- Email: [email protected]
Triple helix formation by oligopurine-oligopyrimidine DNA fragments
J. Mol. Biol. (1990) 213,833-843
Triple Helix Formation by Oligopurine-oligopyrimidine D N A Fragments Electrophoretic and Thermodynamic Behavior G i o r g i o M a n z i n i , L u i g i E. X o d o , D a n i e l a G a s p a r o t t o
Department of Biochemistry, Biophysics and Macromolecular Chemistry University of Trieste, 1-34127, Trieste, Italy Franco Quadrifoglio
Institute of Biology, Faculty of Medicine University of Udine, 1-33100, Udine, Italy G i j s A . v a n der M a r e l a n d J a c q u e s H. v a n B o o m
Gorlaeus Laboratories, State University P.O. Box 9502, 2300 RA Leiden, The Netherlands (Received 10 August 1989; accepted 24 January 1990) The 26met oligodeoxynucleot_ide d(GAAGGAGGAGATTTTTCTCCTCCTTC) adopts in solution a unimolecular hairpin structure (h), with an oligopurine-oligopyrimidine (Pu-Py) stem. When h is mixed with d(CTTCCTCCTCT) (sl) the two strands co-migrate in polyacrylamide gel electrophoresis at pH 5. If sl is substituted with d(TCTCCTCCTTC) (s2), such behavior is not observed and the two strands migrate separately. This supports the suggestion of the formation of a triple-stranded structure by h and sl (h :sl) but not by h and s2, and confirms the strand polarity requirement of the third pyrimidine strand, which is necessary for this type of structure. The formation of a triple helix by h : sl is supported by electrophoretic mobility data (Ferguson plot) and by enzymatic assay with DNase I. Circular dichroism measurements show that, upon triple helix formation, there are two negative ellipticities: a weaker one (As----80M-lcm -1) at 242nm and a stronger one ( A s = 2 1 0 M - l c m -1) at 212nm. The latter has been observed also in triple-stranded polynucleotides, and can be considered as the trademark for a P y : P u : P y DNA triplex. Comparison of ultraviolet absorption at 270 nm and temperature measurements shows that the triple-stranded structure melts with a biphasic profile. The lower temperature transition is bimolecular and is attributable to the breakdown of the triplex to give h and sl, while the higher temperature transition is monomolecular and is due to the transition of hairpin to coil structure. The duplex-to-triplex transition is co-operative, fully reversible and with a hyperchromism of about 10~/o. The analysis of the melting curves, with a three-state model, allows estimation of the thermodynamic parameters of triple helix formation. We found that the duplex-to-triplex transition of h: sl is accompanied by an average change in enthalpy (less the protonation contribution) of -73(_+5)kcal/mol of triplex, which corresponds to -6-6(+_0-4)kcal/mol of binding pyrimidine, attributable to stacking and hydrogen bonding interactions.
in Escherichia coli and some bacteriophages they are not. In nature, Pu-Py sequences are often found
1. Introduction Oligopurine-oligopyrimidine (Pu-Pyt) sequences are present in prokaryotes and eukaryotes. While in eukaryotic genomes the Pu-Py sequences are overrepresented (Behe, 1987; Birnboim et al., 1979),
0022-2836/90/120833-11$03.00/0
tAbbreviations used: Pu, purine; Py, pyrimidine; W-C, Watson-Crick; u.v., ultraviolet; c.d., circular dichroism; PAGE, polyacrylamide gel electrophoresis. 833
© 1990AcademicPressLimited
834
G. M a n z i n i et al.
within the regulatory regions of DNA, such as upstream from the genes (Larsen & Weintraub, 1982) and in recombination hot-spots (Htun et al., 1984). For instance, Pu-Py sequences have been mapped in the chicken fl-globin gene (Schon et al., 1983; Evans et al., 1984), the rat preproinsulin II gene (Evans et al., 1984) and the human thyroglobulin gene (Christophe et al., 1985). A peculiar property of natural Pu-Py sequences is that they are hyperreactive to $1 nuclease. This implies that they may retain a non-canonical Watson-Crick (W-C) double-helical structure. Two major unusual structures have been proposed for Pu-Py sequences. The first is based on a double-stranded helix and explains the $I hypersensitivity of (AG)n sites to nuclease $1 as resulting from a structure in which W-C A" T alternates with Hoogsteen G" C+ basepairs (Pulleyblank et al., 1985); the second structure assumes the formation of triple helices in correspondence with P u - P y sequences (Lyamichev et al., 1986; Mirkin et al., 1987). The formation of DNA triple helices by polynucleotides of defined sequences was widely studied in the 1960s (Riley et al., 1966; Morgan & Wells, 1968). These structures consist, in general, of one oligopurine and two oligopyrimidine strands. X-ray fiber diffraction studies revealed that an oligopurine strand and an oligopyrimidine strand are W-C base-paired, forming a double-stranded helix whose major groove is occupied by a second oligopyrimidine strand, oriented in parallel to the oligopurine strand, and held together by Hoogsteen base-pairs (Arnott & Selsing, 1979). Both arrangements require the hemiprotonation of cytosine residues. There is a large body of evidence supporting the triple helix model: in plasmids the triple helix formation is favored by low pH, superhelical stress and mirror symmetry of the Pu-Py segments (Mirkin et al., 1987). Interruptions in the mirror repeat have been demonstrated to inhibit the triple helix formation substantially, probably because they prevent Hoogsteen base-pairing (Mirkin et al., 1987). Clear evidence of intramolecular DNA triplexes in Pu-Py sequences inserted into plasmids has been obtained by Hanvey et al. {1988), using the specific chemical probes osmium tetroxide and diethyl pyrocarbonate to investigate base-pairing. However, there is little information about triple helix formation by ol~godeoxynucleotides in the absence of superhelical stress. It has been shown that oligopyrimidines bind in the major groove of DNA at homopurine-homopyrimidine stretches by forming local triple helices (Moser & Dervan, 1987; Francois et al., 1988). In addition, two nuclear magnetic resonance analyses on triplexes formed by short oligodeoxynueleotides have been reported (Rajagopal & Feigon, 1989; De los Santos et al., 1989). As yet, the biological function of triple-stranded DNA is not clear; nevertheless, there is a growing conviction that this unusual structure may play an important role in gene expression. An attractive function for the Pu-Py sequences may be that these are specific sites for single-stranded binding proteins
(Lee et al., 1984) or conversely that, by triplex formation, they prevent the binding of ligands of specific Pu-Py DNA stretches. Hence, it is therefore necessary to investigate the physical properties of triple-stranded DNA. We have studied the energetics as well as the influence of base sequence on DNA triple helix formation. The results are presented in this paper.
2. Materials and Methods
(a) Materials The oligodeoxynucleotides investigated in this study are: h 5' d(GAAGGAGGAGATTTTTCTCCTCCTTC); 81 5' d(CTTCCTCCTCT); s2 5' d(TCTCCTCCTTC}. These DNA fragments, as well as others of length varying from 12 to 20 base-pairs, used for reference in electrophoresis experiments, were synthesized in solid phase, according to a modified phosphotriester method as described (Van der Marel et al., 1981). After base deprotection, purification was carried out with Sephadex G-50 resin, by eluting the column with 0'05M-tetra-ethylammonium bicarbonate. Purity was confirmed by both high-pressure liquid chromatography and denaturing electrophoresis (Frank & Koster, 1979). Electrophoresis materials, i.e. acrylamide, bisacrylamide, N,N,N',N'tetramethylethylenediamine and ammonium persulfate, were purchased from Serva Feinbioehemica, and Stains All dye from Sigma. Bovine pancreas deoxyribonuclease I (DNase I) was obtained from Boehringer. (b) Buffer Ultraviolet (u.v.) absorption and electrophoretic experiments were carried out in 0"1 M-sodium acetate (pH 5), 50 mM-NaC1, 10 mM-MgC12. Circular dichroism (c.d.) measurements were done in 1 mM-sodium acetate (pH 5), 100 mM-NaClO,. {c) Polyacrylamide gel electrophoresis PAGE was carried out on g e l s (20cm or 10 cm x 15 cm × 0"15 cm) obtained from buffer solutions containing 20% (10%, 15% or 250/0 (w/v)) acrylamide, 3-3% (w/v) bisacrylamide, 0-07% (w/v) ammonium persulfate. The stacking gel contained 5% acrylamide. In general, the electrophoresis was performed at a constant voltage of 10 V/cm with a current of 20 to 30 mA. The amount of samples loaded on the gel was about 2 to 4 #g. Bromphenol blue dye was used as a marker. The bands were stained with Stains All dye in water/formamide (1 : 1, v/v) and the gels were photographed after 1 h. (d) Circular dichroism c.d. spectra were obtained with a Jasco J-500A speetropolarimeter, equipped with a thermostatically controlled cuvette holder that allows measurements at controlled temperature. The diehrograph was connected to a Jaseo DP500 N data processor. Spectra are presented as AS----(sL--sR) in units of M-I cm-l. The spectra were obtained from sample solutions at a strand concentration of about I00 ~M. Absorbance of the solvent in the short
Triple Helix Formed by OHgodeoxynucleolide8 wavelength region of the spectrum (200 to 230 nm) was minimized by using a 1 mm cuvette and 1 raM-sodium acetate, 100 mM-NaCl04 (pH 5) as buffer.
(e) DenaturaHon experiments The denaturation experiments were performed on a Cary 219 (Varian) spectrophotometer. The temperature was increased at a rate of 0-5 deg.C/min with a Haake PG20 temperature programmer, connected to a Haake water-circulating bath equipped with a refrigerator. The sample temperature was detected by means of a Varian thermistor immersed in an oil-containing cuvette situated in the sample turret. The melting temperature was monitored at 270, 264 and 250 nm and the oligomer concentrations were determined by u.v. absorption at 260 nm at t--90°C, assuming an extinction coefficient of 7500, 8500, 15,000 and 12,500 M-1 cm -1 for C, T, A and G, respectively, in the denatured state.
5'
GAAGGA
5' AA
A
835
GA
/T4 %
T , ,T8 ,TN I N
TT
,
A A"'-~T4
.'CTTCCTCCTCT..
GAA A A T4 3,CT T ~ T ~ T ~ T , . . _ . ~
*.s!
5'CT TCCTCCTCT
st.
5'TCTCCTCCT TC
s2
/~
CH3 ~
CH3
H
H
(g) Ultraviolet absorption spectra versus pH The u.v. absorption spectra of sl at 45°C and of cytidine at fixed temperatures in the range 20 to 42°C were recorded at pH values from 2 to 7. The pH was varied by adding small portions of 5 M-NaOH to a 0-1 M-acetic acid solution, initially brought to pH 2 with a small amount of HCI. The pH was measured at the same temperature for each spectrum using a thermostatically controlled pH meter system (Radiometer, priM52), equipped with a glass electrode, which had been calibrated at each considered temperature with standard buffer (Radiometer, pH 4"01).
T
A
T
C
G
C
Figure 1. Secondary structures accessible to h and an equimolar mixture of h and 81 (h:sl). In the triplestranded structure, sl occupies the major groove of the double-helical domain of hairpin h, forming TAT and CGC+ base triads, sl and the oligopurine strand of h are oriented in parallel. Strand s2 should fully recognize the major groove of h when it is antiparallel to the oligopurine strand. This orientation does not allow a triple-stranded structure.
3. D e s i g n o f T r i p l e - s t r a n d e d D N A In order to design a simple triple-stranded DNA system we exploited the intrinsic propensity of p a r t l y self-complementary sequences to assume a unimolecular hairpin structure (Xodo el al., 1986, 1988; I k u t a et al., 1986; H a a s n o o t el al., 1986; Senior et al., 1988). The sequence of h is characterized b y a 2-fold s y m m e t r y axis at its center, with the 5' arm comprising only purine residues, while the 3' arm includes only pyrimidine residues. Thus, h can a d o p t either a linear dimeric form with a bulge of four thymidine residues at its center or a folded unimolecular hairpin form with a four-T loop and a P u - P y stem of 11 base-pairs. The quasi-palindromic sequences of the h t y p e have been shown to a d o p t mainly a hairpin structure, over a wide range of temperatures, nucleotide concentrations and ionic strengths (Xodo et al., 1988; I k u t a et al., 1986; Senior et al., 1988). The major groove of the double helix domain of hairpin h should be recognized by sl, whose sequence is complementary to the oligopurine stretch of h when both strands are oriented in parallel. According to previous results, we expected for an equimolar mixture o f h and sl (h:sl) a triplestranded structure through the formation of T- A- T and C-G" C + base triads,, as illustrated in Figure 1 (Lyamichev et al., 1986; Mirkin et al., 1987; Wells et al., 1988). B y contrast, when 81 is substituted with 82 the triple-stranded structure should not form, since s2 and the oligopurine stretch of h are complem e n t a r y when t h e y are antiparallel to each other.
In conclusion, of the two strands sl and s2, only the first should bind in the major groove of the doublestranded domain of hairpin h, forming Hoogsteen base-pairs with the oligopurine strand. Another i m p o r t a n t reason why we chose to build a triplestranded DNA using hairpin h is t h a t we expected, due to the high stability of hairpin structures, a two-step melting profile for the complex h : sl, with triplex disruption and hairpin denaturation well resolved.
4. Results and Discussion
(a) Electrophoresi8 The potential of h:81 to form a triple helix was assayed b y non-denaturing P A G E at p H 5. This slightly acidic condition was necessary to obtain the triple-stranded structure. In fact P A G E experiments at p H 7 did not show an appreciable a m o u n t of complex h : 81. Figure 2(a) shows a P A G E analysis, using 2 0 % (w/v) polyacrylamide, of h, sl, 82, h : s l and h+s2. Lane g shows t h a t the 26met h migrates as a hairpin, exhibiting a mobility slightly faster t h a n t h a t of a 14mer (lane 1) and a 12mer (lane h). When h was mixed with 81 in a 1 : 1 strand ratio, a new band of low mobility (triple helix structure), comparable to t h a t of a 20mer running in lane i, appeared in the gel (lane f). I f the m i x t u r e
G. Manzini et al.
836
st or m Self-complexesq
h'$1
|
s l or s 2
(a)
$ f Or $ 2
Self-complexes A:Sl
h s f or s 2
(b)
Figure 2. (a) Polyaerylamide gel electrophoresis of h, sl, s2, h :sl and h + s2 using 20 % polyacrylamide. In each lane approximately 3 pg of DNA was loaded. Lanes h, i and j contain 3 reference duplexes: a 12mer, a 20mer and a 14mer, respectively. Lane g contains h, while lanes e and f contain h : sl, with and without thermal pretreatment. Strands 81 and s2 were loaded in lanes c and d, respectively. Lanes a and b contain the mixture h+s2, with and without thermal pretreatment. The 20% PAGE was in 0.1 M-sodium acetate (pH 5), 50 mM-NaCl, 10 mM-MgCl2. (b) PAGE of sl, s2, h, h:sl and h+82 using 25% polyacrylamide. Strands sl and s2 have been loaded in lanes a and b after being heated at 60°C, and in lanes c and d without any thermal treatment. Lanes e and f contain h+s2 and h : sl, where s2 and sl are in excess with respect to h. Three reference duplexes are in lanes g, h and i: a 12mer, a 20met and a 14mer, respectively. Gel conditions are the same as those reported for the 20 % PAGE.
of h and sl (sl in slight excess) was heated and then cooled to room temperature just prior to loading, the h band disappeared in favor of the low mobility species (lane e). However, when h was mixed with 82 (lane b, in equimolar ratio; lane a, in excess of s2) no influence on the mobility of band h was observed. A close inspection of lanes a and b revealed t h a t s2 migrated separately from h, and, surprisingly, not as one high mobility species, b u t together with twin bands of lower mobility. Lanes c and d show the electrophoretic profiles of sl and s2 alone. These bands are not stained well because sl and s2 do not form W-C secondary structures. However, b y using larger quantities of both sl and
s2, we obtained a clear electrophoretic p a t t e r n of the molecular species formed by sl, s2, h : s l and h + s 2 (Fig. 2(b)). Now it appears t h a t both s l and s2 migrate with three bands (lanes c and d): one with high mobility, corresponding to single-stranded sl and s2, and two close bands of low mobility, which, we believe are due to self-complexes of both sl and s2, stabilized by C ' C + base-pairs as found b y Edwards et ed. (1988) and Sarma et al. (1986) for C,T strands a t low pH. Thus, electrophoresis shows t h a t s l and not s2 is able to recognize the major groove of h and form a triple-helical structure. A further eleetrophoretic characterization of the mixture h : sl was obtained by studying its mobility
Triple Helix Formed by Oligodeoxynucleotides
. . . .
w r - - A -
837
__
\
0
._1 0.1
i.
% x Q
%
|
I0
!
I
20
SO
Acrylarnide (%)
Figure 3. Ferguson plots of an equimolar mixture of h and sl (h:sl) compared with a 20mer and a 12mer DNA duplex. The logarithm of the relative mobility of (O) h:sl, (A) 20mer and {A) 12mer (with respect to that of hairpin h) is been plotted against the concentration of acrylamide in the gel. The slope of these linear plots allows estimation of the retardation coefficient of the 3 samples.
as a function of gel concentration. The results are shown in Figure 3 in the form of a Ferguson plot (Roadbard & Chrambach, 1971): log Rr = log R d 0 ) - KRC, where Rf is the apparent mobility of the charged molecular species in the gel of acrylamide composition C and Rf(0) is the free relative mobility of the charged molecular species for C-*0. The slope, Ks, of the plot provides an estimate of the retardation coefficient of the charged molecular species, a parameter related to the size and shape of the molecular species itself (Roadbard & Chrambach, 1971). F r o m this, it is evident t h a t the 11-triad triplex behaves in a manner similar to t h a t of a 20mer fragment, i.e. with a retardation coefficient higher than t h a t of a 12mer DNA duplex of approximately the same length of h. The plots seem to converge at the same y-intercept value, indicating t h a t the free mobilities of duplex and triplex structures are similar. (b) Effect of DNase I on h and h : sl The effect of DNase I on both hairpin h and complex h :sl was investigated by analyzing the products using denaturing gel electrophoresis. Figure 4 shows the results of PAGE (20% acrylamide) of h and h: sl after t h e y have been incubated with DNase I at 30°C for 1-5 hours. Lanes a and f contain untreated h and h:sl, whereas lanes b to e
Figure 4. Denaturing gel electrophoresis of h and h : 81 after 1"5 h of incubation at 30°C with increasing amounts of DNase I. The incubation buffer was 0"1 M-sodium acetate (pH 5), 50 mM-NaCl, 10 mM-MgCl~. Lanes a and f contain untreated h and h:sl, respectively. Lanes b to e and g to 1contain h and h : sl digested with DNase I at the following /~g DNA//~g DNase I ratios: 11, 5-5, 3-7, 2"7. PAGE conditions: 25% acrylamide, 6 M-urea, 50 mM-Tris • HC1 (pH 8"3), 1 mM-EDTA, t--30°C, 10 V/cm.
and g to 1 contain h and h:sl, respectively, after incubation with increasing amounts of DNase I. The electrophoretic pattern obtained clearly shows that, while h is degraded, h : sl is resistant to the enzyme. This suggests t h a t h :81 forms a triple-stranded structure; such a structure has been reported to be a poor substrate for DNase I (Lee et al., 1984). (c) Circular dichroism c.d. spectroscopy is very suitable for distinguishing between different DNA structures. The formation of triple-stranded DNA by h : s l was followed also by c.d. Figure 5 shows the c.d. spectra of h, sl and the mixture h:sl. The following observations m a y be made. (1) The c.d. spectrum of h is consistent with t h a t of A-DNA, showing a large ellipticity (Ae= 150 M-1 cm -1) at 279 nm and a smaller one (Ae=70 M-1 em -1) at 242 nm (Riazance et al., 1985). (2) The c.d. spectrum of 81, obtained aider thermal treatment, has a shape typical of singlestranded DNA (Lewis & Curtis Johnson, 1974). (3) The c.d. spectrum of h :sl exhibits a large ellipticity (Ae=220 M-1 cm -1) at 279 nm, reminiscent of A-DNA, a crossover at 262 nm and two negative bands at shorter wavelengths: a weak band at 242 nm (A~=80M -1 cm -1) and a strong one at 212 nm (As ----210 M-1 cm-1). Subtracting the c.d. spectra of h and sl from t h a t of h : 81, the ellipticities arising upon DNA triple-stranded formation were obtained. Note t h a t this spectral difference shows clearly t h a t a new DNA structure is formed
838
G. Manzini et al.
3°°I 200
IOC
.o
oe='=~ = "m-% C •o
T
d
T
- .....
I/,.,.,"
II
-
-20(
220
250
280
Wovelength (nm}
Figure 5. c.d. spectra for (curve a), h:sl, (curve b), h, (curve c), sl and (curve d), h : s l - ( h + s l ) . The spectra were recorded at 25°C in 1 mM-sodium acetate (pH 5), 100 mM-NaCI04. The eltipticity is given in units of M-1 em -1, where M is expressed in mol DNA strands/l. by mixing h and sl; this is particularly indicated by the presence of a strong negative band at 212 nm. It is noteworthy that the c.d. spectrum of h :sl is rather similar to those reported by Lee et al. (1979) for triple-stranded polynucleotides. This ellipticity is not observed in A-DNA, B-DNA or Z-DNA (Riazance et al., 1985; Behe, 1986). Hence we are tempted to consider this short wavelength c.d. band as a trademark for P y : P u : P y triple-stranded DNA. Interestingly, the c.d. spectrum of a l : l mixture of h and s2 turned out to be nearly the sum of individual h and s2 spectra, lacking, as expected, an increased negative band at 212 nm. This suggests that no triple helix is formed with this DNA system, in agreement with the electrophoretic results. (d) Ultraviolet spectroscopy and melting experiments u.v. absorption spectra of h, sl and the mixture h: sl, at 25°C and 90°C, have shown that the magni-
rude of the thermally induced hyperchromic effect for h is 16% and for sl 1%, while for h : s l it is 27%. Thus, it follows that the duplex-to-triplex process is characterized by a hypochromism of about 10%. Since this process is accompanied by a partial protonation of the sl cytosine residues (see below), which in turn is a hyperehromic process at 270 nm, the total duplex-to-triplex hypochromism can be evaluated to be around 12%. This suggests that the binding of the pyrimidine strand in the major groove of h, via Hoogsteen base-pairing, results in considerable stacking interactions. The duplex-triplex conformational transition of h : s l was carefully analyzed by thermal denaturation experiments at 270, 264 and 250 nm. Figure 6 shows representative melting curves for h : s l , followed at 264 nm, both in denaturation and renaturation. The shape of the melting profiles is clearly biphasic. We call the transformation occurring at lower and higher temperatures transition 1 and transition 2,
Triple Helix Formed by Oligodeoxynucleotides 0.85
839
b
0.7S B 0.75 0.71
0.67 I
30
I
I
40
I
I
50 60 Ternperoture (°C)
I
70
80
Figure 6. Absorbance at 264 nm versus temperature profiles for the equimolar mixture h :sl in 0"1 M-sodium acetate (pH5), 50 mm-NaCl, 10 m~l-MgC12, in denaturation and renaturation. The nucleotide concentration of h:sl was 0"175 mm/base. Some points (A) from the best-fits of the experimental curves to eqn (5) are reported. The adjusted parameters AH T and ASv with relative errors are reported in Table 1. In each best-fit: (computed (A)--experimental (A))/experimental (A) <0"40/0 at each temperature. respectively. An insight into the molecularity of these transitions was obtained b y v a r y i n g the nucleotide concentrations of h : sl. We observed t h a t the melting t e m p e r a t u r e of transition 2, tl/2 = 72°C, remained c o n s t a n t over an eightfold increase in the nucleotide concentration, suggesting a unimolecular reaction for this transition. On the contrary, the tl/2 of transition 1 was positively affected by increasing the nucleotide concentration. T a b l e 1 gives the tl/2 values a t each concentration considered. This behavior is consistent with a bimolecular process for transition l, which we attribute, on the basis of P A G E and c.d. results, to triple helix formation. This conclusion is confirmed unequivocally by the fact t h a t hairpin h alone melts with a profile corresponding e x a c t l y to transition 2. I t is n o t e w o r t h y t h a t transition 1 is rather co-operative (it spans over a range of a b o u t 15 deg.C) and is fully revers-
ible at a scan rate of 0"5 deg.C/min (Fig. 6). F u r t h e r more, the fact t h a t the duplex-to-triplex transition is quickly reversible suggests t h a t a r a t h e r low activation energy m a y be involved in the process. More compelling evidence t h a t transition 1 is due to the f o r m a t i o n of a triple-stranded structure between h and sl comes from the observation t h a t this transition is absent in the m i x t u r e of h and s2, although s l and s2 have the same central TCCTCCT-stretch. Either the shortness of the central stretch or the hindrance of the dangling ends could be responsible for such behavior.
(e) Thermodynamics of triple helix formation Considering t h a t the processes of triple helix
(h : sl) and hairpin (h) formation are fully reversible, and t h a t their constituent strands are r a t h e r short,
Table
1
Thermodynamic parameters for the triplex h : s l to h + s l transition in sodium acetate buffer Concentration
(mM/base) 0.052 0.086 0.127 0.169 0,175 0,175 0,412 Average AHT Standard deviation
ill2 (°C) 48-0 52,0 52-0 52,5 53,5 53,0 55,0
AHr (kcal/mol)
A8z (e.u.)
Curvet
85,2( + 1.4)~ 80,1 ( + 1.2) 82,2(+ 1 - 2 ) 86,6( + 1,0) 81,4(___ 1,1) 84,8( + 2,0). 90,3( + 1.0)
238.4( + 4-1)~ 220,6( + 3.8) 227,8(+3,7) 241,5( ___2,7) 224-7(__+4.0) 235~)( + 6.0) 249,9( + 3.2)
R D D D R D D
84"3 kcal/mol of h : 81
+5 keal/mol
Sodium acetate buffer is: 0"1 M-SOdiumacetate (pH 5), 10 mM-MgCI2,50 mM-NaCI.Q/2 is the melting temperature; Aftr is the enthalpy change and A8 T the entropy change for the transition.
t R, renaturation; D, denaturation; 1 cal ffi4-184 J; e.u., entropy units. SThe uncertainties of AH T (and A8 T) in parentheses are given by the Marquardt (Bevington, 1969)
algorithm and refer to a single best-fit curve. They are lower than the final standard deviation over the 7 best-fits,since they do not reflectthe high correlationexistingbetween A H r and A 8 T
840
G. Manzini et al.
we could describe transitions I and 2 with a threestate model. The thermodynamic parameters for transition 2, i.e. for the denaturation of hairpin h, were obtained by analyzing the melting curves of sample solutions containing only h. By fitting these curves with a two-state model (DePrisco Albergo et al., 1981), using a non-linear least-squares program (algorithm of Marquardt; Bevington, 1969) we obtained the following values: tl/2=72(+1)°C, AH H= 85( _+4) kcal/mol, AS s = 246( _-4-i-0) e.u. (1 cal =4-184 J). The calculated denaturation enthalpy value for h, based on nearest-neighbor interactions (Breslauer et al., 1986; Gotoh & Tagashira, 1981), is 82 kcal/mol, which is in fairly good agreement with the experimental value. The thermodynamic parameters for transition 1, which reflects the process of triplex formation by an equimolar mixture of h and sl, have been obtained by best-fit analysis of the h : s l melting curves. The melting process for h : s l can be written as: triplex = hairpin + sl = sl + coil.
(1)
Within the approximation of a three-state model, the equilibrium constants for transitions 1 and 2 are given by: Kr=Co(1-a)fl/a=exp (-AHT/RT+ASr/R),
(2)
K s = (1 - ~ - f l ) / f l = e x p ( - - A H H / R T + A S H / R ) , (3) where ~ and fl are the fractions of triplex (T) and hairpin (H) at temperature t and Co the strand concentration of sl (equal to that of h). The thermodynamic parameters AH T, AST and AH H, ASs are the enthalpy and entropy changes for the triplex to ( h + s l ) and h to coil transitions, respectively. The absorbance for mixture h : s l is given by: Ar=%a+(ss+%)~+(%+%)(1--o~--fl),
(4)
where er, En and % are the extinction coefficients for the triplex (h:sl), hairpin (h) and single-stranded (sl) forms, respectively. The extinction coefficients of the three forms showed a slight dependence on temperature, therefore they were assumed to vary linearly according to the equations: ~T=mT(t--tl) + ~(tO, (SH+ ~S)= rnH(t-- t2) + ~(t2), (% + %) = rac(t-- t 3) + ~(t3 ),
(5)
where roT, ran, mc and 8(tl), e(t2) and e(t3) are the slopes and the experimental absorptivities at the temperatures tl, t 2 and t 3. Thus, the absorbance versus termperature curves obtained with h : sl were fitted to equations (2), (3), (4) and (5) with a nonlinear least-squares program (algorithm of Marquardt; Bevin~ton, 1969). The parameters to be adjusted were AH / , AS T and ms, and the remaining ones could be estimated separately. In fact AH" and AS H were calculated by melting h, and mT and m c were evaluated from the pre and post-transition regions of the melting curves. The theoretical curves described the experimental ones with very good accuracy, demonstrating the validity of the threestate model for describing the whole melting process
of h : s l . In general, the differences between computed and experimental AT values were always within 0-4~o, and the standard errors on the adjusted parameters computed by the Marquardt algorithm (Bevington, 1969) were of the order of 1 to 3~/o. Representative best-fits of melting profiles for h : sl are shown in Figure 6. The best-fit analyses of independent melting curves gave the values reported in Table 1. An average enthalpy value of AHT= 84"3( ___5) kcal/mol of triplex was obtained. Thus, the mean enthalpy change value for each sl pyrimidine residue binding in the major groove of h is 7"5(__+0"5)kcal/mol. The observation that a near tenfold increase in h : s l concentration shifts the semi-transition temperature of the duplex-triplex transition from tl/2=48°C to tl/2=55°C is consistent, according to the general bimolecular transitions theory (Cantor & Schimmel, 1980), with a large enthalpy change for this transition of about 80 to 90 kcal/mol. The comparison of AH r with AH N indicates that, for this system, the formation of a triple helix from a duplex and a single strand is enthatpically very similar to the formation of a duplex from two single strands. Since the formation of the triple h : sl requires the cytosine residues of sl to be fully protonated, the estimated AH T value could comprise the contributions from cytosine protonation. Two experiments resolved this problem. First, the protonation of the cytosine residues of sl as a function of pH has been followed by u.v. spectroscopy near the mid-temperature of the triplex h : s l to h + s l transition (45°C). At this temperature sl is essentially in the single-stranded conformation, since its pH-induced secondary structures observed at lower temperatures are nearly disrupted at 45°C, as revealed by u.v.-melting and eleetrophoretie experiments. Figure 7 (insert) shows the u.v.-spectral variations of sl (concn 0.11 mM/base) with pH. Starting from an acetic acid solution of sl at pH 3'03, successive portions of NaOH were added to increase the pH to 7. By plotting the u.v. signal at 290 nm versus pH, a semiprotonation point at pH 4.7 was observed. It seems that, at pH 5, about 30 to 35~/o of the cytidine residues appear to be protonated, which corresponds to about two protonated cytidine residues per sl strand. This finding suggests that the formation of h : sl triplex is accompanied by the protonation of the remaining four cytidine residues of sl. In order to evaluate the enthalpic contribution of cytidine protonation to the overall AH r of triplex formation, we measured by u.v. spectroscopy the protonation of eytidine as a function of pH, at three different temperatures in the range 22°C to 41°C, with a procedure similar to that adopted for sl (except that the initial p H values at about pH 2 were obtained by adding a small portion of HC1 to the acetic acid solution of cytidine). From the absorbances at 270 nm, the protonation degree of cytidine has been obtained as a function of pH at each temperature (Fig. 8). Since at the semi-protonation point pH = p K a, the dependence of pKa on tempera-
Triple Helix Formed by Oligodeoxynueleotides
0-4 -
841
~D
\
0.3 0-6 0"2 3
I
4
I
5
I
I
6
7
i
I
pH
0-5
I
220
I
I
260
I
I
500 Wavelength (nm)
I.
I
540
Figure 7. u.v. spectra of sl as a function of pH. A solution of sl (0"11 mMfoase) was made in dilute acetic acid (pH 3-04) and the pH was increased by adding discrete amounts of NaOH up to pit 7. The spectra, curves a, b, e, d, e and f have been obtained at pH values of 7, 6'07, 4"92, 4"50, 3"94 and 3"03, respectively. In the Figure we omitted the spectrum at pH 3"48 for design clarity. The insert shows the variation of absorbance at 290 nm as a function of pH. The titration was performed at 45°C with a thermostatically controlled pH meter system.
ture (insert in Fig. 8) allowed us to estimate an enthalpy of protonation of --2"8(___0"3)kcal/mol of cytidine. Assuming that this value can be applied to the sl cytosine residues, and considering that in this strand at p H 5 about two out of six cytosine residues are already protonated, the triple helix formation should require, about four protonations, equivalent to a enthalpic change of - 11-2 kcal/mol of sl. This value accounts for only 1 3 o of the total enthalpy change of transition 1, pointing to the major relevance of the stacking interactions in the energy of triple helix formation. A possible contribution due to a B to A conformational change of h
in order to host sl appears to be unlikely on the basis of the e.d. spectra: Figure 5 shows that the differential spectrum d is hardly compatible with such a conformational change. In conclusion, we found that for the h : s l triplex the mean enthalpy change (less the protonation contribution) for each pyrimidine residue binding in the major groove of h is evaluated to be 6"6(_0"4) kcal/mol. Krakauer & Sturtevant (1968) and Neumann & Ackermann (1969) reported a lower enthalpic value (--4keal/mol) for the addition of poly(rU) to poly(rA) : poly(rU), forming RNA triple helixes based exclusively on the U - A - U triad.
842
G. M a n z i n i et al.
4-4
.i/
cL 4-5 4.2 ~
+
._g
....... ~ " ....... ~ ......~..~t....
S 3"2
"~:~:"~.~..
.J
J
I/r
!
5"3
5.4
(K-' x IOs)
h_
% -o'°.
•
2
I
5
I ~':~'~'~
I
4
5
.... "~
6
~-
pH Figure 8. Fraction of protonated cytidine as a function of pH at 3 different temperatures. A solution of cytidine (0"l mM) in dilute acetic acid was titrated with NaOH until the pH reached neutrality. From the spectral variation at 270 nm the fraction of protonated cytidine versus pH was obtained. The semi-transition point (pKa) of 4"33, 4"26 and 4"20 are obtained at 295,304 and 314 K. The insert shows the plot of pK a versus l/T, which allows the estimation of AH for cytidine protonation to be --2"8 kcal/mol. Titrations were performed with a thermostatically controlled pH meter system.
W h e t h e r this difference has to be ascribed to the t y p e of sugar or to the base composition remains to be ascertained. This work was supported by the Italian Ministry of Education, the Italian National Research Council (CNR) and the Netherlands Organization for Pure and Advanced Research.
References
Arnott, S. & Selsing, E. (1979). J. Mol. Biol. 88, 509-521. Behe, M. J. (1986). Biopolymers, 25, 519-523. Behe, M. J. (1987). Biochemistry, 26, 7870-7875. Bevington, P. R. (1969). Data Reduction and Error Analysis for the Physical Sciences, McGraw-Hill, New York. Birnboim, H. C., Sederoff, R. R. & Paterson, M. C. (1979). Eur. J. Biochem. 98, 301-307. Breslauer, K. J., Frank, R., Blocker, H. & Marky, L. A. (1986). Proc. Nat. Acad. Sci., U.S.A. 83, 3746-3750. Cantor, C. R. & Schimmel, P. R. (1980). Biophysical Chemistry, part III, W. H. Freeman and Co., San Francisco. Christophe, D., Cabrer, B., Bacolla, A., Targovnik, H., Pohl, V. & Vassart, G. (1985). Nucl. Acids Res. 13, 5127-5144. De los Santos, C., Rosen, M. & Patel, D. (1989). Biochemistry, 28, 7282-7289. DePrisco Albergo, D., Marky, L. A., Breslauer, K. J. & Turner, D. H. (1981). Biochemistry, 20, 1409-1413. Edwards, E. L., Ratliff, R. L. & Gray, D. M. (1988). Biochemistry, 27, 5166-5174.
Evans, T., Schon, E., Gora-Maslak, G., Patterson, J. ¢¢ Efstratiadis, A. (1984). Nucl. Acids Re,v. 12, 8043-8058. Francois, J.-C., Saison-Behmaaras, T. & Helene, ('. (1988). Nucl. Acids Res. 16, 11431-11440. Frank, R. & Koster, H. (1979). Nucl. Acid,v Re.v. 5, 2069-2087. Freier, S. M., DePrisco Albergo, D. & Turner. D. H. (1983). Biopolymers, 22, 1107-1131. Gotoh, O. & Tagashira, Y. (1981). Biopolymers, 20, 1033-1042. Haasnoot, C. A. G., Hilbers, C. W., van der Marel, G. A., van Boom, J. H., Singh, U. C., Pattabiraman, N. & Kollman, P. A. (1986). J. Biomol. Struct. Dynam. 3, 843-857. Hanvey, J. C., Shimizu, M. & Wells, R. D. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 6292-6296. Htun, H., Lund, E. & Dahlberg, J. E. (1984). Proc. Nat. Acad. Sci., U.S.A. 81, 7288-7292. Ikuta, S., Chattopadhyahya, R., Ito, H., Dickerson, R. E. & Kearns, D. R. (1986). Biochemistry, 25, 5341-5350. Krakauer, H. & Sturtevant, J. M. (1968). Biopolymers, 6, 491-512. Larsen, A. & Weintraub, H. (1982). Cell, 29, 609-622. Lee, J. S., Johnson, D. A. & Morgan, A. R. (1979). Nucl. Acids Res. 6, 3073-3091. Lee, J. S., Woodsworth, M. L., Latimer, L. P. & Morgan, A. R. (1984). Nucl. Acids Res. 12, 6603-6613. Lewis, D. G. & Curtis Johnson, W., Jr (1974). J. Mol. Biol. 86, 91-96. Lyamichev, V. I., Mirkin, S. M. & Frank-Kamenetskii, D. M. (1986). J. Biomol. Struct. Dynam. 3, 667-669. Mirkin, S., Lyamichev, V. I., Drushlyak, K. N., Dobrynin, V. N., Filippov, S. A. &
Triple Helix Formed by Oligodeoxynucleotides Frank-Kamenetskii, M. D. (1987). Nature (London), 330, 495-497. Morgan, A. R. & Wells, R. D. (1968). J. Mol. Biol. 37, 63-80. Moser, H. E. & Dervan, P. B. (1987). Science, 238, 645-650. Neumann, E. & Ackermann, Th. (1969). J. Phys. Chem. 73, 2170-2178. Pulleyblank, D. E., Haniford, D. B. & Morgan, A. R. (1985). Cell, 42, 271-280. Rajagopal, P. & Feigon, J. (1989). Nature (London), 339, 637-640. Riazanee, J. H., Baase, W. A., Curtis Johnson, W. Jr, Hall, K., Cruz, P. & Tinoeo, I., Jr (1985). Nucl. Acids Res. 13, 4983-4989. Riley, M., Maling, B. & Chamberling, M. J. (1966). J. Mol. Biol. 20, 359-389. Roadbard, D. & Chrambach, A. (1971). Anal. Biochem. 40, 94-134.
843
Sarma, M. H., Gupta, G. & Sarma, R. H. (1986). F E B S Letters, 205, 223-229. Senior, M. M., Jones, R. A. & Breslauer, K. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 6242-6246. Schon, E., Evans, T., Welsh, J. & Efstratiadis, A. (1983). Cell, 35, 838-848. Van der Marel, G. A., van Boeckel, C. A. A., Wille, G. & van Boom, J. H. {1981). Tetrahedron Letters, 22, 3887-3890. Wells, R. D., Collier, D. A., Hanvey, J. C., Shimizu, M. & Wohlrab, F. (1988). Fed.Proc. Fed. Amer. Soc. Expt. Biol. 2, 2939-2949. Xodo, L. E., Manzini, G., Quadrifoglio, F., van der Marel, G. A. & van Boom, J. H. (1986). Nucl. Acids Res. 14, 5389-5398. Xodo, L. E., Manzini, G., Quadrifoglio, F., van der Marel, G. A. & van Boom, J. H. (1988). Biochemistry, 27, 6321-6326.
Edited by P. yon Hippel
Triple Helix Formation by Oligopurine-oligopyrimidine D N A Fragments Electrophoretic and Thermodynamic Behavior G i o r g i o M a n z i n i , L u i g i E. X o d o , D a n i e l a G a s p a r o t t o
Department of Biochemistry, Biophysics and Macromolecular Chemistry University of Trieste, 1-34127, Trieste, Italy Franco Quadrifoglio
Institute of Biology, Faculty of Medicine University of Udine, 1-33100, Udine, Italy G i j s A . v a n der M a r e l a n d J a c q u e s H. v a n B o o m
Gorlaeus Laboratories, State University P.O. Box 9502, 2300 RA Leiden, The Netherlands (Received 10 August 1989; accepted 24 January 1990) The 26met oligodeoxynucleot_ide d(GAAGGAGGAGATTTTTCTCCTCCTTC) adopts in solution a unimolecular hairpin structure (h), with an oligopurine-oligopyrimidine (Pu-Py) stem. When h is mixed with d(CTTCCTCCTCT) (sl) the two strands co-migrate in polyacrylamide gel electrophoresis at pH 5. If sl is substituted with d(TCTCCTCCTTC) (s2), such behavior is not observed and the two strands migrate separately. This supports the suggestion of the formation of a triple-stranded structure by h and sl (h :sl) but not by h and s2, and confirms the strand polarity requirement of the third pyrimidine strand, which is necessary for this type of structure. The formation of a triple helix by h : sl is supported by electrophoretic mobility data (Ferguson plot) and by enzymatic assay with DNase I. Circular dichroism measurements show that, upon triple helix formation, there are two negative ellipticities: a weaker one (As----80M-lcm -1) at 242nm and a stronger one ( A s = 2 1 0 M - l c m -1) at 212nm. The latter has been observed also in triple-stranded polynucleotides, and can be considered as the trademark for a P y : P u : P y DNA triplex. Comparison of ultraviolet absorption at 270 nm and temperature measurements shows that the triple-stranded structure melts with a biphasic profile. The lower temperature transition is bimolecular and is attributable to the breakdown of the triplex to give h and sl, while the higher temperature transition is monomolecular and is due to the transition of hairpin to coil structure. The duplex-to-triplex transition is co-operative, fully reversible and with a hyperchromism of about 10~/o. The analysis of the melting curves, with a three-state model, allows estimation of the thermodynamic parameters of triple helix formation. We found that the duplex-to-triplex transition of h: sl is accompanied by an average change in enthalpy (less the protonation contribution) of -73(_+5)kcal/mol of triplex, which corresponds to -6-6(+_0-4)kcal/mol of binding pyrimidine, attributable to stacking and hydrogen bonding interactions.
in Escherichia coli and some bacteriophages they are not. In nature, Pu-Py sequences are often found
1. Introduction Oligopurine-oligopyrimidine (Pu-Pyt) sequences are present in prokaryotes and eukaryotes. While in eukaryotic genomes the Pu-Py sequences are overrepresented (Behe, 1987; Birnboim et al., 1979),
0022-2836/90/120833-11$03.00/0
tAbbreviations used: Pu, purine; Py, pyrimidine; W-C, Watson-Crick; u.v., ultraviolet; c.d., circular dichroism; PAGE, polyacrylamide gel electrophoresis. 833
© 1990AcademicPressLimited
834
G. M a n z i n i et al.
within the regulatory regions of DNA, such as upstream from the genes (Larsen & Weintraub, 1982) and in recombination hot-spots (Htun et al., 1984). For instance, Pu-Py sequences have been mapped in the chicken fl-globin gene (Schon et al., 1983; Evans et al., 1984), the rat preproinsulin II gene (Evans et al., 1984) and the human thyroglobulin gene (Christophe et al., 1985). A peculiar property of natural Pu-Py sequences is that they are hyperreactive to $1 nuclease. This implies that they may retain a non-canonical Watson-Crick (W-C) double-helical structure. Two major unusual structures have been proposed for Pu-Py sequences. The first is based on a double-stranded helix and explains the $I hypersensitivity of (AG)n sites to nuclease $1 as resulting from a structure in which W-C A" T alternates with Hoogsteen G" C+ basepairs (Pulleyblank et al., 1985); the second structure assumes the formation of triple helices in correspondence with P u - P y sequences (Lyamichev et al., 1986; Mirkin et al., 1987). The formation of DNA triple helices by polynucleotides of defined sequences was widely studied in the 1960s (Riley et al., 1966; Morgan & Wells, 1968). These structures consist, in general, of one oligopurine and two oligopyrimidine strands. X-ray fiber diffraction studies revealed that an oligopurine strand and an oligopyrimidine strand are W-C base-paired, forming a double-stranded helix whose major groove is occupied by a second oligopyrimidine strand, oriented in parallel to the oligopurine strand, and held together by Hoogsteen base-pairs (Arnott & Selsing, 1979). Both arrangements require the hemiprotonation of cytosine residues. There is a large body of evidence supporting the triple helix model: in plasmids the triple helix formation is favored by low pH, superhelical stress and mirror symmetry of the Pu-Py segments (Mirkin et al., 1987). Interruptions in the mirror repeat have been demonstrated to inhibit the triple helix formation substantially, probably because they prevent Hoogsteen base-pairing (Mirkin et al., 1987). Clear evidence of intramolecular DNA triplexes in Pu-Py sequences inserted into plasmids has been obtained by Hanvey et al. {1988), using the specific chemical probes osmium tetroxide and diethyl pyrocarbonate to investigate base-pairing. However, there is little information about triple helix formation by ol~godeoxynucleotides in the absence of superhelical stress. It has been shown that oligopyrimidines bind in the major groove of DNA at homopurine-homopyrimidine stretches by forming local triple helices (Moser & Dervan, 1987; Francois et al., 1988). In addition, two nuclear magnetic resonance analyses on triplexes formed by short oligodeoxynueleotides have been reported (Rajagopal & Feigon, 1989; De los Santos et al., 1989). As yet, the biological function of triple-stranded DNA is not clear; nevertheless, there is a growing conviction that this unusual structure may play an important role in gene expression. An attractive function for the Pu-Py sequences may be that these are specific sites for single-stranded binding proteins
(Lee et al., 1984) or conversely that, by triplex formation, they prevent the binding of ligands of specific Pu-Py DNA stretches. Hence, it is therefore necessary to investigate the physical properties of triple-stranded DNA. We have studied the energetics as well as the influence of base sequence on DNA triple helix formation. The results are presented in this paper.
2. Materials and Methods
(a) Materials The oligodeoxynucleotides investigated in this study are: h 5' d(GAAGGAGGAGATTTTTCTCCTCCTTC); 81 5' d(CTTCCTCCTCT); s2 5' d(TCTCCTCCTTC}. These DNA fragments, as well as others of length varying from 12 to 20 base-pairs, used for reference in electrophoresis experiments, were synthesized in solid phase, according to a modified phosphotriester method as described (Van der Marel et al., 1981). After base deprotection, purification was carried out with Sephadex G-50 resin, by eluting the column with 0'05M-tetra-ethylammonium bicarbonate. Purity was confirmed by both high-pressure liquid chromatography and denaturing electrophoresis (Frank & Koster, 1979). Electrophoresis materials, i.e. acrylamide, bisacrylamide, N,N,N',N'tetramethylethylenediamine and ammonium persulfate, were purchased from Serva Feinbioehemica, and Stains All dye from Sigma. Bovine pancreas deoxyribonuclease I (DNase I) was obtained from Boehringer. (b) Buffer Ultraviolet (u.v.) absorption and electrophoretic experiments were carried out in 0"1 M-sodium acetate (pH 5), 50 mM-NaC1, 10 mM-MgC12. Circular dichroism (c.d.) measurements were done in 1 mM-sodium acetate (pH 5), 100 mM-NaClO,. {c) Polyacrylamide gel electrophoresis PAGE was carried out on g e l s (20cm or 10 cm x 15 cm × 0"15 cm) obtained from buffer solutions containing 20% (10%, 15% or 250/0 (w/v)) acrylamide, 3-3% (w/v) bisacrylamide, 0-07% (w/v) ammonium persulfate. The stacking gel contained 5% acrylamide. In general, the electrophoresis was performed at a constant voltage of 10 V/cm with a current of 20 to 30 mA. The amount of samples loaded on the gel was about 2 to 4 #g. Bromphenol blue dye was used as a marker. The bands were stained with Stains All dye in water/formamide (1 : 1, v/v) and the gels were photographed after 1 h. (d) Circular dichroism c.d. spectra were obtained with a Jasco J-500A speetropolarimeter, equipped with a thermostatically controlled cuvette holder that allows measurements at controlled temperature. The diehrograph was connected to a Jaseo DP500 N data processor. Spectra are presented as AS----(sL--sR) in units of M-I cm-l. The spectra were obtained from sample solutions at a strand concentration of about I00 ~M. Absorbance of the solvent in the short
Triple Helix Formed by OHgodeoxynucleolide8 wavelength region of the spectrum (200 to 230 nm) was minimized by using a 1 mm cuvette and 1 raM-sodium acetate, 100 mM-NaCl04 (pH 5) as buffer.
(e) DenaturaHon experiments The denaturation experiments were performed on a Cary 219 (Varian) spectrophotometer. The temperature was increased at a rate of 0-5 deg.C/min with a Haake PG20 temperature programmer, connected to a Haake water-circulating bath equipped with a refrigerator. The sample temperature was detected by means of a Varian thermistor immersed in an oil-containing cuvette situated in the sample turret. The melting temperature was monitored at 270, 264 and 250 nm and the oligomer concentrations were determined by u.v. absorption at 260 nm at t--90°C, assuming an extinction coefficient of 7500, 8500, 15,000 and 12,500 M-1 cm -1 for C, T, A and G, respectively, in the denatured state.
5'
GAAGGA
5' AA
A
835
GA
/T4 %
T , ,T8 ,TN I N
TT
,
A A"'-~T4
.'CTTCCTCCTCT..
GAA A A T4 3,CT T ~ T ~ T ~ T , . . _ . ~
*.s!
5'CT TCCTCCTCT
st.
5'TCTCCTCCT TC
s2
/~
CH3 ~
CH3
H
H
(g) Ultraviolet absorption spectra versus pH The u.v. absorption spectra of sl at 45°C and of cytidine at fixed temperatures in the range 20 to 42°C were recorded at pH values from 2 to 7. The pH was varied by adding small portions of 5 M-NaOH to a 0-1 M-acetic acid solution, initially brought to pH 2 with a small amount of HCI. The pH was measured at the same temperature for each spectrum using a thermostatically controlled pH meter system (Radiometer, priM52), equipped with a glass electrode, which had been calibrated at each considered temperature with standard buffer (Radiometer, pH 4"01).
T
A
T
C
G
C
Figure 1. Secondary structures accessible to h and an equimolar mixture of h and 81 (h:sl). In the triplestranded structure, sl occupies the major groove of the double-helical domain of hairpin h, forming TAT and CGC+ base triads, sl and the oligopurine strand of h are oriented in parallel. Strand s2 should fully recognize the major groove of h when it is antiparallel to the oligopurine strand. This orientation does not allow a triple-stranded structure.
3. D e s i g n o f T r i p l e - s t r a n d e d D N A In order to design a simple triple-stranded DNA system we exploited the intrinsic propensity of p a r t l y self-complementary sequences to assume a unimolecular hairpin structure (Xodo el al., 1986, 1988; I k u t a et al., 1986; H a a s n o o t el al., 1986; Senior et al., 1988). The sequence of h is characterized b y a 2-fold s y m m e t r y axis at its center, with the 5' arm comprising only purine residues, while the 3' arm includes only pyrimidine residues. Thus, h can a d o p t either a linear dimeric form with a bulge of four thymidine residues at its center or a folded unimolecular hairpin form with a four-T loop and a P u - P y stem of 11 base-pairs. The quasi-palindromic sequences of the h t y p e have been shown to a d o p t mainly a hairpin structure, over a wide range of temperatures, nucleotide concentrations and ionic strengths (Xodo et al., 1988; I k u t a et al., 1986; Senior et al., 1988). The major groove of the double helix domain of hairpin h should be recognized by sl, whose sequence is complementary to the oligopurine stretch of h when both strands are oriented in parallel. According to previous results, we expected for an equimolar mixture o f h and sl (h:sl) a triplestranded structure through the formation of T- A- T and C-G" C + base triads,, as illustrated in Figure 1 (Lyamichev et al., 1986; Mirkin et al., 1987; Wells et al., 1988). B y contrast, when 81 is substituted with 82 the triple-stranded structure should not form, since s2 and the oligopurine stretch of h are complem e n t a r y when t h e y are antiparallel to each other.
In conclusion, of the two strands sl and s2, only the first should bind in the major groove of the doublestranded domain of hairpin h, forming Hoogsteen base-pairs with the oligopurine strand. Another i m p o r t a n t reason why we chose to build a triplestranded DNA using hairpin h is t h a t we expected, due to the high stability of hairpin structures, a two-step melting profile for the complex h : sl, with triplex disruption and hairpin denaturation well resolved.
4. Results and Discussion
(a) Electrophoresi8 The potential of h:81 to form a triple helix was assayed b y non-denaturing P A G E at p H 5. This slightly acidic condition was necessary to obtain the triple-stranded structure. In fact P A G E experiments at p H 7 did not show an appreciable a m o u n t of complex h : 81. Figure 2(a) shows a P A G E analysis, using 2 0 % (w/v) polyacrylamide, of h, sl, 82, h : s l and h+s2. Lane g shows t h a t the 26met h migrates as a hairpin, exhibiting a mobility slightly faster t h a n t h a t of a 14mer (lane 1) and a 12mer (lane h). When h was mixed with 81 in a 1 : 1 strand ratio, a new band of low mobility (triple helix structure), comparable to t h a t of a 20mer running in lane i, appeared in the gel (lane f). I f the m i x t u r e
G. Manzini et al.
836
st or m Self-complexesq
h'$1
|
s l or s 2
(a)
$ f Or $ 2
Self-complexes A:Sl
h s f or s 2
(b)
Figure 2. (a) Polyaerylamide gel electrophoresis of h, sl, s2, h :sl and h + s2 using 20 % polyacrylamide. In each lane approximately 3 pg of DNA was loaded. Lanes h, i and j contain 3 reference duplexes: a 12mer, a 20mer and a 14mer, respectively. Lane g contains h, while lanes e and f contain h : sl, with and without thermal pretreatment. Strands 81 and s2 were loaded in lanes c and d, respectively. Lanes a and b contain the mixture h+s2, with and without thermal pretreatment. The 20% PAGE was in 0.1 M-sodium acetate (pH 5), 50 mM-NaCl, 10 mM-MgCl2. (b) PAGE of sl, s2, h, h:sl and h+82 using 25% polyacrylamide. Strands sl and s2 have been loaded in lanes a and b after being heated at 60°C, and in lanes c and d without any thermal treatment. Lanes e and f contain h+s2 and h : sl, where s2 and sl are in excess with respect to h. Three reference duplexes are in lanes g, h and i: a 12mer, a 20met and a 14mer, respectively. Gel conditions are the same as those reported for the 20 % PAGE.
of h and sl (sl in slight excess) was heated and then cooled to room temperature just prior to loading, the h band disappeared in favor of the low mobility species (lane e). However, when h was mixed with 82 (lane b, in equimolar ratio; lane a, in excess of s2) no influence on the mobility of band h was observed. A close inspection of lanes a and b revealed t h a t s2 migrated separately from h, and, surprisingly, not as one high mobility species, b u t together with twin bands of lower mobility. Lanes c and d show the electrophoretic profiles of sl and s2 alone. These bands are not stained well because sl and s2 do not form W-C secondary structures. However, b y using larger quantities of both sl and
s2, we obtained a clear electrophoretic p a t t e r n of the molecular species formed by sl, s2, h : s l and h + s 2 (Fig. 2(b)). Now it appears t h a t both s l and s2 migrate with three bands (lanes c and d): one with high mobility, corresponding to single-stranded sl and s2, and two close bands of low mobility, which, we believe are due to self-complexes of both sl and s2, stabilized by C ' C + base-pairs as found b y Edwards et ed. (1988) and Sarma et al. (1986) for C,T strands a t low pH. Thus, electrophoresis shows t h a t s l and not s2 is able to recognize the major groove of h and form a triple-helical structure. A further eleetrophoretic characterization of the mixture h : sl was obtained by studying its mobility
Triple Helix Formed by Oligodeoxynucleotides
. . . .
w r - - A -
837
__
\
0
._1 0.1
i.
% x Q
%
|
I0
!
I
20
SO
Acrylarnide (%)
Figure 3. Ferguson plots of an equimolar mixture of h and sl (h:sl) compared with a 20mer and a 12mer DNA duplex. The logarithm of the relative mobility of (O) h:sl, (A) 20mer and {A) 12mer (with respect to that of hairpin h) is been plotted against the concentration of acrylamide in the gel. The slope of these linear plots allows estimation of the retardation coefficient of the 3 samples.
as a function of gel concentration. The results are shown in Figure 3 in the form of a Ferguson plot (Roadbard & Chrambach, 1971): log Rr = log R d 0 ) - KRC, where Rf is the apparent mobility of the charged molecular species in the gel of acrylamide composition C and Rf(0) is the free relative mobility of the charged molecular species for C-*0. The slope, Ks, of the plot provides an estimate of the retardation coefficient of the charged molecular species, a parameter related to the size and shape of the molecular species itself (Roadbard & Chrambach, 1971). F r o m this, it is evident t h a t the 11-triad triplex behaves in a manner similar to t h a t of a 20mer fragment, i.e. with a retardation coefficient higher than t h a t of a 12mer DNA duplex of approximately the same length of h. The plots seem to converge at the same y-intercept value, indicating t h a t the free mobilities of duplex and triplex structures are similar. (b) Effect of DNase I on h and h : sl The effect of DNase I on both hairpin h and complex h :sl was investigated by analyzing the products using denaturing gel electrophoresis. Figure 4 shows the results of PAGE (20% acrylamide) of h and h: sl after t h e y have been incubated with DNase I at 30°C for 1-5 hours. Lanes a and f contain untreated h and h:sl, whereas lanes b to e
Figure 4. Denaturing gel electrophoresis of h and h : 81 after 1"5 h of incubation at 30°C with increasing amounts of DNase I. The incubation buffer was 0"1 M-sodium acetate (pH 5), 50 mM-NaCl, 10 mM-MgCl~. Lanes a and f contain untreated h and h:sl, respectively. Lanes b to e and g to 1contain h and h : sl digested with DNase I at the following /~g DNA//~g DNase I ratios: 11, 5-5, 3-7, 2"7. PAGE conditions: 25% acrylamide, 6 M-urea, 50 mM-Tris • HC1 (pH 8"3), 1 mM-EDTA, t--30°C, 10 V/cm.
and g to 1 contain h and h:sl, respectively, after incubation with increasing amounts of DNase I. The electrophoretic pattern obtained clearly shows that, while h is degraded, h : sl is resistant to the enzyme. This suggests t h a t h :81 forms a triple-stranded structure; such a structure has been reported to be a poor substrate for DNase I (Lee et al., 1984). (c) Circular dichroism c.d. spectroscopy is very suitable for distinguishing between different DNA structures. The formation of triple-stranded DNA by h : s l was followed also by c.d. Figure 5 shows the c.d. spectra of h, sl and the mixture h:sl. The following observations m a y be made. (1) The c.d. spectrum of h is consistent with t h a t of A-DNA, showing a large ellipticity (Ae= 150 M-1 cm -1) at 279 nm and a smaller one (Ae=70 M-1 em -1) at 242 nm (Riazance et al., 1985). (2) The c.d. spectrum of 81, obtained aider thermal treatment, has a shape typical of singlestranded DNA (Lewis & Curtis Johnson, 1974). (3) The c.d. spectrum of h :sl exhibits a large ellipticity (Ae=220 M-1 cm -1) at 279 nm, reminiscent of A-DNA, a crossover at 262 nm and two negative bands at shorter wavelengths: a weak band at 242 nm (A~=80M -1 cm -1) and a strong one at 212 nm (As ----210 M-1 cm-1). Subtracting the c.d. spectra of h and sl from t h a t of h : 81, the ellipticities arising upon DNA triple-stranded formation were obtained. Note t h a t this spectral difference shows clearly t h a t a new DNA structure is formed
838
G. Manzini et al.
3°°I 200
IOC
.o
oe='=~ = "m-% C •o
T
d
T
- .....
I/,.,.,"
II
-
-20(
220
250
280
Wovelength (nm}
Figure 5. c.d. spectra for (curve a), h:sl, (curve b), h, (curve c), sl and (curve d), h : s l - ( h + s l ) . The spectra were recorded at 25°C in 1 mM-sodium acetate (pH 5), 100 mM-NaCI04. The eltipticity is given in units of M-1 em -1, where M is expressed in mol DNA strands/l. by mixing h and sl; this is particularly indicated by the presence of a strong negative band at 212 nm. It is noteworthy that the c.d. spectrum of h :sl is rather similar to those reported by Lee et al. (1979) for triple-stranded polynucleotides. This ellipticity is not observed in A-DNA, B-DNA or Z-DNA (Riazance et al., 1985; Behe, 1986). Hence we are tempted to consider this short wavelength c.d. band as a trademark for P y : P u : P y triple-stranded DNA. Interestingly, the c.d. spectrum of a l : l mixture of h and s2 turned out to be nearly the sum of individual h and s2 spectra, lacking, as expected, an increased negative band at 212 nm. This suggests that no triple helix is formed with this DNA system, in agreement with the electrophoretic results. (d) Ultraviolet spectroscopy and melting experiments u.v. absorption spectra of h, sl and the mixture h: sl, at 25°C and 90°C, have shown that the magni-
rude of the thermally induced hyperchromic effect for h is 16% and for sl 1%, while for h : s l it is 27%. Thus, it follows that the duplex-to-triplex process is characterized by a hypochromism of about 10%. Since this process is accompanied by a partial protonation of the sl cytosine residues (see below), which in turn is a hyperehromic process at 270 nm, the total duplex-to-triplex hypochromism can be evaluated to be around 12%. This suggests that the binding of the pyrimidine strand in the major groove of h, via Hoogsteen base-pairing, results in considerable stacking interactions. The duplex-triplex conformational transition of h : s l was carefully analyzed by thermal denaturation experiments at 270, 264 and 250 nm. Figure 6 shows representative melting curves for h : s l , followed at 264 nm, both in denaturation and renaturation. The shape of the melting profiles is clearly biphasic. We call the transformation occurring at lower and higher temperatures transition 1 and transition 2,
Triple Helix Formed by Oligodeoxynucleotides 0.85
839
b
0.7S B 0.75 0.71
0.67 I
30
I
I
40
I
I
50 60 Ternperoture (°C)
I
70
80
Figure 6. Absorbance at 264 nm versus temperature profiles for the equimolar mixture h :sl in 0"1 M-sodium acetate (pH5), 50 mm-NaCl, 10 m~l-MgC12, in denaturation and renaturation. The nucleotide concentration of h:sl was 0"175 mm/base. Some points (A) from the best-fits of the experimental curves to eqn (5) are reported. The adjusted parameters AH T and ASv with relative errors are reported in Table 1. In each best-fit: (computed (A)--experimental (A))/experimental (A) <0"40/0 at each temperature. respectively. An insight into the molecularity of these transitions was obtained b y v a r y i n g the nucleotide concentrations of h : sl. We observed t h a t the melting t e m p e r a t u r e of transition 2, tl/2 = 72°C, remained c o n s t a n t over an eightfold increase in the nucleotide concentration, suggesting a unimolecular reaction for this transition. On the contrary, the tl/2 of transition 1 was positively affected by increasing the nucleotide concentration. T a b l e 1 gives the tl/2 values a t each concentration considered. This behavior is consistent with a bimolecular process for transition l, which we attribute, on the basis of P A G E and c.d. results, to triple helix formation. This conclusion is confirmed unequivocally by the fact t h a t hairpin h alone melts with a profile corresponding e x a c t l y to transition 2. I t is n o t e w o r t h y t h a t transition 1 is rather co-operative (it spans over a range of a b o u t 15 deg.C) and is fully revers-
ible at a scan rate of 0"5 deg.C/min (Fig. 6). F u r t h e r more, the fact t h a t the duplex-to-triplex transition is quickly reversible suggests t h a t a r a t h e r low activation energy m a y be involved in the process. More compelling evidence t h a t transition 1 is due to the f o r m a t i o n of a triple-stranded structure between h and sl comes from the observation t h a t this transition is absent in the m i x t u r e of h and s2, although s l and s2 have the same central TCCTCCT-stretch. Either the shortness of the central stretch or the hindrance of the dangling ends could be responsible for such behavior.
(e) Thermodynamics of triple helix formation Considering t h a t the processes of triple helix
(h : sl) and hairpin (h) formation are fully reversible, and t h a t their constituent strands are r a t h e r short,
Table
1
Thermodynamic parameters for the triplex h : s l to h + s l transition in sodium acetate buffer Concentration
(mM/base) 0.052 0.086 0.127 0.169 0,175 0,175 0,412 Average AHT Standard deviation
ill2 (°C) 48-0 52,0 52-0 52,5 53,5 53,0 55,0
AHr (kcal/mol)
A8z (e.u.)
Curvet
85,2( + 1.4)~ 80,1 ( + 1.2) 82,2(+ 1 - 2 ) 86,6( + 1,0) 81,4(___ 1,1) 84,8( + 2,0). 90,3( + 1.0)
238.4( + 4-1)~ 220,6( + 3.8) 227,8(+3,7) 241,5( ___2,7) 224-7(__+4.0) 235~)( + 6.0) 249,9( + 3.2)
R D D D R D D
84"3 kcal/mol of h : 81
+5 keal/mol
Sodium acetate buffer is: 0"1 M-SOdiumacetate (pH 5), 10 mM-MgCI2,50 mM-NaCI.Q/2 is the melting temperature; Aftr is the enthalpy change and A8 T the entropy change for the transition.
t R, renaturation; D, denaturation; 1 cal ffi4-184 J; e.u., entropy units. SThe uncertainties of AH T (and A8 T) in parentheses are given by the Marquardt (Bevington, 1969)
algorithm and refer to a single best-fit curve. They are lower than the final standard deviation over the 7 best-fits,since they do not reflectthe high correlationexistingbetween A H r and A 8 T
840
G. Manzini et al.
we could describe transitions I and 2 with a threestate model. The thermodynamic parameters for transition 2, i.e. for the denaturation of hairpin h, were obtained by analyzing the melting curves of sample solutions containing only h. By fitting these curves with a two-state model (DePrisco Albergo et al., 1981), using a non-linear least-squares program (algorithm of Marquardt; Bevington, 1969) we obtained the following values: tl/2=72(+1)°C, AH H= 85( _+4) kcal/mol, AS s = 246( _-4-i-0) e.u. (1 cal =4-184 J). The calculated denaturation enthalpy value for h, based on nearest-neighbor interactions (Breslauer et al., 1986; Gotoh & Tagashira, 1981), is 82 kcal/mol, which is in fairly good agreement with the experimental value. The thermodynamic parameters for transition 1, which reflects the process of triplex formation by an equimolar mixture of h and sl, have been obtained by best-fit analysis of the h : s l melting curves. The melting process for h : s l can be written as: triplex = hairpin + sl = sl + coil.
(1)
Within the approximation of a three-state model, the equilibrium constants for transitions 1 and 2 are given by: Kr=Co(1-a)fl/a=exp (-AHT/RT+ASr/R),
(2)
K s = (1 - ~ - f l ) / f l = e x p ( - - A H H / R T + A S H / R ) , (3) where ~ and fl are the fractions of triplex (T) and hairpin (H) at temperature t and Co the strand concentration of sl (equal to that of h). The thermodynamic parameters AH T, AST and AH H, ASs are the enthalpy and entropy changes for the triplex to ( h + s l ) and h to coil transitions, respectively. The absorbance for mixture h : s l is given by: Ar=%a+(ss+%)~+(%+%)(1--o~--fl),
(4)
where er, En and % are the extinction coefficients for the triplex (h:sl), hairpin (h) and single-stranded (sl) forms, respectively. The extinction coefficients of the three forms showed a slight dependence on temperature, therefore they were assumed to vary linearly according to the equations: ~T=mT(t--tl) + ~(tO, (SH+ ~S)= rnH(t-- t2) + ~(t2), (% + %) = rac(t-- t 3) + ~(t3 ),
(5)
where roT, ran, mc and 8(tl), e(t2) and e(t3) are the slopes and the experimental absorptivities at the temperatures tl, t 2 and t 3. Thus, the absorbance versus termperature curves obtained with h : sl were fitted to equations (2), (3), (4) and (5) with a nonlinear least-squares program (algorithm of Marquardt; Bevin~ton, 1969). The parameters to be adjusted were AH / , AS T and ms, and the remaining ones could be estimated separately. In fact AH" and AS H were calculated by melting h, and mT and m c were evaluated from the pre and post-transition regions of the melting curves. The theoretical curves described the experimental ones with very good accuracy, demonstrating the validity of the threestate model for describing the whole melting process
of h : s l . In general, the differences between computed and experimental AT values were always within 0-4~o, and the standard errors on the adjusted parameters computed by the Marquardt algorithm (Bevington, 1969) were of the order of 1 to 3~/o. Representative best-fits of melting profiles for h : sl are shown in Figure 6. The best-fit analyses of independent melting curves gave the values reported in Table 1. An average enthalpy value of AHT= 84"3( ___5) kcal/mol of triplex was obtained. Thus, the mean enthalpy change value for each sl pyrimidine residue binding in the major groove of h is 7"5(__+0"5)kcal/mol. The observation that a near tenfold increase in h : s l concentration shifts the semi-transition temperature of the duplex-triplex transition from tl/2=48°C to tl/2=55°C is consistent, according to the general bimolecular transitions theory (Cantor & Schimmel, 1980), with a large enthalpy change for this transition of about 80 to 90 kcal/mol. The comparison of AH r with AH N indicates that, for this system, the formation of a triple helix from a duplex and a single strand is enthatpically very similar to the formation of a duplex from two single strands. Since the formation of the triple h : sl requires the cytosine residues of sl to be fully protonated, the estimated AH T value could comprise the contributions from cytosine protonation. Two experiments resolved this problem. First, the protonation of the cytosine residues of sl as a function of pH has been followed by u.v. spectroscopy near the mid-temperature of the triplex h : s l to h + s l transition (45°C). At this temperature sl is essentially in the single-stranded conformation, since its pH-induced secondary structures observed at lower temperatures are nearly disrupted at 45°C, as revealed by u.v.-melting and eleetrophoretie experiments. Figure 7 (insert) shows the u.v.-spectral variations of sl (concn 0.11 mM/base) with pH. Starting from an acetic acid solution of sl at pH 3'03, successive portions of NaOH were added to increase the pH to 7. By plotting the u.v. signal at 290 nm versus pH, a semiprotonation point at pH 4.7 was observed. It seems that, at pH 5, about 30 to 35~/o of the cytidine residues appear to be protonated, which corresponds to about two protonated cytidine residues per sl strand. This finding suggests that the formation of h : sl triplex is accompanied by the protonation of the remaining four cytidine residues of sl. In order to evaluate the enthalpic contribution of cytidine protonation to the overall AH r of triplex formation, we measured by u.v. spectroscopy the protonation of eytidine as a function of pH, at three different temperatures in the range 22°C to 41°C, with a procedure similar to that adopted for sl (except that the initial p H values at about pH 2 were obtained by adding a small portion of HC1 to the acetic acid solution of cytidine). From the absorbances at 270 nm, the protonation degree of cytidine has been obtained as a function of pH at each temperature (Fig. 8). Since at the semi-protonation point pH = p K a, the dependence of pKa on tempera-
Triple Helix Formed by Oligodeoxynueleotides
0-4 -
841
~D
\
0.3 0-6 0"2 3
I
4
I
5
I
I
6
7
i
I
pH
0-5
I
220
I
I
260
I
I
500 Wavelength (nm)
I.
I
540
Figure 7. u.v. spectra of sl as a function of pH. A solution of sl (0"11 mMfoase) was made in dilute acetic acid (pH 3-04) and the pH was increased by adding discrete amounts of NaOH up to pit 7. The spectra, curves a, b, e, d, e and f have been obtained at pH values of 7, 6'07, 4"92, 4"50, 3"94 and 3"03, respectively. In the Figure we omitted the spectrum at pH 3"48 for design clarity. The insert shows the variation of absorbance at 290 nm as a function of pH. The titration was performed at 45°C with a thermostatically controlled pH meter system.
ture (insert in Fig. 8) allowed us to estimate an enthalpy of protonation of --2"8(___0"3)kcal/mol of cytidine. Assuming that this value can be applied to the sl cytosine residues, and considering that in this strand at p H 5 about two out of six cytosine residues are already protonated, the triple helix formation should require, about four protonations, equivalent to a enthalpic change of - 11-2 kcal/mol of sl. This value accounts for only 1 3 o of the total enthalpy change of transition 1, pointing to the major relevance of the stacking interactions in the energy of triple helix formation. A possible contribution due to a B to A conformational change of h
in order to host sl appears to be unlikely on the basis of the e.d. spectra: Figure 5 shows that the differential spectrum d is hardly compatible with such a conformational change. In conclusion, we found that for the h : s l triplex the mean enthalpy change (less the protonation contribution) for each pyrimidine residue binding in the major groove of h is evaluated to be 6"6(_0"4) kcal/mol. Krakauer & Sturtevant (1968) and Neumann & Ackermann (1969) reported a lower enthalpic value (--4keal/mol) for the addition of poly(rU) to poly(rA) : poly(rU), forming RNA triple helixes based exclusively on the U - A - U triad.
842
G. M a n z i n i et al.
4-4
.i/
cL 4-5 4.2 ~
+
._g
....... ~ " ....... ~ ......~..~t....
S 3"2
"~:~:"~.~..
.J
J
I/r
!
5"3
5.4
(K-' x IOs)
h_
% -o'°.
•
2
I
5
I ~':~'~'~
I
4
5
.... "~
6
~-
pH Figure 8. Fraction of protonated cytidine as a function of pH at 3 different temperatures. A solution of cytidine (0"l mM) in dilute acetic acid was titrated with NaOH until the pH reached neutrality. From the spectral variation at 270 nm the fraction of protonated cytidine versus pH was obtained. The semi-transition point (pKa) of 4"33, 4"26 and 4"20 are obtained at 295,304 and 314 K. The insert shows the plot of pK a versus l/T, which allows the estimation of AH for cytidine protonation to be --2"8 kcal/mol. Titrations were performed with a thermostatically controlled pH meter system.
W h e t h e r this difference has to be ascribed to the t y p e of sugar or to the base composition remains to be ascertained. This work was supported by the Italian Ministry of Education, the Italian National Research Council (CNR) and the Netherlands Organization for Pure and Advanced Research.
References
Arnott, S. & Selsing, E. (1979). J. Mol. Biol. 88, 509-521. Behe, M. J. (1986). Biopolymers, 25, 519-523. Behe, M. J. (1987). Biochemistry, 26, 7870-7875. Bevington, P. R. (1969). Data Reduction and Error Analysis for the Physical Sciences, McGraw-Hill, New York. Birnboim, H. C., Sederoff, R. R. & Paterson, M. C. (1979). Eur. J. Biochem. 98, 301-307. Breslauer, K. J., Frank, R., Blocker, H. & Marky, L. A. (1986). Proc. Nat. Acad. Sci., U.S.A. 83, 3746-3750. Cantor, C. R. & Schimmel, P. R. (1980). Biophysical Chemistry, part III, W. H. Freeman and Co., San Francisco. Christophe, D., Cabrer, B., Bacolla, A., Targovnik, H., Pohl, V. & Vassart, G. (1985). Nucl. Acids Res. 13, 5127-5144. De los Santos, C., Rosen, M. & Patel, D. (1989). Biochemistry, 28, 7282-7289. DePrisco Albergo, D., Marky, L. A., Breslauer, K. J. & Turner, D. H. (1981). Biochemistry, 20, 1409-1413. Edwards, E. L., Ratliff, R. L. & Gray, D. M. (1988). Biochemistry, 27, 5166-5174.
Evans, T., Schon, E., Gora-Maslak, G., Patterson, J. ¢¢ Efstratiadis, A. (1984). Nucl. Acids Re,v. 12, 8043-8058. Francois, J.-C., Saison-Behmaaras, T. & Helene, ('. (1988). Nucl. Acids Res. 16, 11431-11440. Frank, R. & Koster, H. (1979). Nucl. Acid,v Re.v. 5, 2069-2087. Freier, S. M., DePrisco Albergo, D. & Turner. D. H. (1983). Biopolymers, 22, 1107-1131. Gotoh, O. & Tagashira, Y. (1981). Biopolymers, 20, 1033-1042. Haasnoot, C. A. G., Hilbers, C. W., van der Marel, G. A., van Boom, J. H., Singh, U. C., Pattabiraman, N. & Kollman, P. A. (1986). J. Biomol. Struct. Dynam. 3, 843-857. Hanvey, J. C., Shimizu, M. & Wells, R. D. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 6292-6296. Htun, H., Lund, E. & Dahlberg, J. E. (1984). Proc. Nat. Acad. Sci., U.S.A. 81, 7288-7292. Ikuta, S., Chattopadhyahya, R., Ito, H., Dickerson, R. E. & Kearns, D. R. (1986). Biochemistry, 25, 5341-5350. Krakauer, H. & Sturtevant, J. M. (1968). Biopolymers, 6, 491-512. Larsen, A. & Weintraub, H. (1982). Cell, 29, 609-622. Lee, J. S., Johnson, D. A. & Morgan, A. R. (1979). Nucl. Acids Res. 6, 3073-3091. Lee, J. S., Woodsworth, M. L., Latimer, L. P. & Morgan, A. R. (1984). Nucl. Acids Res. 12, 6603-6613. Lewis, D. G. & Curtis Johnson, W., Jr (1974). J. Mol. Biol. 86, 91-96. Lyamichev, V. I., Mirkin, S. M. & Frank-Kamenetskii, D. M. (1986). J. Biomol. Struct. Dynam. 3, 667-669. Mirkin, S., Lyamichev, V. I., Drushlyak, K. N., Dobrynin, V. N., Filippov, S. A. &
Triple Helix Formed by Oligodeoxynucleotides Frank-Kamenetskii, M. D. (1987). Nature (London), 330, 495-497. Morgan, A. R. & Wells, R. D. (1968). J. Mol. Biol. 37, 63-80. Moser, H. E. & Dervan, P. B. (1987). Science, 238, 645-650. Neumann, E. & Ackermann, Th. (1969). J. Phys. Chem. 73, 2170-2178. Pulleyblank, D. E., Haniford, D. B. & Morgan, A. R. (1985). Cell, 42, 271-280. Rajagopal, P. & Feigon, J. (1989). Nature (London), 339, 637-640. Riazanee, J. H., Baase, W. A., Curtis Johnson, W. Jr, Hall, K., Cruz, P. & Tinoeo, I., Jr (1985). Nucl. Acids Res. 13, 4983-4989. Riley, M., Maling, B. & Chamberling, M. J. (1966). J. Mol. Biol. 20, 359-389. Roadbard, D. & Chrambach, A. (1971). Anal. Biochem. 40, 94-134.
843
Sarma, M. H., Gupta, G. & Sarma, R. H. (1986). F E B S Letters, 205, 223-229. Senior, M. M., Jones, R. A. & Breslauer, K. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 6242-6246. Schon, E., Evans, T., Welsh, J. & Efstratiadis, A. (1983). Cell, 35, 838-848. Van der Marel, G. A., van Boeckel, C. A. A., Wille, G. & van Boom, J. H. {1981). Tetrahedron Letters, 22, 3887-3890. Wells, R. D., Collier, D. A., Hanvey, J. C., Shimizu, M. & Wohlrab, F. (1988). Fed.Proc. Fed. Amer. Soc. Expt. Biol. 2, 2939-2949. Xodo, L. E., Manzini, G., Quadrifoglio, F., van der Marel, G. A. & van Boom, J. H. (1986). Nucl. Acids Res. 14, 5389-5398. Xodo, L. E., Manzini, G., Quadrifoglio, F., van der Marel, G. A. & van Boom, J. H. (1988). Biochemistry, 27, 6321-6326.
Edited by P. yon Hippel