Eur. J. Biochem. 96,267-276 (1979)

Helix-Coil Transition of the Self-Complementary dG-dG-dA-dA-dT-dT-dC-dC Duplex Dinshaw J. PATEL and Lita L. CANUEL Bell Laboratories, Murray Hill, New Jersey (Received July 19, 1978/January 8, 1979)

The helix-coil transition of the octanucleotide self-complementary duplex dG-dG-d A-dA-dT-dTdC-dC has been monitored at the Watson-Crick protons, the base and sugar nonexchangeable protons and the backbone phosphates by high-resolution nuclear magnetic resonance (NMR) spectroscopy. The melting transition of the octanucleotide monitored by ultraviolet absorbance spectroscopy is characterized by the thermodynamic parameters AH" = - 216.7 kJ/mol and AS" (25 "C) = -0.632 kJ mol-' K-' in 0.1 M NaCl, 10 mM phosphate solution. Correlation of the transition midpoint values monitored by the ultraviolet absorbance studies at strand concentrations below 0.2 mM and by NMR studies at 5.3 mM suggest that both methods are monitoring the octanucleotide duplex-to-strand transition. The NMR spectra of the Watson-Crick ring NH protons of the octanucleotide duplex have been followed as a function of temperature. The resonance from the terminal dG-dC base pairs broadens out at room temperature while the resonances from the other base pairs broaden simultaneously with the onset of the melting transition. The nonexchangeable base and sugar H-1' protons are resolved in the duplex and strand states and shift as average peaks through the melting transition. The experimental shifts on duplex formation have been compared with calculated values based on ring-current and atomic diamagnetic anisotropy contributions for a B-DNA base-pair-overlap geometry in solution. Several nonexchangeable proton resonances broaden in the fast-exchange region during the duplex-to-strand transition and the excess widths yield a duplex dissociation rate constant for the octanucleotide of 1.9 x lo3 s-' at 32 "C (fraction of duplex = 0.86) in 0.1 M NaCl, 10 mM phosphate buffer. The 31Presonances of the seven internucleotide phosphates are distributed over 0.6 ppm in the duplex state, shift downfield during the duplex-to-strand transition and undergo additional downfield shifts during the stacked-to-unstacked strand transition with increasing temperature. The application of nuclear magnetic resonance (NMR) spectroscopy to monitor the duplex-to-strand transition of oligonucleotide and polynucleotide sequences has been recently reviewed and demonstrates the power of this technique to differentiate between individual base pairs in a duplex and, in addition, to monitor each base pair at the Watson-Crick hydrogen bonds, the nonexchangeable base and sugar protons and the backbone phosphates in solution [1,2]. These investigations include studies on the exchangeable and nonexchangeable protons of the deoxy pentanucleotide (dT-dT-dG-dT-dT) . (dA-dA-dC-dAdA) [3,4] and deoxy block polymer (dC,,-dA,,) . (dT,,-dG,,) [5], and the self-complementary sequences dA-dT-dG-dC-dA-dT [6] and poly(dA-dT) [7]. In addition, the exchangeable protons in double-stranded and triple-stranded oligomer duplexes containing Watson-Crick and Hoogsteen base pair have been Abbreviation. NMR, nuclear magnetic resonance. Enzyme. Restriction endonuclease EcoRI from Escherichiu coli

(EC 3.1.4.-).

monitored and characterized in solution by NMR methods [8]. Considerable effort has also been devoted to the rib0 analogs and include detailed studies on the selfcomplementary tetranucleotide C-C-G-G [9] and hexanucleotide A-A-G-C-U-U [lo] sequences and the complementary oligo (A). oligo (U) duplex [ 111. These studies have been extended to monitor and compare the NMR parameters for the synthetic RNAs, poly(1-C) and poly(A-U), and for the synthetic DNAs, poly(d1-dC) and poly(dA-dU), with the same alternating purine-pyrimidine sequence [12,13]. We have recently reported on NMR studies of the netropsin .poly(dA-dT) complex, phosphate/drug = 50 and have monitored the nucleic-acid proton resonances through the biphasic duplex-to-strand transition [14]. We were unable to monitor the resonances of the antibiotic and could not probe the intermolecular hydrogen bonds between the antibiotic peptide groups and the nucleic-acid base-pair edges in the complex [15,16] at the synthetic DNA level. It became necessary therefore to initiate an investigation of stable oligonucleo-

268 dG-dG - d A o

o

r

n

dA e

-

d T - d T - d C -dC e

r

n

o

-

o

dC-dC-dT-dT-dA-dA-dG-dG 1

2

3

4

Scheme 1. The self-complementary octanucleotide sequence dG-dGdA-dA-dT-(IT-dC-dC.The numbers are used to refer to the base 0, W, 0 ) refer to the four pairs in III~. I C Y I . The symbols (0, different b,ixi. piiirs in the duplex

tide sequences due to their narrower line widths compared to those found for synthetic DNA in solution. Previous studies have demonstrated that self-complementary G .C-containing tetranucleotide sequences at millimolar concentrations form stable duplexes at 0 "C [17]. By contrast, the transition midpoints of the corresponding A.T-containing self-complementary tetranucleotide sequences is much below 0 "C in aqueous solution. We have therefore studied the self-complementary octanucleotide sequence dG-dGdA-dA-dT-dT-dC-dC (Scheme 1) which contains a central cluster of four dA.dT base pairs flanked on either side by two stabilizing, d G . d C base pairs. Since netropsin and distamycin complex specifically at dA.dT sites [15,16], the central cluster of dA.dT base pairs serves as a potential binding site for the peptide antibiotics. The dG-dG-dA-dA-dT-dT-dC-dC duplex is of additional interest since it is recognized by the restriction enzyme EcoRI which cleaves it assymmetrically at the dG-dA site on each strand.

Helix-Coil Transition of the dG-dG-dA-dA-dT-dT-dC-dC Duplex

RESULTS Transition Midpoint

The duplex-to-strand transition of the self-complementary dG-dG-d A-dA-dT-dT-dC-dC octanucleotide has been investigated by monitoring the 260-nm absorbance band as a function of temperature at total strand concentrations ranging from 0.2 mM to 8 pM in 10 mM phosphate buffer, 1 mM EDTA, H,O, pH 6.85, in the absence and presence of 0.1 M NaCl and 1 M NaCl. Typical melting curves normalized to an absorbance of 1.Oat high temperature for a total strand concentration of 0.17 mM at the three buffer conditions are presented in Fig. 1. The transition midpoints are 21 "C (no added NaCI), 34.5 "C (0.1 M NaCI) and 44.5 "C (1 M NaCl) in 10 mM phosphate buffer solutions. The dependence of the transition midpoint, t,,, , on total strand concentration of the octanucleotide, in 10 mM phosphate buffer, in the absence and presence of 0.1 M NaCl and 1 M NaCl are presented in Fig. 2 and summarized in Table 1. The NMR studies of the nonexchangeable protons to be presented below were run on 5.3 mM (strand concentration) dG-dG-dA-dA-dT-dT-dC-dC in 0.1 M NaCl, 10 mM phosphate, 'H,O, and the plots are extended to this concentration range to compare the transition midpoints from the NMR spectra with those evaluated from the ultraviolet absorbance melting curves (Fig. 2). Watson-Crick Protons

EXPERIMENTAL PROCEDURE Materials dG-dG-dA-dA-dT-dT-dC-dCwas purchased from Collaborative Research (Waltham, Mass. U.S.A.). It was passed twice through Sephadex G-10 columns to remove any salt present in the commercially prepared sample. The absorption coefficient (260 nm) was determined to be 54x lo3 M-' cm-i (with respect to strand concentration') at 23 "C. Spectra

Proton NMR spectra in H,O were recorded in the continuous-wave mode while proton and phosphorus NMR spectra in 2 H 2 0 were recorded in the Fouriertransform mode on a Bruker HX-360 spectrometer interfaced to a BNC-12 computer. Proton chemical shifts are referenced relative to internal sodium 2,2dimethyl-2-silapentane-5-sulfonate, (CH3)3Si(CH2)3 S03Na,, while phosphorus chemical shifts are referenced relative to internal trimethyl phosphate, (CH,O),PO. Homonuclear spin-decoupling of the cytosine H-5 and H-6 resonances on the octanucleotide were undertaken in the Fourier-transform mode.

The integrity of the dG-dG-dA-dA-dT-dT-dC-dC self-complementary duplex can be probed by monitoring theguanine H-1 and thymine H-3 hydrogen-bonded exchangeable protons in H,O solution at temperatures below the midpoint of the melting transition [4,18,19]. The spectra of the exchangeable protons in the octanucleotide duplex (10.5 mM strand concentration) in 0.1 M NaCI, 10 mM phosphate buffer, H,O, pH 6.4 at 4.7"C, 23.6"C and 32.5"C are presented in Fig. 3. Four resolvable resonances are observed between 11.5 ppm and 14 ppm at 4.7 "C, with the resonance at ~ 1 2 . 9 ppm 5 broadening out on raising the temperature to 23.6 "C (Fig. 3). The three remaining resonances broadened significantly in a concerted manner on raising the temperature to 32.5 "C (Fig. 3). Two exchangeable resonances are observed at 8.4 ppm at 4 "C with one of these resonances shifting upfield with increasing temperature. These resonances probably correspond to side-chain NH, Watson-Crick hydrogen-bonded protons. Nonexchangeable Proton Chemical Shifts

The 360-MHz proton NMR spectra of 5.3 mM (strand concentration) dG-dG-dA-dA-dT-dT-dC-dC

269

D. J. Pate1 and L. L. Canuel

0.80 I

0

10

I 20

I

I

I

30

40

70

I I 50 60 Temperature ("C)

80

r1LULl

~

I iLlLLlLL

~

L 1

L 1

10.~

uLLIIILu;J 10.~

10-2

Total strand concentration (M)

Fig. 2. A plot ofthe inverse ofthe transition midpoint, (t,,,-', against the total strand concentration for the dG-dG-dA-dA-dT-dT-dC-dC octanucleotide in IOmMphosphate buffer, l m M EDTA, H20,pH6.85, in the absence (@), andpresence of 0.1 M NaCl ( 0 ) and I M NaCl ( A ) . The data points are from ultraviolet absorbance melting curves while the transition midpoint deduced from NMR studies at 5.3 mM octanucleotide strand concentration is designated by (I)

Table 1. The concentration dependence of the transition midpoint, t,, deduced from ultraviolet absorbance melting curves of'the dG-dGdA-dA-dT-dT-dC-dC octanucleotide duplex The octanucleotide was dissolved in 10 mM phosphate buffer, 1 mM EDTA, H,O, pH 6.85 (no salt) or in 0.1 M NaCl, 10 mM phosphate buffer, 1 mM EDTA, H,O, pH 6.85 (low salt) or in 1.0 M NaCI, 10 mM phosphate buffer, 1 mM EDTA, H,O, pH 6.85 (high salt) Total stand concn

t , in solution with ___ no salt low salt .

PM

"C

170 90.4 51.12 34.07 17.04 8.52

21 19 16.5 15.5 13.5 12

34.5 33.5 29.5 28.5 25.5 24.5

12

6 (ppm)

Fig. 1. Ultraviolet absorbunce melting curves (normalized to an absorbance of 1.0 at high temperature) for 0.17 m M (strand concentru10 mMphosphate. I m M E D T A , tion) dG-dG-dA-dA-dT-dT-dC-dCin H,O, pH 6.85, in the absence ( I ) and presence o f 0 . l M NaCl ( I I ) and I M NaCl ( I I I ) . The transition midpoints increase with salt concentration

3 .o

13

14

__-

high salt

43.5 43.0 39.5 38.5 37.5 33.0

Fig. 3. The 360-MHz continuous-wave proton N M R spectra ( 1 2 - 14 ppm) of 10.5 m M (strandconcentration) dG-dG-dA-dA-dT-dT-dC-dC in 0.1 M NaCl, I0 mM phosphate buffer, I mM EDTA, H,O. p f i 6.4 at 4 ° C (non-spin), 23.6 C arid 32.5 c' (with spinning)

in 0.1 M NaCl, 10 mM phosphate, 2H,0, pH 5.05 show well-resolved spectra at low temperature in the duplex state and at high temperature in the strand state, with the resonances shifting as average peaks during the duplex-to-strand transition between 7.0 and 8.4 ppm (Fig. 4). This spectral region consists of the cytosine H-6 doublets and the adenine H-8, guanine H-8, adenine H-2 and thymine H-6 singlets. The adenine H-8 and guanine H-8 singlets can be differentiated on the basis of the former protons resonating at lower field at high temperature in the unstacked strand state. The purine H-8 and H-2 protons can be differentiated since the former can be deuterated following heating at high temperature while the latter protons exhibit the longest spin-lattice relaxation time. The thymine H-6 singlets are broader than the other singlets due to a small four-bond coupling to the thymine CH,-5 protons (Fig. 4). The corresponding 360-MHz proton NMR spectra of the octanucleotide in the same buffer between 5.2 and 6.4 ppm as a function of temperature are plotted in Fig. 5. The spectra are partially resolved, which permits a differentiation between the cytosine H-5 doublets and the sugar H-1' multiplets (Fig. 5). The thymine CH,-5 singlets resonate between 1.2 and 1.9 ppm in the octanucleotide as a function of temperature and are resolved over the entire temperature range. The temperature dependence of the base protons (adenine H-8 and H-2, thymine H-6 and CH,-5) in the two dA .dT base pairs in the octanucleotide duplex have been monitored through the melting transition as a function of temperature (Fig. 6). The resonances shift upfield as average peaks on duplex formation except for one adenine H-8 resonance which undergoes a small downfield shift on lowering the temperature. It is not possible to differentiate between the two

270

Helix-Coil Transition of the dG-dG-dA-dA-dT-dT-dC-dCDuplex dA(H-8)

tI

dGfH-8) dA ( H - 8 )

m

8.4

8.2

I

I

1

I

8 .O

7.8

7.6

7.4

1

7.2

7 .O

8 (ppm)

Fig. 4. The 360-MHz Fourier-transformproton NMR spectra (7.0 - 8.4 ppm) of 5.3 mM (strand concentration) dG-dG-dA-dA-dT-dT-dC-dC in 0.1 M NaCI. I0 mMphosphate, I mM EDTA, 'H,O, pH 5.05 at 14.2"C. 37.5 "C and 71.1 "C

,

I

6.4

6.2

6.0

5.8

I

I

I

5.6

5.4

5.2

8 (ppm)

Fig. 5 . The 360-MHz Fourier-transformproton NMR spectra (5.2- 6.4 ppm) of 5.3 rnM (strand concentration) dG-dG-dA-dA-dT-dT-dC-dC in 0.1 M NaC1, 10 mMphosphate, I m M EDTA, 2 H 2 0 .pH 5.05 at 14.2"C. 37.5 "C and 71.1 "C

dA.dT base pairs numbered 3 and 4 (Scheme 1) for each type of resonance. The chemical shift at 90 "C, the chemical shift difference between 90 "C and 10 "C, and the transition midpoint for the eight transitions in Fig. 8 are summarized in Table 2. The midpoints for opening of the central cluster of dA-dT base pairs in 5.3 mM (strand concentration) dG-dG-dA-dA-dTdT-dC-dC self-complementary duplex in 0.1 M NaCl, 10 mM phosphate buffer cover the range 46 3 "C (Table 2). We have checked the chemical shift dependence of the observable resonances through a heating, cooling and reheating cycle with the results for the upfield

thymine CH,-5 resonance plotted in Fig. 7. The data points fall on the same melting curve with excellent reproducibility and reversibility for the self-complementary octanucleotide. The temperature-dependent chemical shifts of the base protons (guanine H-8 and cytosine H-5 and H-6) in the two d G - d C base pairs in the dG-dG-dA-dAdT-dT-dC-dC duplex between 5 "C and 95 "C are plotted in Fig. 8. The cytosine H-5 and H-6 protons move upfield while smaller downfield shifts are observed at the guanine H-8 protons on duplex formation (Fig. 8). The cytosine H-5 and H-6 protons which continue to shift upfield between 30°C and 0°C are

D. J. Patel and L. L. Canuel 7.1-

72

1

t

'

1

27 1 I

j -

1.2

?

7.3 -

1

1

1

1

1

1

1

1

1

-

- 7.5 -8 7.6 -

-

-..-1.4 =._ !. Q

E

c v,

7.7 7.8

1

1.3

7.4 -

'0

1

1.5

3

._ E

-

6

1.6

7.9 -~ 1.7

8.0 -

,

8.11

I

I

I

I

I

,

I

I

48.3

1.8

0 2 0 4 0 63 8 0 1 0 0 0 20 40 60 80 1 0 0 Temperature ("C)

Fig. 6. The temperature-dependent chemical shifts of the adenine H-8 and H-2 and the thymine H-6 and CH3-5 resonances of 5.3 mM (strand concentration) dG-dG-dA-dA-dT-dT-dC-dC in 0.1 M NaCI. I0 mM phosphate, I mM EDTA, 'H,O. pH 5.05 between 5 "C and 95 "C

Temperature ("C)

Fig. 7.A plot of the chemical shift of the thymine CH3-5as a function of temperature during the first heating cycle (0).first cooling cycle (0).andsecond heating cycle (A) for 5.3 mM (strand concentration) dG-dG-dA-dA-dT-dT-dC-dC in 0.1 M NaCl, I0 mM phosphate, I mM EDTA, ' H 2 0 , pH 5.05

Table 2. The chemical shift parameters and transition midpoints for the melting of the dA ' dT base pairs in the dG-dG-dA-dA-dT-dTdC-dC octanucleotide duplex Octanucleotide strand concentration was 5.3 mM. The buffer was 0.1 M NaC1, 10 mM phosphate, 1 mM EDTA, 'H,O, pH 5.05 Proton

6 at 90 "C

A S for 90 "C to 10 "C

PPm dT CH3-5 dT CH3-5 dT H-6 dT H-6 dA H-2 dA H-2 dA H-8 dA H-8

1.722 1.824 7.473 7.553 7.968 7.988 8.128 8.258

-

I,

- 5.9 "C

0.444 0.271 0.293 0.148 0.662 0.324

-- 0.047 0.083

5.7

- 5.8

44.5 45 47 45.5 43.5 47.5 -

49

tentatively assigned to the terminal dG.dC base pairs (pair 1 in Scheme 1) and reflect the contributions from end-to-end aggregation of octanucleotide duplexes at low temperature [17]. The cytosine H-5 and H-6 protons of the other dG.dC base pair (position 2 in Scheme 1) show a decreased temperature dependence of their chemical shifts between 30 "C and 0 "C (Fig. 8) as do the adenine H-2 and thymine H-6 and CH,-5 protons of the dA-dT base pairs numbered 3 and 4 in Scheme 1 (Fig. 6). Spin-decoupling studies demonstrated that the downfield cytosine H-5 and H-6 protons in each set (at temperatures > 20 "C) were coupled to each other as were the corresponding upfield protons (at temperatures > 20 "C) in each set.

-

&!E

ro

- 6.0 8.0 -

I

6.1 I

I

I

I

I

I

0 x,406080100

I

I

I

I

I

I

I

I

0 20 40 60 80 100

Temperature ("C)

Fig. 8. The temperature-dependent chemical shifts of the guanine H-8. cytosine H-5 and cytosine H-6 resonances of 5.3 mM (strand concentration) dG-dG-dA-dA-dT-dT-dC-dC in 0.I M NaCI, 10 mM phosphate, I mM EDTA, ' H 2 0 ,pH 5.05

The chemical shift at 90 "C, the chemical shift difference between 90 " and 10 "C, and the transition midpoints for the base protons on the dG.dC base pairs in the octanucleotide in 0.1 M NaCl, 10 mM phosphate buffer are summarized in Table 3. The cytosine H-5 resonances could not be accurately monitored through the entire melting transition (Fig. 8) since they shift through the sugar H-1' region (Fig. 5). The eight sugar H-1 ' multiplets in the dG-dG-dAdA-dT-dT-dC-dC sequence are plotted as a function of temperature in Fig. 9. These resonances are spread out between 5.4 and 6.2 ppm in the duplex state and over a somewhat smaller chemical shift range (0.5 ppm) in the strand state. The sugar H-1' resonances cannot be assigned to specific positions in the sequence at

272

Helix-Coil Transition of the dC-dC-dA-dA-dT-dT-dC-dC Duplex

Table 3. The chemicalshift parameters and transition midpointsfor the melting of the dG . dC base pairs in the dG-dG-dA-dA-dT-dT-dC-dC octanucleotide duplex Octanucleotide strand concentration was 5.3 mM. The buffer was 0.1 M NaCI, 10 mM phosphate, 1 m M EDTA, 'H,O, pH 5.05

6 at 90 "C

Proton

A 6 for 90 "C to 10 "C

t,

PPm dC H-5 dC H-5 dC H-6 dC H-6 dG H-8 d C H-8

"C

6.074 6.095 7.844 7.883 7.760 7.800

0.389 0.455 0.264 0.289 - 0.024 - 0.020

-42.5 -45.5

I

5.5 5.6 -

-g 5.7

0

a

5 5.9

f,

6.0

40 60 80 Temperature ("C)

100

Fig. 10. The temperature-dependence of the line widths of the downfield ( 0 )and upfield ).( thymine H-6 resonance and ofthe downfield ( 0 )and upfield (@) thymine CH3-5 resonance of 5.3 m M (strand concentration) dG-dG-dA-dA-dT-dT-dC-dC in 0.1 M NaCI, 10 mM phosphate, I m M EDTA, , H 2 0 , p H 5.05 between 5 "C and 95 " C

5.8-

s

20

-

1

2

5t

6.1

tt

6.2

6.3

I I I I I / I I I I I I 0 20 40 60 80 100 Tern perat ure ("C)

Fig. 9. The temperature-dependent chemical shifts of the sugar H-I' resonances of 5.3 m M (strand concentration) dG-dG-dA-dA-dT-dTdC-dC in 0.1 M NaCI, I0 m M phosphate, I m M EDTA. 'H'O, p H 5.05, between 5 "C and 95 "C

this time. The two sugar H-1' resonances at highest field undergo the largest upfield shifts on duplex formation (Fig. 9). Nonexchangeable Proton Line Widths

We have evaluated the line widths of the base proton of dG-dG-dA-dA-dT-dT-dC-dC duplex during their average chemical shift changes through the melting transition with increasing temperature. The adenine H-8 and guanine H-8 resonances, which undergo small shifts during the duplex-to-strand transition, remain narrow between 20 "C and 95 "C and broaden gradually below 20 "C due to end-to-end aggregation and viscosity effects. By contrast, resonances exhibit-

ing large chemical shift differences between duplex and strand states broaden significantly during the melting transition, as can be observed for the adenine H-2 resonance at = 7.5 ppm and the thymine H-6 resonance at ~ 7 . ppm 2 in the octanucleotide spectrum at 37.5 "C (Fig. 4). We have plotted the line widths of the two thymine H-6 protons and the two thymine CH,-5 protons of 5.3 mM (strand concentration) dG-dG-dA-dA-dTdT-dC-dC duplex in 0.1 M NaCI, 10 mM phosphate buffer between 5 "C and 95 "C in Fig. 10. There is an increase in line widths between 20 "C and 50 "C with the widths reaching their maximum value at ~ 3 "C. 0 The extent of broadening is dependent on the chemical shift difference between states so that the upfield thymine CH,-5 which exhibits A6 = 0.44 ppm (90 "C to 10 "C) broadens to a greater extent than the downfield thymine CH,-5 (Fig. 10)which undergoesasmaller chemical shift difference of A 6 = 0.271 ppm. The upfield adenine H-2 resonance which undergoes the largest chemical shift difference between duplex and strand states is very broad in the spectrum at 37.5 "C in Fig. 4. The line widths changes correspond to uncertainty broadening contributions in the fast-exchange region and can be used to deduce the dissociation rate constants for the duplex-to-strand transition of the octa-

D. J. Patel and L. L. Canuel

273 4.6

I

.c

c

c .-

m

.a

-m

ir"

3.6

4.5

-

4.4

-

-E 4.3

-

3 4.2

-

a

A ' m i

~

i

~

l

l

l

l

~

l

~

u)

c

L

3.8

4.0 4.2 4.4 7 I (ppm)

4.6

c

4.1

-

$, 4.0

-

-m

Y)

0

._

5

3.9 -

3.8 -

4.8

3.7 -

Fig. 11. The 145.72-MHz Fourier-transjorm phosphorus NMR spectra (3.6 -4.8 ppm) of 5.3 mM (strand concentrution) dG-dGdA-dA-dT-dT-dC-dC in 0.1 M NaCI, I0 mM phosphate, I mM EDTA, 'H,O, p H 5.05 at 30.8 " C

3.6

'

I

0

20

I

'

I

I

I

I

I

I

40 60 80 100

nucleotide duplex. It should be noted that the maximum width for the upfield thymine CH,-5 resonance is observed at 32 "C for a transition with a midpoint of 44.5 "C. Backbone Phosphates There are seven internucleotide phosphates in the dG-dG-dA-dA-dT-dT-dC-dC octanucleotide sequence and the temperature dependence of their 145.72-MHz 31 P chemical shifts have been followed as a function of temperature for 5.3 mM (strand concentration) in 0.1 M NaCl, 10 mM phosphate, 1 mM EDTA, 2H20, pH 5.05. A typical spectrum recorded at 30.8 "C is presented in Fig. 11 with five resolvable peaks between 3.9 and 4.4 ppm upfield from internal standard trimethyl phosphate. The temperature dependence of the resolvable 31Pchemical shifts are plotted in Fig. 12 with the resonances spread over 0.6 ppm (3.9 - 4.5 ppm) in the duplex state at 5 "C and over 0.25 ppm (3.7-3.95 ppm) in the strand state at 90 "C. The resolvable resonances shift upfield on duplex formation with the largest change being 0.6 ppm between 90 "C and 5 "C (Fig. 12). We are unable to assign the resolvable 31P resonances to individual internucleotide phosphates in the octanucleotide sequence.

of 0.1 M NaCl and of ~ 2 "C3 on addition of 1 M NaCl to theoctanucleotide in 10 mM phosphate, 1 mM EDTA in H,O, pH 6.85, over the strand concentration range 0.2 mM to 8 pM (Fig. 2). Semilogarithmic plots of the octamer (t,)-' values from ultraviolet absorbance studies against the total strand concentration, c, , under the three buffer conditions are linear in the above concentration range and provide an estimate of the enthalpy AH" of the transition [2] according to the relationship

and the free energy AC" of the transition [2] according to the relationship AGO = Rt,lnc,.

The thermodynamic parameters for the dG-dG-dAdA-dT-dT-dC-dC melting transition in 0.1 M NaCI, 10 mM phosphate, 1 mM EDTA, H,O, pH 6.85, are AC;5 -28.4 kJ/mol, AH'' = - 216.7 kJ/mol, and AS"= -0.631 kJ mol-' K-'. The transition midpoints at the thymine CH,-5 and H-6 and the adenine H-8 and H-2 resonances of both dA.dT base pairs in 5.3 mM (strand concentration) dG-dG-dA-dA-dT-dT-dC-dC duplex deduced from NMR spectroscopy cover the range 46 i3 "C (Table 2). The corresponding values for the two d G . d C base pairs cannot be accurately monitored at the guanine nc=

DISCUSSION Concentration Dependence oft, Values The melting transition of the dG-dG-dA-dA-dTdT-dC-dC octanucleotide self-complementary duplex is stabilized by salt; this is demonstrated by an increase in the transition midpoint, t,, of z 13.5 "C on addition

(2)

r

274

H-8 resonance due to the small chemical shift change nor at the cytosine H-5 and H-6 resonances of terminal d G - d C base pairs due to end-to-end aggregation at low temperatures. The t , values for the cytosine H-5 and H-6 resonances of the penultimate d G - d C (pair 2 in Scheme 1) base pairs are ~ 4 4 1.5 f "C (Table 3). These t , values for the octanucleotide in 0.1 M NaCl, 10 mM phosphate, 1 mM EDTA, 'H20, pH 5.05 deduced by NMR spectroscopy are in good agreement with an extension of the ultraviolet absorbance t , values in this buffer to the NMR concentration of 5.3 mM in strands (Fig. 2). Such an approach correlating ultraviolet absorbance and NMR studies was reported earlier for the melting transition of tetraribonucleotide and hexaribonucleotide duplexes [9, lo]. These data require that the same melting transition is being monitored by the ultraviolet absorbance studies at low concentrations and the NMR studies at higher concentrations. The octamer sequence can either form a hairpin loop structure stabilized by two dG .dC base pairs or a self-complementary eight-basepair duplex. The NMR data to be discussed below demonstrate that the dA.dT base pairs are intact in the duplex state and hence the melting transition monitored by the ultraviolet absorbance profiles (Fig. 1) corresponds to a self-complementary duplex opening to strands. Hydrogen Bonding

The chemical shift of the pyrimidine H-3 WatsonCrick resonance in an isolated dA.dT base pair is predicted to resonate ~1 ppm downfield from the intrinsic chemical shift of the purine H-1 Watson-Crick proton in an isolated dG . dC base pair [19-221. Therefore, the two resonances at % 13.75 and % 13.85 ppm are assigned to dA .dT base pairs while the two resonances at ~ 1 2 . 7 5and ~ 1 2 . 9 ppm 5 are assigned to dG .dC base pairs in the octanucleotide spectrum in 0.1 M NaCl, 10 mM phosphate buffer, H,O, pH 6.85 at 4.7 "C (Fig. 3). The dG.dC resonance at ~ 1 2 . 9 ppm 5 broadens and shifts upfield to the greatest extent on raising the temperature to 10 "C, which permits assignment of this resonance to the guanine H-1 proton of the terminal d G - d C base pairs (pair 1 in Scheme 1). The fraying at the ends of the duplex [6,8,10,23,24] coupled with the exposure of one face of the terminal base pair to solvent accounts for the exchange characteristics observed for the Watson-Crick proton of this base pair. The remaining three resonances of the octanucleotide duplex are well resolved in the spectrum at 23.6 "C and broaden to the same extent on raising the temperature to 32.5 *C. They monitor the conversion of the octanucleotide duplex to strands. The exchangeable ring NH resonances of the octanucleotide broaden out

Helix-Coil Transition of the dG-dG-dA-dA-dT-dT-dC-dC Duplex Table 4. Calculated upfield shifis of duplex formation The calculations are based on ring-current and atomic diamagnetic anisotropy contributions at a distance of 0.34 nm for a B-DNA helix [25,26]. The numbering of the base pairs is that of Scheme 1 Proton

Upfield shift for base pair dG . dC (1) d G . dC (2)

dA dT (3)

dA . dT (4)

0.15 1.55 0.75 0.15 0.25

0.2 0.65 1.25 0.25 0.4

PPm dGH-8 dG H-1 dC H-6 dC H-5 dA H-8 dA H-2 dT H-3 dT H-6 dT CH,-5

0 0.4 0.1 0.45

0.15 0.9 0.15 0.4

at temperatures ( % 36 "C) below the transition midpoint monitored by the nonexchangeable protons (46 f3 "C) as has been reported previously for oligonucleotide duplex-to-strand transitions [4,23]. Two resolvable resonances are observed between 8.1 and 8.3 ppm for the octanucleotide with the downfield resonance exhibiting a temperature-independent chemical shift while the other resonance shifts upfield with increasing temperature. These protons probably originate in a side-chain-amino proton participating in Watson-Crick hydrogen bonds [21] and are assigned to the cytosine H-4 protons based on earlier studies of dG .dC containing tetranucleotide duplexes [17]. Base-Pair Overlap Geometries

The experimental shifts on formation of the octanucleotide duplex (Tables 2 and 3) may be compared with computed values based on ring current and atomic diamagnetic anisotropy contributions from nearestneighbor base pairs [25] for a B-DNA helix [26] (Table 4). A detailed comparison is not possible since we are unable to differentiate between the two dA.dT base pairs and only partially differentiate between the two dG.dC base pairs at this time. There are also limits to the accurate estimation of the chemical shifts in the duplex state due to aggregation at low temperatures and an estimation of the chemical shifts in the strand state due to the equilibrium between stacked and unstacked strands at high temperature. The chemical shift differences between 90 "C and 10 "C (Tables 2 and 3) thus represent an appropriate estimate of the experimental shifts associated with the octanucleotide duplex-to-strand transition. There is good agreement between the calculated upfield shifts at the thymine H-6 and CH,-5 protons on duplex formation (Table 4) and those observed

D. J . Patel and L. L. Canuel

275

Table 5. dG-dG-dA-dA-dT-dT-dC-dC octanucleotide duplex dissocialion rate constants at 32 "C (fd = 0.86) 5.3 mM strand concentration in 0.1 M NaCI, 10 mM phosphate, 1 mM EDTA, *H,O, pH 5.05

.

Resonance

Upfield dT CH3-5 Downfield dT CH,-5 Upfield dT H-6

A6

0.444 0.271 0.293

Line width observed

control

22 11 14

4.5 4.5 6.5

experimentally at these positions (Table 2). The adenine H-8 protons are predicted to exhibit small shifts on duplex formation (Table 4)as observed experimentally (Table 2). By contrast, the calculated adenine H-2 upfield shifts of 1.55 ppm and 0.65 ppm (Table 4) overestimate the experimental values of 0.66 ppm and 0.32 ppm at these positions (Table 2). Similarly, the exchangeable thymine H-3 protons are calculated to differ by 0.5 ppm (Table 4) between the two dA.dT base pairs in the duplex state while they differ by 0.1 ppm in the experimental spectrum (Fig. 3). We do not attempt to compare the experimental and calculated values for the resonances of the terminal dG.dC base pairs (position 1 in Scheme 1) since the contributions of end-to-end aggregation cannot be partitioned from the duplex-to-strand shifts. The calculated upfield shifts at the cytosine H-5 and H-6 protons of the penultimate dG.dC base pairs (position 2 in Scheme 1) are 0.4 ppm and 0.15 ppm (Table 4) compared to experimental values of 0.39 ppm and 0.26 ppm (Table 3). The calculations predict small shifts on duplex formation at the guanine H-8 resonances and this conclusion is observed experimentally. A further comparison between the experimental shifts and those computed for B-DNA overlap geometries require additional knowledge for differentiation between the two types of dA.dT base pairs. Duplex Dissociation Rate Constants

Several proton resonances that undergo chemical shift differences of >0.2 ppm and shift as average peaks during the duplex-to-strand transition of dGdG-dA-dA-dT-dT-dC-dC in 0.1 M NaCl, 10 mM phosphate solution exhibit uncertainty broadening contributions during the transition (Fig. 4). These line width changes are proportional to the chemical shift separations with the largest changes observed at the upfield adenine H-2 resonance which undergoes the largest shift on duplex formation. The excess width can be related to the duplex dissociation rate constant according to the relationship : Excess width =4n~2fi(A6)2(z, + z),

%fd

= zdfs

9

kd

1.9 x 103 1.95 x lo3 1.9 x lo3

where& andf, are the fraction of duplex and strands, respectively, A 6 is the chemical shift separation between duplex and strands (in Hz), and z, and z, are the lifetimes of the duplex and strand states, respectively. The duplex dissociation rate constant for the octanucleotide, k , = (zd)-', has been evaluated to be z 1.9 x lo3 s p l at 32 "C (Table 5) at the temperature (32 *C,fd = 0.86) corresponding to maximum broadening of the upfield and downfield thymine CH,-5 resonances, and upfield thymine H-6 resonance. Stacked- Unstacked Strand Equilibrium

The duplex-to-strand transition monitored by the nonexchangeable protons of thymine H-6 and CH,-5 (Fig. 6) and cytosine H-6 and H-5 (Fig. 8) approaches completion by 55 "C. The ,'P resonances shift downfield (relative to internal trimethyl phosphate) between 10 "C and 50 "C during the duplex-to-strand transition (Fig. 12). Further, all the 31P resonances continue to shift downfield on increasing the temperature above 50 "C,conditions under which the equilibrium between stacked and unstacked strands shifts towards the latter (Fig. 12). The above trends in 31Pshifts for the stacked-tounstacked strand transition have been previously observed for the dG .dC containing tetradeoxynucleotide self-complementary duplexes dC-dG-dC-dG, dC-dCdG-dG and dG-dG-dC-dC [17] and for the synthetic DNA poly(dA-dT) [7]. They have been assigned [17] to changes in the w,w' angles on conversion from stacked strands (gauche,gauche) to unstacked strands (gauche, trans) [27 - 291. The 0.6-ppm variation in the 31Pshifts of the seven internucleotide phosphates in the duplex state (Fig. 11 and 12) may indicate the existence of sequence-dependent variations in the w ,w' angles in the double-helical state. REFERENCES 1 . Kearns, D. R. (1977) Annu. Rev. Biophys. Bioeng. 6,471-523. 2. Kallenbach, N. R. & Berman, H. M. (1977) Q. Rev. Biophys. 10, 138 - 236. 3. Cross, A. D. & Crothers, D. M. (1971) Biochemistry, 10,40154023.

2 76

D. J. Patel and L. L. Canuel: Helix-Coil Transition of the dG-dG-dA-dA-dT-dT-dC-dC Duplex

4. Crothers, D. M., Hilbers, C. W. & Shulman, R. G. (1973) Proc. Nutl Acud. Sci. U.S.A. 70, 2899 - 2901. 5. Early, T. A., Kearns, D. R., Burd, J. F., Larson, J. E.& Wells, R. D. (1977) Biochemistry, 16, 541 -551. 6. Patel, D. J. (1974) Biochemistry, 13, 2396 - 2402. 7. Patel, D. J. & Canuel, L. L. (1976) Proc. Nutl Acud. Sci. U . S . A . 73, 674 - 678. 8. Kallenbach, N. R., Daniel, W. E.,Jr. & Kaminker, M. A. (1976) Biochemistry, 15, 1218- 1224. 9. Arter, D. B., Walker, G. C., Uhlenbeck, 0. C. & Schmidt, P. G. (1974) Biochem. Biophys. Res. Commun. 61, 10891094. 10. Borer, P. N., Kan, L. S. & T’so, P. 0. P. (1975) Biochemistry, 14,4847 - 4863 & 4864 - 4869. 1 1 . Heller, M. J.. Tu, A. T. & Maciel, G. E. (1974) Biochemistry, 13, 1623-1631. 12. Patel, D. J. (1978) Eur. J . Biochem. 83, 453-464. 13. Patel, D J. (1978) J . Po/j,m. Sci. Synp. 62, 117-141. 14. Patel, D. J. & Canuel, L. L. (1977) Proc. Natl Acud. Sci. U.S.A. 74, 5207-5211. 15. Zimmer, Ch. (1975) Progr. Nucleic Acids Res. Mol. Biol. I S , 285 - 318. 16. Wartell, R. M., Larson, J. E. & Wells, R. D. (1975) J . Biol. Chem. 250. 2698 - 2702.

D. Patel and L. L. Canuel, Bell Laboratories, 600 Mountain Avenue, Murray Hill, New Jersey, U.S.A. 07974

17a.Pate1, D. J. (1976) Biopolymers, IS, 533-558. 17b.Patel. D. J. (1977) Biopolymers, 16, 1635- 1656. 18. Kearns, D. R., Patel, D. J. & Shulman, R. G. (1971) Nature (Lond.) 229, 338 - 339. 19. Patel, D. J. & Tonelli, A. E. (1974) Biopolymers, 13, 19431964. 20. Kearns, D. R. & Shulman, R. G. (1974) Acc. Chem. Rex 7, 33 - 39. 21. Kearns, D. R. (1976) Progr. Nucleic Acid.7 Res. Mol. Biol. 18, 91 - 149. 22. Reid, B. R. & Hurd, R. E.(1977) Ace. Chem. Res. 10,396-402. 23. Patel, D. J. & Hilbers, C. W. (1975) Biochemistry, 14, 2651 2656. 24. Hilbers, C. W. & Patel, D. J. (1975) Biochemistry, 14, 26562660. 25. Giessner-Prettre, C. & Pullman, B. (1976) Biochem. Biophys. Res. Commun. 70. 578 - 581. 26. Arnott, S. & Hukins, D. W. L. (1973) J . Mol. Biol. 81, 93- 105. 27. Tewari, R., Nanda, R. K. & Govil, G. (1974) Biopolymers, I S , 201 5 - 2035. 28. Olson, W. K. (1975) Biopolymers, 14, 1797-1810. 29. Yathindra, N. & Sundaralingam, M. (1974) Proc. Nut1 Acud. Sci. U . S . A . 71, 3325 - 3328.