Skip Navigation

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (120K) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (21)
Right arrowRequest Permissions
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Mukerji, I
Right arrow Articles by Spiro, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Mukerji, I
Right arrow Articles by Spiro, T.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© 1996 Oxford University Press 5013-5020

Footnote

A UV resonance Raman study of hairpin dimer helices of d(A-G)10 at neutral pH containing intercalated dA residues and alternating dG tetrads

A UV resonance Raman study of hairpin dimer helices of d(A-G) 10 at neutral pH containing intercalated dA residues and alternating dG tetrads Ishita Mukerji+, Mary Claire Shiber1, Jacques R. Fresco1,* and Thomas G. Spiro

Departments of Chemistry and 1Molecular Biology, Princeton University, Princeton, NJ 08544-1014, USA

Received July 16, 1996; Revised and Accepted November 6, 1996

ABSTRACT

The structure of the oligonucleotide d(A-G)10 in 0.6 M Na+, pH 7.0 has been investigated with UV resonance Raman (UVRR) spectroscopy. Variable wavelength excitation was used to distinguish the spectral contributions of dG and dA residues. Both classes of residues show UVRR hyperchromism with increasing temperature, reflecting unstacking of the bases. The dG residues melt relatively cooperatively with a Tm of ~42oC. Unstacking is non-cooperative for the dA residues, increasing linearly between 4 and 80oC. G-tetrads at low temperature are indicated by UVRR frequency shifts of modes associated with C6=O and C2-NH2 of the dG residues, and of vibrations involving N7, all sites of H-bonding. However, there are no indications of interbase H-bonds for the dA residues, showing they do not form H-bonded tetrads. Most of the bases are oriented anti about the glycosyl bond, but at 4oC a fraction of the residues are syn. These results, together with the findings by Shiber et al. [Shiber,M.C., Braswell,E.H., Klump,H. and Fresco,J.R. (1996) Nucleic Acids Res. 24, 5004-5012] that d(A-G)10 under comparable conditions has the molecular weight of a dimer, support a model in which two hairpins interact to form a helical structure with G-tetrads and intercalated dA residues.

INTRODUCTION

Approximately 0.4% of the mammalian genome consists of homopurine.homopyrimidine sequences (2 ) which have been implicated in genomic processes such as transcriptional regulation and recombination (3 ,4 ). Concentration and pH strongly influence the structure of these sequences; e.g. at acidic pH homopurine. homopyrimidine tracts can form intramolecular triplexes (3 ,5 ). Similarly, dG-rich sequences, as in Oxytricha telomere DNA, are thought to form tetraplex structures stabilized by H-bonds that involve the same H-bonding sites on dG residues as in Hoogsteen and Watson-Crick base pairs (Fig. 1 , top) (6 ,7 ).

Telomere DNA (8 -10 ) is typically characterized by multiple repeats of two to five consecutive dG residues that occur at the ends of eukaryotic chromosomes. It has long been known that homopolymers of rG as well as 5'-GMP and other guanosine derivatives form helices with G-tetrads (11 -15 ). It has been shown that dG-rich oligomers, modeled after telomeric repeat sequences, adopt tetraplex structures which are thermodynamically and kinetically stable (8 ). This ability of telomere DNA to form tetraplex-type structures in vitro and the associated stability of these structures has elicited considerable interest because of their possible involvement in recombinational events (16 ,17 ) and in telomere function (18 ,19 ).

G-tetrad structures are characterized by a cyclic network of eight H-bonds (Fig. 1 , top), which confer a high degree of stability, the half-life for unfolding being on the order of hours (20 ). Additionally, NMR characterization has revealed very slow rates for imino proton exchange, and in certain cases imino protons have still been observed at 90oC (21 ). Interestingly, chemical modification and native gel electrophoresis experiments (22 ,23 ) indicate that telomeric sequences containing adenine form tetraplexes that are less stable than those formed by non-adenine-containing sequences. However, the presence of adenine in the telomeric repeat of humans and other vertebrates, d(TnAGGG) (9 ), raises questions regarding the effect of dA residues on tetraplex structure and stability and their possible role in telomere function.


Figure 1. H-bond pattern expected for G-tetrad (top), and an isostructural A-tetrad (bottom) that is unstable because of van der Waals overlap of the `internalized amino hydrogens'.

To enhance our understanding of tetrad formation and polymorphism of G-rich sequences in the presence of dA residues, we have undertaken an investigation of d(A-G)10. This sequence, which adopts a single-stranded helical structure at acidic pH (24 -27 ), exhibits a completely different conformation at pH 7.0, as shown by CD spectroscopy (24 ) and by UV absorption, native gel electrophoresis, equilibrium sedimentation and nuclease sensitivity measurements [described in the preceding paper (1 )]. Shiber et al. (1 ) have thereby shown that d(A-G)10 adopts a structure at 4oC, which is dependent on both ionic strength and oligomer concentration, their experimental observations being explained by a single-hairpin duplex <-> double-hairpin tetraplex equilibrium. In this study, UV resonance Raman spectroscopy (UVRR) is employed to further examine the melting behavior of the oligomer under ionic conditions where the double-hairpin is highly favored at low temperature, and to identify the functional groups forming H-bonds at 4oC. In addition, the sugar-pucker and orientation of the bases with respect to the ribose moieties have been determined by monitoring conformationally sensitive Raman bands. From the UVRR results and those reported in the previous paper, a structure for d(A-G)10 is proposed in which two strands interact to form a hairpin-dimer helix with alternating stacked G-tetrads and intercalated dA residues.

MATERIALS AND METHODS

Nucleotides and oligonucleotides

2'-deoxyadenosine-5'-monophosphate-free acid (dAMP) and 2'-deoxyguanosine-5'-monophosphate-disodium salt (dGMP) (Peninsula Laboratories) were used without further purification. d(A-G)10 was synthesized, purified to homogeneity (24 ) and desalted (26 ). All concentrations are reported in total nucleotide residues.

Sample preparation

d(A-G)10 (2 mM) was in 0.3 M Na2SO4/0.01 M cacodylate or phosphate (pH 7.0). dAMP (1 mM) and dGMP (1 mM) were in 0.3 M Na2SO4/0.01 M cacodylate (pH 7.0). All samples (2 ml) were directly titrated at 25oC using a combination pH electrode (MI-410, Microelectrodes, Inc.) and filtered using a 0.2 mm Nylon-66 membrane (Rainin). Samples in D2O solvent were obtained by freezing samples in H2O, lyophilizing and dissolving at 4oC in D2O. Oligonucleotide integrity was verified after irradiation (27 ).

UVRR spectroscopy

Laser radiation was generated by frequency doubling the output of a 300 Hz excimer-pumped dye laser (Lambda Physik LPX130/FL3002). To generate 240 and 250 nm excitation, coumarin 480 (Exciton) was used as the lasing dye. For 210 nm excitation, stilbene 420 (Exciton) was the lasing dye. UV radiation at an average power of 0.3 mW was incident on the sample using a 135o backscattering geometry. The UVRR spectrometer has been described previously (28 ). Samples in a stirred quartz cuvette were maintained at specified temperatures with an aluminum block cell holder and circulating bath, with sample temperature monitored by a copper-constantan thermocouple in the block.

Spectra resulting from 0.5 to 1 h of data collection were corrected for the response of the monochromator and the detector by measuring the throughput of a deuterium lamp. Curve-fitting of the spectra was accomplished using Labcalc software (Galactic Instruments Corp.). The curves were generated assuming a 50% Lorentzian and 50% Gaussian line shape with a fixed width of 20 cm-1. The quality of the fit was assessed by both visual inspection and examination of the [chi]2 values resulting from the deviations between the fit simulations and the actual data.

RESULTS

Stacking of d(A-G)10

The 250 nm-excited UVRR spectrum of the homopurine oligomer d(A-G)10 (Fig. 2 ) closely resembles the co-added monomer spectrum, except that the intensity is diminished by ~65% relative to the monomer intensities at 4oC (Table 1 and 2 ). The bands arise from in-plane ring stretching vibrations, and the suppression of Raman intensity results from base stacking interactions (29 ).


Figure 2. UVRR spectra of d(A-G)10, dAMP and dGMP measured with 250 nm excitation. The spectrum for dAMP + dGMP was obtained by adding the two monomer spectra. All spectra are plotted on the same scale and have been normalized to the 981 cm-1 band of the sodium sulfate internal intensity standard.

Table 1 . Hypochromic ratiosa determined with 250 nm excitation
Temp oC

 1337 cm-1

 1485 cm-1

 1576 cm-1
4

 0.30

 0.23

 0.25
20

 0.33

 0.26

 0.28
40

 0.42

 0.35

 0.38
60

 0.47

 0.44

 0.44
80

 0.53

 0.54

 0.54
aIntensity of the indicated band relative to the corresponding band of the added constituent monomers.

Table 2 . Hypochromic ratiosa determined with 240 nm excitation
Temp oC

 1337 cm-1

 1485 cm-1

 1576 cm-1
4

 0.43

 0.33

 0.40
20

 0.45

 0.40

 0.44
40

 0.55

 0.48

 0.56
60

 0.65

 0.60

 0.68
80

 0.73

 0.75

 0.78
aIntensity of the indicated band relative to the corresponding band of the added constituent monomers.

UVRR spectra of d(A-G)10, taken at discrete intervals from 4 to 80oC (Fig. 3 ), demonstrate that Raman intensity increases with increasing temperature. Melting profiles were completely reversible and no significant differences were detected between the spectrum of the original and renatured samples at 4oC (data not shown). The Raman data are consistent with UV absorption melting profiles, which show a similar loss of hypochromicity upon thermal denaturation (1 ). Over this wide temperature range, the factor of two increase in Raman intensity (Tables 1 and 2 ) is consistent with a marked reduction in base stacking.


Figure 3. UVRR spectra of d(A-G)10 measured at 4, 20, 40, 60 and 80oC with 250 nm excitation. An average wavenumber value is shown in cases where the frequency of the Raman band is temperature dependent. All spectra are plotted on the same scale and normalized as in Figure 2.

With 250 nm excitation, the 1337 cm-1 band is dominated by scattering from dA residues (Fig. 2 ). As seen in Figure 4 , the intensity of this band in the d(A-G)10 spectra increases in an essentially linear manner to 80oC. The band intensities of the mononucleotides show no dependence on temperature over the same range. We infer from the behavior of the 1337 cm-1 band that the dA residues in d(A-G)10 become progressively less stacked with increasing temperature. This must be due in the lower temperature region to the removal from their intercalated position between G-tetrads (see below) to an extrahelical locus; so in that same temperature range, as the G-tetrads begin to melt, the dG residues can stack with nearest neighbor dA residues. It is the unstacking of these A-G neighbors that results in the continuing increase of the 1337 cm-1 Raman band and the more gradual increase in UV hyperchromicity at high temperature (1 ).


Figure 4. Intensity of the 1337 cm-1 band (250 nm excitation) as a function of temperature for d(A-G)10 ([squ]), dAMP (z) and dGMP ([squf]). Intensities for d(A-G)10 are those shown in Figure 3.

In contrast, the dG residues unstack in a cooperative manner, as can be observed with 240 nm excitation, at which wavelength Raman scattering from dG is enhanced preferentially relative to dA residues (30 ). The temperature-dependent spectra for d(A-G)10 at this wavelength are shown in Figure 5 , while the thermal profile of the 1575 cm-1 band, a coupled C4=C5/C4-N3 stretching mode of dG (31 ), is shown in Figure 6 . In dGMP, this band actually decreases slightly with temperature, reflecting a small thermally-induced red-shift in the absorption spectrum (26 ). The dA residues give rise to a weak band at the same frequency, but no temperature dependence is observed for this band in dAMP (Fig. 6 ). Other bands in the 240 nm-excited spectra of d(A-G)10 give s-shaped temperature profiles similar to that of the 1575 cm-1 band.


Figure 5. UVRR spectra of d(A-G)10 measured at 4, 20, 40, 60 and 80oC with 240 nm excitation. An average wavenumber value is shown in cases where the frequency of the Raman band is temperature dependent. Spectra were plotted on the same scale and normalized as in Figure 2.

Similar cooperative melting behavior is seen when the UV absorbance at 260 nm (due 2/3 to dA residues and 1/3 to dG residues) is plotted against temperature (Fig. 6 , bottom) (1 ), and Tm is ~42oC, as it is in the 240 nm-excited UVRR intensity profiles (Fig. 6 , top). That the melting profiles measured by the two methods are not identical is due in part to the fact that subtle changes in the absorption spectrum can produce more dramatic ones in the UVRR spectrum. This is because resonance Raman intensity is approximately proportional to the square of the absorptivity (30 ,32 ). The UVRR results establish, nevertheless, that the more cooperative behavior is limited to dG residues, the dA residues melting non-cooperatively.

We infer from this disparate melting that the A and G bases are not paired to each other in the structure at low temperature. Hence, if there is base pairing in that structure, then dG must pair with dG. As seen below, there is no evidence that dA residues are H-bonded to each other.


Figure 6. (Top) Intensity at 1575 cm-1 plotted as a function of temperature for 2 mM d(A-G)10 ([squ]), 1 mM dAMP (z) and 1 mM dGMP ([squf]) in 0.3 M Na2SO4 and 0.01 M cacodylate, pH 7.0. Intensities for d(A-G)10 taken from the spectra shown in Figure 5. (Bottom) UV absorbance of 0.64 mM d(A-G)10 in 0.3 M Na2SO4 and 0.01 M cacodylate, pH 7.0 at 260 nm as a function of temperature.

H-bonding of the dG residues

Involvement of the C6 carbonyl. The nature of base-pairing was investigated with 210 nm excitation, in order to resonantly enhance scattering from the carbonyl moieties of the dG residues (31 ,33 ). The C6=O stretching coordinate is the main contributor to the d(A-G)10 band seen at 1680 cm-1 in H2O (Fig. 7 ) or at 1654 cm-1 in D2O (Fig. 8 ); the D2O downshift reflects a substantial contribution from the NH bend of the adjacent imine group (32 ). Both bands shift up substantially when the temperature is raised from 4 to 80oC, 7 cm-1 in H2O (Fig. 7 ) and 13 cm-1 in D2O (Fig. 8 ). These upshifts imply a significant strengthening of the C6=O bond at high temperature, consistent with a loss of H-bonding (34 ). dGMP by itself shows an upshift in this band (data not shown), but the extent of the shift between 4 and 80oC is only 5 cm-1 in either H2O or D2O. Weakening of the H-bonding to H2O molecules can account for the temperature effect on dGMP. The larger shifts seen for d(A-G)10 reflect disruption of stronger H-bonds, consistent with G.G base pairing at low temperature. The same magnitude of shift was also seen with dGMP, which is known to form helically wound G-tetrads (11 ) when the concentration is brought to 10 mg/ml (data not shown). We note that in helical structures of poly(G) and poly(dG-dC), the C6=O band splits and the shift pattern is more complex because of strong interbase coupling between C=O modes of adjacent residues (35 -37 ). The absence of these effects in d(A-G)10 indicates that adjacent bases along the strands are not coupled vibrationally, consistent with the uncorrelated fluctuations of dG and dA residues implied by the melting behavior, and that the bases themselves deviate significantly from coplanarity (32 ).


Figure 7. (Top) 210 nm-excited UVRR spectra of d(A-G)10 in the carbonyl and -NH2 scissors region. The spectra are normalized to the peak maximum at 1580 cm-1. (Bottom) Difference spectrum generated between 80 and 4oC spectra.


Figure 8. (Top) 210 nm-excited UVRR spectra of d(A-G)10 in D2O taken at 4, 20, 40, 60 and 80oC. All spectra have been normalized to the sodium sulfate intensity standard. The large band at ~1200 cm-1 arises from D2O. (Bottom) Difference spectrum generated between 80 and 4oC spectra. All spectra are plotted on the same scale; the difference spectrum has been multiplied by a y-scale factor of 2.

Involvement of the exocyclic NH2. The exocyclic -NH2 scissors mode gives rise to weakly enhanced UVRR bands near 1600 cm-1 for both dA and dG residues in DNA (31 ). In D2O, exchange of NH2 for ND2 shifts the mode to ~1195 cm-1 for dA residues and ~1210 cm-1 for dG residues. It is therefore revealing that the UVRR difference spectrum of d(A-G)10 between 80 and 4oC in D2O (Fig. 8 ) shows a bisignate band with a trough at 1235 cm-1 and a peak at 1198 cm-1. We interpret this signal as arising from an upshift of the -ND2 scissors mode of the dG residues at low temperature, due to stronger H-bonding, an interaction which is expected to increase the N-D bending force constant. Comparable upshifts have been reported for 9-ethyladenine in proton-accepting solvents and support our assignment of frequency upshifts to H-bond formation (38 ). The H-bonding effect is less obvious in the H2O spectra, because of overlap between the G and A -NH2 scissors modes. Nevertheless, a close examination of the 1600 cm-1 region (Fig. 9 ) reveals a significant broadening of the composite band at low temperature. Curve resolution yields two components, 1598 and 1609 cm-1, at 4oC, but only one component at 1601 cm-1, at 80oC. We interpret the broadening at low temperature as an upshift in the -NH2 scissors frequency of the dG residues due to H-bonding, just as in the D2O spectrum.


Figure 9. Curve-fitting of the 240 nm-excited spectral region from 1540 to 1640 cm-1; data are depicted with a solid line and the curve-fit spectrum with a dotted line. Original spectra are shown in Figure 5. Spectra are normalized to the peak maximum at 1577 cm-1. Curve-fit bands are generated with a fixed width of 20 cm-1 and a fixed 50/50 Gaussian-Lorentzian ratio.

Involvement of guanine N7. The spectra also provide evidence for H-bonding to N7 of the dG residues. When excited at 240 nm, the strong band at 1484 cm-1 shifts to 1487 cm-1 as the temperature is raised from 4 to 80oC (Fig. 5 , inset). This band is due mainly to a G ring mode involving N7=C8 stretching and C8-H bending coordinates (Fig. 2 , inset) (31 ,34 ). The frequency of this mode is sensitive to the C8 substituent and to any molecular interactions at N7 (34 ,37 ). Such a downshift was also observed for 3',5' GpG (37 ) at low temperature (1490 to 1481 cm-1 between 80 and 22oC) and attributed to N7[middot][middot][middot]H2N intermolecular H-bonding in helical aggregates of stacked GpG tetramers. Likewise, downshifts of this band were observed in helices of dGMP gels and of poly(dG) (29 ). It is important to note that with 250 nm excitation, where an A ring mode dominates the 1485 cm-1 band, there is no temperature effect on the frequency (Fig. 3 ). The temperature effect observed with 240 nm excitation is therefore attributed to H-bond involvement by the N7 atoms of G at low temperature.

Further evidence for H-bonding at N7 of G is detected in the course of melting in D2O (Fig. 8 ); a band at 1338 cm-1 shifts to 1345 cm-1 upon thermal denaturation. A new band at 1348 cm-1 in the 80-4oC difference spectrum additionally confirms this frequency upshift (Fig. 8 ). Fodor et al. (31 ) assigned the G vibrational mode observed at 1364 cm-1 in H2O to in-phase stretching motions of the triene linkage: C8=N7-C5=C4-N3=C2. The intensity of this mode is amplified with 210 nm excitation and its frequency is lowered in D2O (31 ,39 ); thus, the band observed at 1338 cm-1 in d(A-G)10 is assigned to this stretching vibration. 15N substitution in model compounds has shown that N7 motion significantly affects the frequency of this band (39 ). Hence, we suggest that the upshift of this mode in the oligomer at higher temperature reflects a loss of H-bonding at N7.

dA residues do not form H-bonded tetrads

UVRR melting studies were also examined for evidence of H-bonding of the dA residues. Previous studies (27 ,38 ) have shown that H-bonding or protonation at the acceptor sites of dA residues (N1, N3 and N7) induces an upshift of ring frequencies above 1400 cm-1 with concomitant downshifts of the modes at 1335 and 1310 cm-1. For d(A-G)10, when dA residues are preferentially selected by 250 nm excitation, the frequencies of these bands remain constant during thermal denaturation, implying that H-bonding at these sites is extremely weak or non-existent. As described above, the downshift of the -NH2 scissors mode upon deuteration and subsequent thermally-induced wavenumber shift implicates -NH2 groups of dG but not dA residues in H-bonding at low temperature (Fig. 8 ). Figure 1 (bottom) depicts a model for A-tetrads that is isostructural with the proposed G-tetrads, but it requires H-bonding between N1 and -NH2 of dA residues of adjacent strands. The UVRR studies show no evidence for H-bonding at those sites, nor at any other H-bond acceptor sites on the dA residues. Therefore, we conclude that H-bonded A-tetrads are not present in d(A-G)10.

Conformation of the ribosyl ring

As a consequence of vibrational coupling between base and ribose rings, UVRR spectra in the 600-900 cm-1 region (Fig. 10 ) contain bands arising from ring vibrations, which are characteristic of glycosyl bond orientation and sugar pucker. The bands observed at 682 and 863 cm-1 in the 4oC spectrum reflect dG residues which are in an anti orientation with respect to deoxyribose rings in the C2'-endo conformation (Fig. 10 ) (32 ,40 ). When the temperature is raised, these bands become broader and weaker, reflecting increased flexibility of the backbone and a broadening of the range of accessible conformers. Figure 11 shows that the diminution of the 683 cm-1 band peak height exhibits the same cooperative melting behavior as do the high frequency G mode enhancements (Fig. 6 ). Thus, the conformational flexibility of the dG residues correlates with unstacking. The 4oC spectrum (Fig. 10 ) also shows weak bands at 628 and 847 cm-1 positions characteristic of dG residues in a C3'-endo/syn configuration. Thus, a fraction of the dG residues in d(A-G)10 may adopt syn orientations at low temperature, but by 20oC, these bands have disappeared. As the UV Raman cross sections for these bands are not well known, the population of dG residues that are C3'-endo/syn cannot be accurately estimated; however, the observation of both bands confirms that some of the dG residues do adopt the syn configuration.


Figure 10. 210 nm-excited UVRR spectra of d(A-G)10 in the conformation-sensitive region measured at 4, 20, 40, 60 and 80oC. Raman bands at 628 and 847 cm-1 correspond to dG residues in the C3'-endo/syn configuration, while the band at 863 cm-1 corresponds to dG residues in the C2'-endo/anti configuration. The band at 730 cm-1 is attributed to dA residues in the C2'-endo/anti configuration. Spectra have been normalized to the sodium sulfate intensity standard.

The 730 cm-1 band in the d(A-G)10 spectrum is characteristic of dA residues in an anti orientation with C2'-endo sugar conformation (41 ). This band becomes stronger with increasing temperature, just as do the high-frequency bands of the dA residues (Fig. 4 ). However, the 730 cm-1 band augmentation is not linear over the whole temperature range (Fig. 11 ), but only begins when the temperature is raised above 20oC. One possible interpretation is that some of the dA residues are in the syn orientation at 4oC. Such residues are expected to contribute to the 628 cm-1 UVRR band (41 ), along with any dG syn residues. Since the 628 cm-1 band disappears when the temperature is raised to 20oC, the syn residues (being in a higher energy conformation) melt before the main population of dA residues, which remain in the C2'-endo/anti configuration. Theseresidues, which had melted out of the structure, would sample a mixture of configurations and therefore, not contribute to any particular Raman band. Both melting stages contribute to the high frequency A mode hyperchromism, which has an overall linear temperature dependence. This Raman band is composed primarily of pyrimidine modes, and with 250 nm excitation exhibits little conformational sensitivity (39 ).


Figure 11. Intensity of the dG and dA conformational marker bands at 683 and 730 cm-1 respectively, plotted as a function of temperature. Intensities were taken from the spectra depicted in Figure 10.

DISCUSSION

The UVRR observations presented here are fully consistent with the presence of G-tetrads in d(A-G)10 at low temperature. Complex formation from two strands (1 ) implicates base-base association between strands, but the uncorrelated melting of the dA and dG residues would seem to rule out H-bonding interactions between them. Moreover, the involvement of the -NH2 substituent of dG but not of dA residues in interbase H-bonding is implicated by low-temperature upshifts in only the G-associated NH2 or ND2 scissors mode frequencies; and H-bond acceptance by the C6=O carbonyl groups and N7 atoms of G is indicated by frequency downshifts of appropriate modes. These H-bond interactions are entirely consistent with the G-tetrad model in Figure 1 (top), in which the C6=O and N1H groups form H-bonds, as do the exocyclic -NH2 groups and the N7 atoms.

Some uncertainty exists regarding the assignment of the -NH2 scissors vibration. In a recent report (42 ), the spectral effects of D2O exchange of the amino and imino protons in the oligomer (dG)12 were examined and the 1603 cm-1 band was attributed to N1H in-plane bending, while a band at 1644 cm-1 (not observed in our data) was attributed to scissoring of the NH2 amino group. We note that the ~1604 and 1640 cm-1 bands both arise primarily from ring vibrations which have variable contributions from N1H bending and -NH2 scissoring motions (43 ). In the case of adenine and its N9-derivatives, the relative contribution of -NH2 scissoring to the modes at ~1604 and 1635 cm-1 strongly depends upon the degree of H-bonding. -NH2 scissoring motion primarily contributes to the lower frequency band in weak to moderate H-bonding solutions, but to the higher frequency band in stronger H-bonding solutions or in the crystalline state (38 ). In the case of (dG)12, as for linear homo G tetraplexes in general, the H-bonding is considerably stronger and the melting more cooperative (e.g. ref. 12 ) than in d(A-G)10 (1 ), which argues for a less stable structure for d(A-G)10 and consequently, weaker H-bonds. It is possible that in guanine as well as adenine the relative -NH2 contribution to the 1640 and 1604 cm-1 modes is strongly dependent on the degree of H-bonding and could be reversed for the two sequences, d(A-G)10 and (dG)12. However, we note that even if the band observed arises mainly from N1H bending rather than -NH2 scissors vibrations, the detected frequency shifts would reflect H-bond interactions of the C6=O and N1H groups and would still be consistent with G-tetrad formation.


Figure 12. One of several possible models for a two-hairpin tetraplex of d(A-G)10 at 4oC, with alternating dG-tetrads and intercalated dA residues. H-bonding between the dG residues in the tetrads is represented by the dotted lines.

In contrast, the UVRR spectra exhibit no evidence for interbase H-bonding of the dA residues in d(A-G)10, even at low temperature, as shown by the relatively temperature-insensitive frequencies of dA vibrational modes. These findings are consistent with the conclusion (23 ) that in telomeric sequences containing dA residues, there are G-tetrads but not A-tetrads.

Melting of the G-tetrads in d(A-G)10 occurs more or less cooperatively, with a 42oC transition temperature, and leads to unstacking of the dG residues, as seen in the UVRR hyperchromism of the G ring modes. Melting also disorders the glycosyl bond orientations and the sugar pucker, as revealed by the conformationally-sensitive G vibrations. Because of the absence of anchoring H-bond interactions, the dA residues melt gradually. Nevertheless, it is apparent that the dA residues, like the dG residues, contribute to the UVRR hypochromism at low temperature. This must mean that at low temperature the dA residues incur some stabilization by intercalating and stacking with their nearest neighbor G-tetrads.

The fact that the dA residues do not form H-bonded tetrads is probably because they cannot form a tetrad of sufficient stability that is isostructural with the G-tetrads (Fig. 1 ). Hence, as indicated by the UV melting profiles of Shiber et al. (1 ), the dA residues melt gradually over the whole temperature range; at low temperatures by becoming extrahelical, then by stacking with nearest neighbor dG residues as those melt, and then at still higher temperatures by unstacking of the dA-dG stacked neighbors.

Although UVRR observations shed light on the nature of base stacking and the interbase interactions, they do not directly identify the relative orientation of the strands. From the results of Shiber et al. (1 ) it is possible to distinguish between the four-stranded linear and two-stranded hairpin duplex structures, and it is the hairpin duplex structure that is consistent with experimental observations (Fig. 12 ). Our UVRR observations are fully compatible with this model. Although Rippe et al. (44 ) proposed parallel-stranded duplexes for alternating d(G-A) sequences, the indication that some of the residues are in the syn orientation (bands at 628 and 847 cm-1 in Fig. 10 ) strongly supports an antiparallel strand orientation. Moreover, the model of Rippe et al. (44 ) does not anticipate our findings regarding the presence of G-tetrads. From a Raman study of d(G4T4G4), (45 ) it was concluded that syn orientations are associated only with the antiparallel hairpin conformation and not with a linear parallel conformation. Other spectroscopic investigations (8 ) have also shown that bases involved in the same tetrad have opposite glycosyl torsions if the bases are on antiparallel strands. In those studies, however, G-residues adopt a C2'-endo/syn configuration and not the C3'-endo/syn configuration observed in the present study. Definitive evidence for the orientation of the strands must await further spectroscopic investigation.

ACKNOWLEDGEMENTS

We thank Horst Klump for helpful discussions. This work was supported by grants from the National Institutes of Health (GM 42936) to J.R.F. and (GM 25158) to T.G.S. M.C.S. was supported in part by a Sterling postdoctoral fellowship, and I.M. by a NRSA fellowship (GM 14324).

REFERENCES

1 Shiber,M.C., Braswell,E.H., Klump,H. and Fresco,J.R. (1996) Nucleic Acids Res. 24, 5004-5012.

2 Manor,H., Rao,B.S. and Martin,R.G. (1988) J. Mol. Evol., 27, 96-101. MEDLINE Abstract

3 Palecek,E. (1991) CRC Crit. Rev. Biochem. Mol. Biol., 26, 151-226.

4 Yagil,G. (1991) CRC Crit. Rev. Biochem. Mol. Biol., 26, 475-559.

5 Shimizu,M., Hanvey,J.C. and Wells,R.D. (1989) J. Biol. Chem., 264, 5944-5949. MEDLINE Abstract

6 Kang,C., Zhang,X., Ratliff,R., Moyzis,R. and Rich,A. (1992) Nature, 356, 126-131. MEDLINE Abstract

7 Smith,F.W. and Feigon,J. (1992) Nature, 356, 164-168. MEDLINE Abstract

8 Williamson,J.R. (1994) Annu. Rev. Biophys. Biomol. Struct., 23, 703-730. MEDLINE Abstract

9 Blackburn,E.H. (1991) Nature, 350, 569-573. MEDLINE Abstract

10 Blackburn,E.H. (1992) Annu. Rev. Biochem., 61, 113-129. MEDLINE Abstract

11 Gellert,M., Lipsett,M.N. and Davies,D.R. (1962) Proc. Natl. Acad. Sci. USA, 48, 2013-2018.

12 Fresco,J.R. and Massoulié,J. (1963) J. Am. Chem. Soc., 85, 1352-1353.

13 Tougard,P., Chantot,J.-F. and Guschlbauer,W. (1973) Biochim. Biophys. Acta, 308, 9-16. MEDLINE Abstract

14 Arnott,S., Chandrasekaran,R. and Marttila,C.M. (1974) Biochem. J., 141, 537-543. MEDLINE Abstract

15 Zimmerman,S.B., Cohen,G.H. and Davies,D.R. (1975) J. Mol. Biol., 92, 181-192. MEDLINE Abstract

16 Sen,D. and Gilbert,W. (1988) Nature, 334, 364-366. MEDLINE Abstract

17 Sen,D. and Gilbert,W. (1990) Nature, 344, 410-414. MEDLINE Abstract

18 Williamson,J.R., Raghuraman,M.K. and Cech,T.R. (1990) Cell, 59, 871-880.

19 Sundquist,W.I. and Klug,A. (1989) Nature, 342, 825-829. MEDLINE Abstract

20 Raghuraman,M.K. and Cech,T.R. (1990) Nucleic Acids Res., 18, 4543-4552. MEDLINE Abstract

21 Hardin,C.C., Henderson,E., Watson,T. and Prosser,J.K. (1991) Biochemistry, 30, 4460-4472. MEDLINE Abstract

22 Guo,Q., Lu,M. and Kallenbach,N.R. (1992) J. Biol. Chem., 267, 15293-15300. MEDLINE Abstract

23 Murchie,A.I.H. and Lilley,D.M.J. (1994) EMBO J., 13, 993-1001.

24 Dolinnaya,N.G. and Fresco,J.R. (1992) Proc. Natl. Acad. Sci. USA, 89, 9242-9246. MEDLINE Abstract

25 Dolinnaya,N.G., Braswell,E.H., Fossella,J.A., Klump,H. and Fresco,J.R. (1993) Biochemistry, 32, 10263-10270. MEDLINE Abstract

26 Shiber,M.C., Lavelle,L., Fossella,J.A. and Fresco,J.R. (1995) Biochemistry,34, 14293-14299. MEDLINE Abstract

27 Mukerji,I., Shiber,M.C., Spiro,T.G. and Fresco,J.R. (1995) Biochemistry,34, 14300-14303. MEDLINE Abstract

28 Rodgers,K.R., Su,C., Subramaniam,S. and Spiro,T.G. (1992) J. Am. Chem. Soc., 114, 3697-3709.

29 Small,E.W. and Peticolas,W.L. (1971) Biopolymers, 10, 1377-1416. MEDLINE Abstract

30 Perno,J.R., Grygon,C.A. and Spiro,T.G. (1989) J. Phys. Chem., 93, 5672-5678.

31 Fodor,S.P.A., Rava,R.P., Hays,T.R. and Spiro,T.G. (1985) J. Am. Chem. Soc., 107, 1520-1529.

32 Fodor,S.P.A. and Spiro,T.G. (1986) J. Am. Chem. Soc., 108, 3198-3205.

33 Ziegler,L.D., Hudson,B., Strommen,D.P. and Peticolas,W.L. (1984) Biopolymers, 23, 2067-2081. MEDLINE Abstract

34 Tsuboi,M., Nishimura,Y., Hirakawa,A.Y. and Peticolas,W.L. (1987) In Spiro,T.G. (ed.), Biological Applications of Raman Spectroscopy. John Wiley & Sons, Chichester, Vol. 2, pp. 109-179.

35 Tsuboi,M., Takahashi,S. and Harada,I. (1973) In Duchesne,J. (ed.), Physico-Chemical Properties of Nucleic Acids. Academic Press, New York, Vol. 2, pp. 91-145.

36 Tsuboi,M. and Nishimura,Y. (1982) In Lascombe,J. and Huang,P.V. (eds) Raman Spectroscopy-Linear and Nonlinear. John Wiley & Sons, Chichester, pp. 683-693.

37 Savoie,R., Klump,H. and Peticolas,W.L. (1978) Biopolymers, 17, 1335-1345.

38 Toyama,A., Takeuchi,H. and Harada,I. (1991) J. Mol. Struct., 242, 87-98.

39 Toyama,A., Takino,Y., Takeuchi,H. and Harada,I. (1993) J. Am. Chem. Soc., 115, 11092-11098.

40 Benevides,J.M. and Thomas,G.J., Jr. (1983) Nucleic Acids Res., 16, 5747-5761.

41 Thomas,G.J., Jr. and Tsuboi,M. (1993) In Bush,C.A. (ed.) Advances in Biophysical Chemistry. JAI Press, Greenwich, Vol. 3, pp. 1-70.

42 Miura,T. and Thomas,G.J., Jr. (1995) Biochemistry, 34, 9645-9654. MEDLINE Abstract

43 Delabar,J.-M. and Majoube,M. (1978) Spectrochimica Acta, 34A, 129-140.

44 Rippe,K., Fritsch,V., Westhof,E. and Jovin,T.M. (1992) EMBO J., 11, 3777-3786. MEDLINE Abstract

45 Miura,T. and Thomas,G.J., Jr. (1994) Biochemistry, 33, 7848-7856. MEDLINE Abstract

This is paper No. 25 in the series Polynucleotides, of which the last is ref. 42 .

Return

*To whom correspondence should be addressed. Tel: +1 609 258 3927; Fax: +1 609 258 6730; Email: jrfresco@princeton.edu or spiro@princeton.edu

+Present Address: Department of Molecular Biology and Biochemistry, Wesleyan University, Middletown, CT 06459-0175, USA
Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 Proc. Natl. Acad. Sci. USAHome page
W. Suen, T. G. Spiro, L. C. Sowers, and J. R. Fresco
Identification by UV resonance Raman spectroscopy of an imino tautomer of 5-hydroxy-2'-deoxycytidine, a powerful base analog transition mutagen with a much higher unfavored tautomer frequency than that of the natural residue 2'-deoxycytidine
PNAS, April 13, 1999; 96(8): 4500 - 4505.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (120K) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (21)
Right arrowRequest Permissions
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Mukerji, I
Right arrow Articles by Spiro, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Mukerji, I
Right arrow Articles by Spiro, T.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?