Search

Sleeping Beauty(SB) is the most active Tc1/mariner-like transposon in vertebrate species. Each of the terminal inverted repeats (IRs) of SB contains two transposase-binding sites (DRs). This feature, termed the IR/DR structure, is conserved in a group of Tc1-like transposons. The DNA-binding region of SB transposase, similar to the paired domain of Pax proteins, consists of two helix-turn-helix subdomains (PAI + RED = PAIRED). The N-terminal PAI subdomain was found to play a dominant role in contacting the DRs. Transposase was able to bind to mutant sites retaining the 3′ part of the DRs; thus, primary DNA binding is not sufficient to determine the specificity of the transposition reaction. The PAI subdomain was also found to bind to a transpositional enhancer-like sequence within the left IR of SB, and to mediate protein-protein interactions between transposase subunits. A tetrameric form of the transposase was detected in solution, consistent with an interaction between the IR/DR structure and a transposase tetramer. We propose a model in which the transpositional enhancer and the PAI subdomain stabilize complexes formed by a transposase tetramer bound at the IR/DR. These interactions may result in enhanced stability of synaptic complexes, which might explain the efficient transposition of Sleeping Beautyin vertebrate cells.

Ahnak is a ubiquitously expressed giant protein of 5643 amino acids implicated in cell differentiation and signal transduction. In a recent study, we demonstrated the association of ahnak with the regulatory β2 subunit of the cardiac L‐type Ca2+channel. Here we identify the most carboxyl‐terminal ahnak region (aa 5262–5643) to interact with recombinant β2a as well as with β2 and β1a isoforms of native muscle Ca2+channels using a panel of GST fusion proteins. Equilibrium sedimentation analysis revealed Kdvalues of 55 ± 11 nM and 328 ± 24 nM for carboxyl‐terminal (aa 195–606) and amino‐terminal (aa 1–200) truncates of the β2a subunit, respectively. The same carboxylterminal ahnak region (aa 5262–5643) bound to G‐actin and cosedimented with F‐actin. Confocal microscopy of human left ventricular tissue localized the carboxylterminal ahnak portion to the sarcolemma including the T‐tubular system and the intercalated disks of cardiomyocytes. These results suggest that ahnak provides a structural basis for the subsarcolemmal cytoarchitecture and confers the regulatory role of the actin‐based cytoskeleton to the L‐type Ca2+channel.—Hohaus, A., Person, V., Behlke, J., Schaper, J., Morano, I., Haase, H. The carboxyl‐terminal region of ahnak provides a link between cardiac L‐type Ca2+channels and the actin‐based cytoskeleton. FASEB J. 16, 1205–1216 (2002)

At the transcriptional level, the pSM19035‐encoded ω protein coordinates the expression of proteins required for control of copy number and maintenance of plasmids. Using circular dichroism, fluorescence spectroscopy, ultracentrifugation and an electrophoretic mobility shift assay, the wild‐type ω protein and a variant with a C‐terminal hexa‐histidine tag (ω‐H6) were characterized. The ω protein is mainly α‐helical (42%), occurs as homodimer in solution, unfolds thermally with half transition temperatures, Tm, between ∼43 and ∼78°C depending on the ionic strength of the buffer, and binds PcopS‐DNA with high affinity. The ω‐H6protein has a modified conformation with lower α‐helix content (29%), lower thermal stability, and strongly reduced affinity to PcopS‐DNA.

At the transcriptional level, the pSM19035-encoded ω protein coordinates the expression of proteins required for control of copy number and maintenance of plasmids. Using circular dichroism, fluorescence spectroscopy, ultracentrifugation and an electrophoretic mobility shift assay, the wild-type ω protein and a variant with a C-terminal hexa-histidine tag (ω-H6) were characterized. The ω protein is mainly α-helical (42%), occurs as homodimer in solution, unfolds thermally with half transition temperatures, Tm, between ∼43 and ∼78°C depending on the ionic strength of the buffer, and binds PcopS-DNA with high affinity. The ω-H6protein has a modified conformation with lower α-helix content (29%), lower thermal stability, and strongly reduced affinity to PcopS-DNA.

The first crystal structure of a protein, the Zα high affinity binding domain of the RNA editing enzyme ADAR1, bound to left-handed Z-DNA was recently described. The essential set of residues determined from this structure to be critical for Z-DNA recognition was used to search the database for other proteins with the potential for Z-DNA binding. We found that the tumor-associated protein DLM-1 contains a domain with remarkable sequence similarities to ZαADAR. Here we report the crystal structure of this DLM-1 domain bound to left-handed Z-DNA at 1.85 Å resolution. Comparison of Z-DNA binding by DLM-1 and ADAR1 reveals a common structure-specific recognition core within the binding domain. However, the domains differ in certain residues peripheral to the protein–DNA interface. These structures reveal a general mechanism of Z-DNA recognition, suggesting the existence of a family of winged-helix proteins sharing a common Z-DNA binding motif.

ABSTRACTThe binary iota-toxin is produced by Clostridium perfringenstype E strains and consists of two separate proteins, the binding component iota b (98 kDa) and an actin-ADP-ribosylating enzyme component iota a (47 kDa). Iota b binds to the cell surface receptor and mediates the translocation of iota a into the cytosol. Here we studied the cellular uptake of iota-toxin into Vero cells. Bafilomycin A1, but not brefeldin A or nocodazole, inhibited the cytotoxic effects of iota-toxin, indicating that toxin is translocated from an endosomal compartment into the cytoplasm. Acidification (pH ≤ 5.0) of the extracellular medium enabled iota a to directly enter the cytosol in the presence of iota b. Activation by chymotrypsin induced oligomerization of iota b in solution. An average mass of 530 ± 28 kDa for oligomers was determined by analytical ultracentrifugation, indicating heptamer formation. The entry of iota-toxin into polarized CaCo-2 cells was studied by measuring the decrease in transepithelial resistance after toxin treatment. Iota-toxin led to a significant decrease in resistance when it was applied to the basolateral surface of the cells but not following application to the apical surface, indicating a polarized localization of the iota-toxin receptor.

Analytical ultracentrifugation was used to determine the molecular mass, M, of hexameric DNA‐helicase RepA at pH 5.8 and 7.6. At pH 7.6, a molecular mass of 179.5±2.6 kDa was found, consistent with the known hexameric state of RepA, (RepA)6. At pH 5.8, (RepA)6associates to form a dimer with a molecular mass of 366.2±4.1 kDa. Analytical ultracentrifugation was also applied to characterize the interaction of single‐stranded DNA (ssDNA) with the two different oligomeric states of (RepA)6at pH 5.8 and 7.6. The dissociation constants, Kd, for the equilibrium binding of (dA)30to the (RepA)6dimer at pH 5.8 and to (RepA)6at pH 7.6 were determined at 10°C in the presence of 0.5 mM ATPγS, 10 mM MgCl2and 60 mM NaCl as Kd5.8=0.94±0.13 μM at pH 5.8 and Kd7.6=25.4±6.4 μM at pH 7.6. The stoichiometries, n, for the two complexes (dA)30/(RepA)6dimer and (dA)30/(RepA)6at pH 5.8 and 7.6 were calculated from the corresponding binding curves. At pH 5.8 one (dA)30molecule was bound per (RepA)6dimer, while at pH 7.6 one (dA)30molecule was bound to one (RepA)6. Binding curves were compatible with a single ssDNA binding site present on the (RepA)6dimer and on (RepA)6, respectively, with no indication of cooperativity. (RepA)6tends to form larger aggregates under acidic conditions (pH<6.0) which are optimal for ssDNA binding. In contrast, at pH 5.8 in the presence of 60 mM NaCl, only the (RepA)6dimer was observed both in the absence and presence of (dA)30.

EcoRII is a type IIE restriction endonuclease characterized by a highly cooperative reaction mechanism that depends on simultaneous binding of the dimeric enzyme molecule to two copies of its DNA recognition site. Transmission electron microscopy provided direct evidence that EcoRII mediates loop formation of linear DNA containing two EcoRII recognition sites. Specific DNA binding of EcoRII revealed a symmetrical DNase I footprint occupying 16–18 bases. Single amino acid replacement of Val258by Asn yielded a mutant enzyme that was unaffected in substrate affinity and DNase I footprinting properties, but exhibited a profound decrease in cooperative DNA binding and cleavage activity. Because the electrophoretic mobility of the mutant enzyme-DNA complexes was significantly higher than that of the wild-type, we investigated if mutant V258N binds as a monomer to the substrate DNA. Analysis of the molecular mass of mutant V258N showed a high percentage of protein monomers in solution. The dissociation constant of mutant V258N confirmed a 350-fold decrease of the enzyme dimerization capability. We conclude that Val258is located in a region of EcoRII involved in homodimerization. This is the first report of a specific amino acid replacement in a restriction endonuclease leading to the loss of dimerization and DNA cleavage while retaining specific DNA binding.

The actin-ADP-ribosylating binaryClostridium botulinumC2 toxin consists of two individual proteins, the binding/translocation component C2II and the enzyme component C2I. To elicit its cytotoxic action, C2II binds to a receptor on the cell surface and mediates cell entry of C2I via receptor-mediated endocytosis. Here we report that binding of C2II to the surface of target cells requires cleavage of C2II by trypsin. Trypsin cleavage causes oligomerization of the activated C2II (C2IIa) to give SDS-stable heptameric structures, which exhibit a characteristic annular or horseshoe shape and form channels in lipid bilayer membranes. Cytosolic delivery of the enzyme component C2I is blocked by bafilomycin but not by brefeldin A or nocodazole, indicating uptake from an endosomal compartment and requirement of endosomal acidification for cell entry. In the presence of C2IIa and C2I, short term acidification of the extracellular medium (pH 5.4) allows C2I to enter the cytosol directly. Our data indicate that entry of C2 toxin into cells involves (i) activation of C2II by trypsin-cleavage, (ii) oligomerization of cleaved C2IIa to heptamers, (iii) binding of the C2IIa oligomers to the carbohydrate receptor on the cell surface and assembly with C2I, (iv) receptor-mediated endocytosis of both C2 components into endosomes, and finally (v) translocation and release of C2I into the cytosol after acidification of the endosomal compartment.

Plasmid pIP501 encoded transcriptional repressor CopR is one of the two regulators of plasmid copy number. It acts as a transcriptional repressor at the essential repRpromoter. Furthermore, CopR prevents convergent transcription from the repRand the antisense promoter, thereby indirectly increasing the amount of antisense‐RNA, the second regulatory component. CopR binds as a dimer to a nearly palindromic operator with the consensus sequence 5`CGTG. Previously, a CopR structural model was built and used to identify amino acids involved in DNA binding. These data showed that CopR is a HTH protein belonging to the lambda repressor superfamily and allowed the identification of two amino acids involved in specific DNA recognition. Here, we describe site‐directed mutagenesis in combination with EMSA, dimerization studies using sedimentation equilibrium, and CD measurements to verify the model predictions concerning amino acids involved in dimerization. With this approach, the dimeric interface could be located between amino acids I44 and L62. F5 located at the N‐terminus is additionally required for proper folding, and could, therefore, not be unequivocally assigned to the dimeric interface. CD measurements at protein concentrations well below KDimerrevealed that the monomer of CopR is folded. Proteins 2000;39:408–416. © 2000 Wiley‐Liss, Inc.

Plasmid pIP501 encoded transcriptional repressor CopR is one of the two regulators of plasmid copy number. It acts as a transcriptional repressor at the essential repR promoter. Furthermore, CopR prevents convergent transcription from the repR and the antisense promoter, thereby indirectly increasing the amount of antisense-RNA, the second regulatory component. CopR binds as a dimer to a nearly palindromic operator with the consensus sequence 5`CGTG. Previously, a CopR structural model was built and used to identify amino acids involved in DNA binding. These data showed that CopR is a HTH protein belonging to the lambda repressor superfamily and allowed the identification of two amino acids involved in specific DNA recognition. Here, we describe site-directed mutagenesis in combination with EMSA, dimerization studies using sedimentation equilibrium, and CD measurements to verify the model predictions concerning amino acids involved in dimerization. With this approach, the dimeric interface could be located between amino acids I44 and L62. F5 located at the N-terminus is additionally required for proper folding, and could, therefore, not be unequivocally assigned to the dimeric interface. CD measurements at protein concentrations well below KDimer revealed that the monomer of CopR is folded. Proteins 2000;39:408–416. © 2000 Wiley-Liss, Inc.

Analytical ultracentrifugation studies indicated that the C‐terminal domains of IF2 comprising amino acid residues 520–741 (IF2 C) and 632–741 (IF2 C‐2) bind fMet‐tRNA with similar affinities (Kdat 25°C equal to 0.27 and 0.23 μM, respectively). Complex formation between fMet‐tRNAfMetand IF2 C or IF2 C‐2 is accompanied by barely detectable spectral changes as demonstrated by a comparison of the Raman spectra of the complexes with the calculated sum of the spectra of the individual components. These results and the temperature dependence of the Kdof the protein–RNA complexes indicate that complex formation is not accompanied by obvious conformational changes of the components, and possibly depends on a rather small binding site comprising only a few interacting residues of both components.

The plasmid pIP501 encoded transcriptional repressor CopR is one of the two regulators of plasmid copy number. CopR binds as a dimer to a nearly palindromic operator with the consensus sequence 5`-CGTG. Intermediate sequence searches revealed a significant structural relationship between CopR and the bacteriophage P22 c2 and the 434 c1 repressors. In this report we describe the experimental verification of a CopR homology model, which is based on a fairly low-sequence identity of 13.8 % to P22 c2 repressor. A model for the complex of CopR with the deoxyribonucleic acid (DNA) target was built on the basis of experimental footprinting data, the above-mentioned CopR homology model, and the crystal structure of the 434 c1 repressor-DNA complex. Site-directed mutagenesis was used to test the function of amino acids involved in sequence and nonsequence-specific DNA recognition and amino acids important for correct protein folding. CD measurements were performed to detect structural changes caused by the mutations. Exchanges of residues responsible for sequence-specific DNA recognition reduced binding to a nonspecific level. Mutations of amino acids involved in nonspecific DNA binding lead to decreased binding affinity while maintaining selectivity. Substitution of amino acids necessary for proper folding caused dramatic structural changes. The experimental data support the model of CopR as a helix-turn-helix protein belonging to the λ repressor superfamily. Proteins 2000;38:393–406. © 2000 Wiley-Liss, Inc.

The plasmid pIP501 encoded transcriptional repressor CopR is one of the two regulators of plasmid copy number. CopR binds as a dimer to a nearly palindromic operator with the consensus sequence 5`‐CGTG. Intermediate sequence searches revealed a significant structural relationship between CopR and the bacteriophage P22 c2 and the 434 c1 repressors. In this report we describe the experimental verification of a CopR homology model, which is based on a fairly low‐sequence identity of 13.8 % to P22 c2 repressor. A model for the complex of CopR with the deoxyribonucleic acid (DNA) target was built on the basis of experimental footprinting data, the above‐mentioned CopR homology model, and the crystal structure of the 434 c1 repressor‐DNA complex. Site‐directed mutagenesis was used to test the function of amino acids involved in sequence and nonsequence‐specific DNA recognition and amino acids important for correct protein folding. CD measurements were performed to detect structural changes caused by the mutations. Exchanges of residues responsible for sequence‐specific DNA recognition reduced binding to a nonspecific level. Mutations of amino acids involved in nonspecific DNA binding lead to decreased binding affinity while maintaining selectivity. Substitution of amino acids necessary for proper folding caused dramatic structural changes. The experimental data support the model of CopR as a helix‐turn‐helix protein belonging to the λ repressor superfamily. Proteins 2000;38:393–406. © 2000 Wiley‐Liss, Inc.

The plasmid pIP501 encoded transcriptional repressor CopR is one of the two regulators of plasmid copy number. CopR binds as a dimer to a nearly palindromic operator with the consensus sequence 5`-CGTG. Intermediate sequence searches revealed a significant structural relationship between CopR and the bacteriophage P22 c2 and the 434 c1 repressors. In this report we describe the experimental verification of a CopR homology model, which is based on a fairly low-sequence identity of 13.8 % to P22 c2 repressor. A model for the complex of CopR with the deoxyribonucleic acid (DNA) target was built on the basis of experimental footprinting data, the above-mentioned CopR homology model, and the crystal structure of the 434 c1 repressor-DNA complex. Site-directed mutagenesis was used to test the function of amino acids involved in sequence and nonsequence-specific DNA recognition and amino acids important for correct protein folding. CD measurements were performed to detect structural changes caused by the mutations. Exchanges of residues responsible for sequence-specific DNA recognition reduced binding to a nonspecific level. Mutations of amino acids involved in nonspecific DNA binding lead to decreased binding affinity while maintaining selectivity. Substitution of amino acids necessary for proper folding caused dramatic structural changes. The experimental data support the model of CopR as a helix-turn-helix protein belonging to the λ repressor superfamily. Proteins 2000;38:393–406. © 2000 Wiley-Liss, Inc.

The plasmid pIP501 encoded transcriptional repressor CopR is one of the two regulators of plasmid copy number. CopR binds as a dimer to a nearly palindromic operator with the consensus sequence 5`-CGTG. Intermediate sequence searches revealed a significant structural relationship between CopR and the bacteriophage P22 c2 and the 434 c1 repressors. In this report we describe the experimental verification of a CopR homology model, which is based on a fairly low-sequence identity of 13.8 % to P22 c2 repressor. A model for the complex of CopR with the deoxyribonucleic acid (DNA) target was built on the basis of experimental footprinting data, the above-mentioned CopR homology model, and the crystal structure of the 434 c1 repressor-DNA complex. Site-directed mutagenesis was used to test the function of amino acids involved in sequence and nonsequence-specific DNA recognition and amino acids important for correct protein folding. CD measurements were performed to detect structural changes caused by the mutations. Exchanges of residues responsible for sequence-specific DNA recognition reduced binding to a nonspecific level. Mutations of amino acids involved in nonspecific DNA binding lead to decreased binding affinity while maintaining selectivity. Substitution of amino acids necessary for proper folding caused dramatic structural changes. The experimental data support the model of CopR as a helix-turn-helix protein belonging to the λ repressor superfamily. Proteins 2000;38:393–406. © 2000 Wiley-Liss, Inc.

The Zα domain of the human RNA editing enzyme double‐stranded RNA deaminase I (ADAR1) binds to left‐handed Z‐DNA with high affinity. We found by analytical ultracentrifugation and CD spectroscopy that two Zα domains bind to one d(CG)3T4(CG)3hairpin which contains a stem of six base pairs in the Z‐DNA conformation. Both wild‐type Zα and a C125S mutant show a mean dissociation constant of 30 nM as measured by surface plasmon resonance and analytical ultracentrifugation. Our data suggest that short (≥6 bp) segments of Z‐DNA within a gene are able to recruit two ADAR1 enzymes to that particular site.

The size and shape of the recombinant murine small heat shock protein, hsp25, have been analyzed by hydrodynamic and electron microscopic methods. According to these studies recombinant hsp25 exists in large complexes with a sphere-like shape and diameters of 15–18 nm. The molecular mass of these complexes amounts to about 730 kDa indicating that they are composed of about 32 monomers.

The Zα domain of the human RNA editing enzyme double-stranded RNA deaminase I (ADAR1) binds to left-handed Z-DNA with high affinity. We found by analytical ultracentrifugation and CD spectroscopy that two Zα domains bind to one d(CG) 3T 4(CG) 3hairpin which contains a stem of six base pairs in the Z-DNA conformation. Both wild-type Zα and a C125S mutant show a mean dissociation constant of 30 nM as measured by surface plasmon resonance and analytical ultracentrifugation. Our data suggest that short (≥6 bp) segments of Z-DNA within a gene are able to recruit two ADAR1 enzymes to that particular site.

The size and shape of the recombinant murine small heat shock protein, hsp25, have been analyzed by hydrodynamic and electron microscopic methods. According to these studies recombinant hsp25 exists in large complexes with a sphere‐like shape and diameters of 15–18 nm. The molecular mass of these complexes amounts to about 730 kDa indicating that they are composed of about 32 monomers.

Pyridoxal phosphate, cofactor of several enzymes, possesses linking properties to induce oligomerization of identical proteins such as adrenodoxin-reductase or adrenodoxin. The capability to get such self-assemblies is slightly lower than that for obtaining the heterologous complex between adrenodoxin-reductase and adrenodoxin which was found to be essential for electron transfer in the cytochrome P450 system. The influence of pyridoxal phosphate on the complex formation between adrenodoxin-reductase and adrenodoxin as well as the oligomerization reaction of the isolated proteins and possible consequences is discussed.

The CopR protein is one of the two regulators of pIP501 copy number. It acts as transcriptional repressor at the essential repR promoter pII. Previously, we found that CopR contacts two consecutive major grooves (site I and site II) on the same face of the DNA. In spite of identical sequence motifs in these sites, neighboring bases were contacted differently. Furthermore, we showed that CopR can dimerize in solution. We demonstrate by two independent methods that CopR binds the DNA as a dimer. We present data that suggest that the sigmoidal CopR-DNA binding curve published previously is the result of two coupled equilibria: dimerization of CopR monomers and CopR dimer-DNA binding. A KD-value of 1.44(±0.49) × 10−6 M for CopR dimers was determined by analytical ultracentrifugation. Based on this value and the binding curve, the equilibrium dissociation constant K2 for the CopR-DNA complex was calculated to be 4(±1.3) × 10−10 M. Quantitative Western blot analysis was used to determine the intracellular concentration of CopR in Bacillus subtilis. This value, 20 × 10−6 to 30 × 10−6 M, is 10 to 20-fold higher than the equilibrium constant for dimer dissociation, suggesting that CopR binds in vivo as a preformed dimer.

‘Mammary-derived growth inhibitor (MDGI)’ is a 14.5 kDa polypeptide with growth-inhibitory activity for various mammary epithelial cells in vitro which is highly homologous to cardiac fatty acid-binding protein (H-FABP). Here we describe a new biological activity of MDGI: Inhibition of L(+)-lactate-, arachidonic acid- and 15-S-hydroxyeicosatetraenoic acid-induced supersensitivity of neonatal rat heart cells for beta-adrenergic stimulation, concerning particularly a small population of beta2 receptors. Synthetic peptides corresponding to the MDGI-sequence, residue 121–131 mimic the effect of MDGI. Measurements of lipid-binding to MDGI and synthetic peptides excluded the binding of arachidonic acid, 15-S-hydroxyeicosatetraenoic acid or beta-adrenergic agonists to MDGI or the peptides as the mechanism for this effect. Also, no direct interference of MDGI and the synthetic peptides with the binding of the beta-adrenergic agent CGP 12177 to its receptor on A431 cells could be detected.

The amino‐teminal laminin G‐like domain of human sex hormone‐binding globulin (SHBG), which contains the steroid‐binding site and the dimerization domain, has been produced in Escherichia coli, purified to homogeneity and crystallized in complex with 5α‐­dihydrotestosterone (DHT) in two different crystal forms. Native data sets have been collected for tetragonal crystals (space group P4122 or P4322; unit‐cell parameters a= 52.2, c= 148.4 Å) diffracting to 3.3 Å and trigonal crystals (R32; a= 104.0, c= 84.4 Å) diffracting to better than 1.6 Å. Since both crystal forms can only accommodate a single monomer in the asymmetric unit and share twofold rotational symmetry, it is proposed that the homodimer of this truncated form of SHBG, as observed in ultracentrifugation experiments, displays C2point‐group symmetry.

Sedimentation experiments on cytochrome P450cam (CYP101) has been performed to compare the molecular mass of the protein in the oxidized state and as carbon monoxide complex. The oxidized protein in the absence of β-mercaptoethanol is a dimer with a molecular mass of 92 kDa. Addition of mercaptoethanol avoids completely the dimerization. Dithionite reduced P450cam in the presence of carbon monoxide has been found to be a monomeric protein.

The 24mer deoxyoligonucleotide 3′-d(T)10-5′-5′-d(C)4-d(A)10-3′ (psC4) with an uncommon 5′-p-5′ phosphodiester linkage was designed to enable the formation of a hairpin structure with unusual parallel-stranded stem. As reference hairpin structure with an antiparallel-stranded stem, the 24mer 5′-d(T)10d(C)4-d(A)10-3′ (apsC4) was chosen. The behaviour of these oligonucleotides at different temperatures, DNA and salt concentrations was characterised by a combination of UV melting, CD, CD melting, infrared and Raman spectroscopy, infrared melting and analytical ultracentrifugation. The parallel-stranded hairpin structure was found to be formed by psC4 only under conditions of low DNA concentration and low salt concentration. Increase of the NaCl concentration beyond the physiological level or high DNA concentration supports the formation of intermolecular multi-stranded structures. The experimental data are in agreement with a four-stranded complex formed by two molecules of psC4. The base pairing model of this asymmetric four-stranded complex is based on the pyrimidine motif of a triple helix with two bifurcated hydrogen bonds at the O4 of the thymine each directed towards one of the amino protons of both adenines. In contrast, the reference oligonucleotide apsC4 forms only an antiparallel-stranded hairpin under all experimental conditions.