular G-quadruplex

The Ff gene 5 protein (g5p) is classified as a single-stranded DNA-binding protein. However, we previously showed that g5p binds with high affinity to a SELEX-selected G-rich 58-mer DNA oligomer, I-3, that forms an intramolecular G-quadruplex [Wen, J.-D., Gray, C. W., and Gray, D. M. (2001) Biochemistry 40, 9300-9310]. In 200 mM NaCl at 37 degrees C, g5p binds to I-3 in two stages, the first stage being the formation of a discrete intermediate complex that appears to be a precursor to a saturated g5p x I-3 complex. For the present paper, CD spectroscopy and DMS methylation techniques were used to investigate the binding of g5p to the I-3 oligomer and to the truncated 26-nucleotide core of the I-3 oligomer. The core sequence, called I-3c26, was d(GGGGTCAGGCTGGGGTTGTGCAGGTC). Results were the following: (1) The g5p binds in one stage to I-3c26 in 200 mM NaCl at 37 degrees C. (2) The intermediate complex of g5p.I-3 is formed by the binding of g5p to the core sequence. (3) G-quadruplex structures are maintained in both the g5p x I-3 and g5p x I-3c26 complexes, but the bound G-quadruplex structures are altered from their respective steady-state folded forms in 200 mM NaCl. (4) CD kinetics measurements showed that the I-3c26 quadruplex folds in two stages and that a transiently folded form is apparently the same as the altered structure to which g5p binds. (5) DMS methylation protection and interference experiments identified two guanines that are differentially involved in the steady-state folded and g5p-bound G-quadruplex structures. A model for a possible I-3c26 G-quadruplex structure is described.     

Selective recognition of a DNA G-quadruplex by an engineered antibody

Particular guanine rich nucleic acid sequences can fold into stable secondary structures called G-quadruplexes. These structures have been identified in various regions of the genome that include the telomeres, gene promoters and UTR regions, raising the possibility that they may be associated with biological function(s). Computational analysis has predicted that intramolecular G-quadruplex forming sequences are prevalent in the human genome, thus raising the desire to differentially recognize genomic G-quadruplexes. We have employed antibody phage display and competitive selection techniques to generate a single-chain antibody that shows >1000-fold discrimination between G-quadruplex and duplex DNA, and furthermore >100-fold discrimination between two related intramolecular parallel DNA G-quadruplexes. The amino acid sequence composition at the antigen binding site shows conservation within the light and heavy chains of the selected scFvs, suggesting sequence requirements for G-quadruplex recognition. Circular dichroism (CD) spectroscopic data showed that the scFv binds to the prefolded G-quadruplex and does not induce G-quadruplex structure formation. This study demonstrates the strongest discrimination that we are aware of between two intramolecular genomic G-quadruplexes.     

Hairpin formation in Tn5 transposition

The initial chemical steps in Tn5 transposition result in blunt end cleavage of the transposon from the donor DNA. We demonstrate that this cleavage occurs via a hairpin intermediate. The first step is a 3' hydrolytic nick by transposase. The free 3'OH then attacks the phosphodiester bond on the opposite strand, forming a hairpin at the transposon end. In addition to forming precise hairpins, Tn5 transposase can form imprecise hairpins. This is the first example of imprecise hairpin formation on transposon end DNA. To undergo strand transfer, the hairpin must to be resolved by a transposase-catalyzed hydrolytic cleavage. We show that both precise and imprecise hairpins are opened by transposase. A transposition mechanism utilizing a hairpin intermediate allows a single transposase active site to cleave both 3' and 5' strands without massive protein/DNA rearrangements.     

The putative single-stranded DNA-binding protein of the filamentous bacteriophage, Ifl. Amino acid sequence of the protein and structure of the gene.

The protein product corresponding to the gene located in the region of the coliphage Ifl genome shown to contain the code for the single-stranded DNA (ssDNA)-binding proteins of all filamentous phages so far studied has been isolated from infected bacterial cells and its amino acid sequence determined. The mature protein contains 95 amino acids (calculated molecular mass 10553 Da). Its sequence corresponds to that predicted from the DNA sequence but lacks the initiating methionine residue. Although there is little direct sequence homology between the phage Ifl protein and the ssDNA-binding proteins of the other filamentous phages that have been studied, computer-based comparisons of various physical and structural parameters showed that the phage Ifl protein contains a domain that is closely related to domains in the coliphage T4 gene 32 protein and the Pseudomonas phage Pfl ssDNA-binding protein and suggest that the Ifl protein does have a ssDNA-binding function although we were unable to show this directly.     

The effects of loop size on Sac7d-hairpin DNA interactions

Hairpin structure is a common feature of DNA molecules. They are located near functional loci, such as regulation and promotion sites, as well as in cruciform structures, and they provide potential binding sites for endogenous proteins. The effects of different hairpin loops that are composed of one to five thymidines, designated as L1-L5, and have a common self-complementary stem, CTATATAG, on the interactions with Sac7d were studied. In thermostability studies, Sac7d stabilized a tetra-loop hairpin DNA and hairpin DNA with GTTC tetra-loop regions better than it stabilized tri- and penta-loops. Circular dichroism (CD) spectra showed that hairpins retained primarily a B-type conformation upon Sac7d binding. Intermolecular interactions between hairpins were likely decreased, due to the Sac7d-induced kinks, as shown by an increase at 220nm in the CD spectra. Surface plasmon resonance (SPR) observations suggested that the rates of Sac7d binding to hairpin DNA depend on the loop size of the hairpin duplexes. At a fixed stem length, Sac7d binds to tetra-loop hairpin DNA duplexes with a higher association rate and lower dissociation rate, compared with their tri- and penta-loop counterparts. In addition, the tri-loop and GTC tri-loop hairpin DNA had lower affinity for Sac7d because of the smaller and tighter loop size. Our study indicates that Sac7d binding affinity to hairpin DNA is primarily determined by loop size and stem integrity, and the results presented here provide a model for studies concerning other minor groove DNA-binding proteins that kink hairpin DNA.     

Effects of single-base bulges on intercalator binding to small RNA and DNA hairpins and a ribosomal RNA fragment

The way in which a single-base bulge might affect the structure of an RNA helix has been examined by preparing a series of six RNA hairpins, all with seven base pairs and a four-nucleotide loop. Five of the hairpins have single-base bulges at different positions. The intercalating cleavage reagent (methidiumpropyl)-EDTA-Fe(II) [MPE-Fe(II)] binds preferentially at a CpG sequence in the helix lacking a bulge and in four of the five hairpins with bulges. Hairpins with a bulge one or two bases to the 3' side of the CpG sequence bind ethidium 4-5-fold more strongly than the others. V1 RNase, which is sensitive to RNA backbone conformation in helices, detects a conformational change in all of the helices when ethidium binds; the most dramatic changes, involving the entire hairpin stem, are in one of the two hairpins with enhanced ethidium affinity. Only a slight conformational change is detected in the hairpin lacking a bulge. A bulge adjacent to a CpG sequence in a 100-nucleotide ribosomal RNA fragment enhances MPE-Fe(II) binding by an order of magnitude. These results extend our previous observations of bulges at a single position in an RNA hairpin [White, S. A., & Draper, D.E. (1987) Nucleic Acids Res. 15, 4049] and show that (1) a structural change in an RNA helix may be propagated for several base pairs, (2) bulges tend to increase the number of conformations available to a helix, and (3) the effects observed in small RNA hairpins are relevant to larger RNAs with more extensive structure. A bulge in a DNA hairpin identical in sequence with the RNA hairpins does not enhance MPE-Fe(II) binding affinity, relative to a control DNA hairpin. The effects of bulges on ethidium intercalation are evidently modulated by helix structure.     

Structure of nuclear pre-mRNA. VI. "Reversed repeats" in animal DNA and their hybridization with double-stranded regions of pre-mRNA.

About 2% of mouse DNA reassociates at extremely low Cot values (10-7-10-6 mole-liter-1-sec). This DNA fraction was isolated with the aid of nuclease S1 and purified by chromatography on hydroxyapatite. The study of this fraction suggests that it has a structure of hairpin type. The DNA of the hairpins can hybridize with sequences forming the double-stranded regions in pre-mRNA. Probably at least some of the DNA of the hairpins, described as "reversed repeating sequences" of DNA, is transcribed with the formation of structures of hairpin type in pre-mRNA.     

Head-to-head bis-hairpin polyamide minor groove binders and their conjugates with triplex-forming oligonucleotides: studies of interaction with target double-stranded DNA

Two hairpin hexa(N-methylpyrrole)carboxamide DNA minor groove binders (MGB) were linked together via their N-termini in head-to-head orientation. Complex formation between these bis-MGB conjugates and target DNA has been studied using DNase I footprinting, circular dichroism, thermal dissociation, and molecular modeling. DNase I footprint revealed binding of these conjugates to all the sites of 492 b.p. DNA fragment containing (A/T)(n)X(m)(A/T)(p) sequences, where n>3, p>3; m=1,2; X = A,T,G, or C. Binding affinity depended on the sequence context of the target. CD experiments and molecular modeling showed that oligo(N-methylpyrrole)carboxamide moieties in the complex form two short antiparallel hairpins rather than a long parallel head-to-head hairpin. Binding of bis-MGB also stabilized a target duplex thermodynamically. Sequence specificity of bis-MGB/DNA binding was validated using bis-conjugates of sequence-specific hairpin (N-methylpyrrole)/(N-methylimidazole) carboxamides. In order to increase the size of recognition sequence, the conjugates of bis-MGB with triplex-forming oligonucleotides (TFO) were synthesized and compared to TFO conjugated with single MGB hairpin unit. Bis-MGB-oligonucleotide conjugates also bind to two blocks of three and more A.T/T.A pairs similarly to bis-MGB alone, independently of the oligonucleotide moiety, but with lower affinity. However, the role of TFO in DNA recognition was demonstrated for mono-MGB-TFO conjugate where the binding was detected mainly in the area of the target sequence consisting of both MGB and TFO recognition sites. Basing on the molecular modeling, three-dimensional models of both target DNA/bis-MGB and target DNA/TFO-bis-MGB complexes were built, where bis-MGB forms two antiparallel hairpins. According to the second model, one MGB hairpin is in the minor groove of 5'-adjacent A/T sequence next to the triplex-forming region, whereas the other one occupies the minor groove of the TFO binding polypurine tract. All these data together give a key information for the construction of MGB-MGB and MGB-oligonucleotide conjugates possessing high specificity and affinity for the target double-stranded DNA.