3274
J. Phys. Chem. 1984,88, 3274-3282
interaction strength must be derived from the first (or possibly first two) solvation steps for which steric factors are minimal. Finally, much of the data of Table I, as well as other matrix solvation results, are understandable from this composit view. In particular, the surprisingly small (1-2) step for phen can be interpreted as reflecting a stage for which the cation charge (and therefore (6Av3)bonding)has been strongly reduced by the first bidentate phen coordination, but for which steric effects are still quite small. The subsequent step (2-3) is large because, as anticipated, steric crowding is significant for phen3.LiN03. It is interesting that limiting coordination numbers are larger for the bidentate ligands (six for phen and bpy vs. five for py; six for glyme vs. four for THF) while the residual Au3 values are only slightly smaller. This is attributed to the smaller steric interactions per coordinate bond for the bidentate ligands. For example, there are no solvent-solvent steric forces in the phen.LiN03 solvate, a situation that also tends to minimize any solvent-anion repulsions for the n = 2 case. It is also interesting that the only conceivable solvate structure for phen3.LiN03 (and thus bpy3.LiN03 and glyme3.LiN03), based on space-filling models, has the appearance of a paddle wheel with “flat” solvent (anion) units, with the long molecular axes parallel, spaced at -90° intervals.22 Such an arrangement of ligands creates a cavity for the Li+ ion with an -2.5 A diameter, which suggests a Li+-N (or 0) bond of -2.2 A in length. This can be compared with cation-to-ligand bond lengths of 1.9 A estimated for the lower H 2 0 and NH3 solvates of Li+N03-.9 Several points may also be made regarding previous reports on solutions of lithium nitrate in py(1). On the basis of the positions and appearance of certain bands in the far-infrared spectra of lithium salts in py(l), Popov proposed that there is no direct contact between ions in dilute solutions of lithium nitrate in ~ y ( l ) .Other ~~
-
(22) The planar nitrate structure may constitute the basis for a different solvate structure compared to the bis-coordinated Phen solvate of lithium perchlorate (Schmidt, E. et al. Inorg. Chim. Acta 1981, 52, 91-5).
researchers, however, have reported the mid-infrared spectrum for solutions of lithium nitrate in py (at similar concentration^),^^ and comparison of the gross features in the nitrate v3 region of that spectrum with the spectra for codeposits of lithium nitrate and py implies the existence of contact ion pairs in the liquid solutions. The solution assignments for the u3a and vjb bands of the totally solvated Li+N0< species, given as 1330 and 1410 cm-’, compare favorably with the values reported here for those bands, Le., 1334 and 1411 cm-’? As the preparative method used in the present work excludes the presence of ion triplets, no direct comment may be made concerning the bands reported a t 1354 and 1440 cm-’, for the solution of lithium nitrate in py, which have been assigned to such triplets.24 Although we cannot refute these assignments, it should be noted that, unless deuterated py is used, the presence of py bands in this region makes band assignments difficult, particularly near the intense 1439-cm-’ band of py. Also, two difficulties in the reasoning which led to their assignments should be pointed out. First, they assume that the u3 modes for the two ion triplets, ( ~ y ) ~ L i ( N 0(triplet-1) ~)~and ( ~ y ) ~ L i ~ N (triplet 0 ~ + + l ) , are identical. This seems highly unlikely considering the documented sensitivity of these modes to the NO3- environment. Second, their assertion that inner-sphere solvent molecules have no effect on the v3 modes of the ion pair is in direct contradiction to the results of the present work and that of previous matrix isolation experiments.2-6
Acknowledgment. Support of this research by the National Science Foundation under grant CHE-8209702 is gratefully acknowledged. Registiy No. LiN03, 7790-69-4;py, 110-86-1;phen, 66-71-7. (23) (a) Handy, P. R.; Popov, A. I. Spectrochim. Acta, Part A 1972,28, 1545-53. (b) Popov, A. I. Pure Appl. Chem. 1975, 41, 275-89. (24) Perelygin, I. S.; Klimchuk, M. A.; Belabrodova, N. N. Russ. J . Phys. Chem. 1980, 54, 605-6. (25) We have checked the v3 band positions for a dilute solution of LiN03 in liquid py and find that qa= 1415 cm-l and v3b = 1329 cm-I.
The Effect of Barbiturates on the Hydrogen Bonds of Nucleotide Base Pairs R. Buchet and C. Sandorfy* Ddpartement de Chimie, Universitd de Montrdal, and National Foundation f o r Cancer Research at Universitd de Montrdal, Montrtal, Qudbec, Canada H3C 3Vl (Received: September 23, 1983)
Infrared, near-infrared, and proton NMR investigations show that the hydrogen-bond equilibrium in adenine/thymine type base pairs can be perturbed by barbiturates. The effect of the perturbation consists of an increased number of “free” species resulting from the dissociation of the hydrogen bonds in the base pair while the barbiturate replaces thymine. It is further suggested that adenine/barbiturate/thymine trimer formation occurs as a step in the mechanism of the perturbation and that this may lead to an increase in the proportion of Hoogsteen type base pairs. It is speculated that such an event may be damageable to DNA and, eventually, lead to cancer.
Introduction In a previous paper’ the way in which the hydrogen-bond equilibrium can be perturbed in the l-cyclohexyluracil/9-ethyladenine dimer was examined by infrared and proton N M R spectroscopic techniques. The perturbers used were halofluorocarbons and barbiturates. While the hydrogen-bond-“breaking” properties of such compounds were observed in the course of a study related to anesthesia, the present work is not related to it. The intensity changes of the free and associated infrared NH stretching bands and the N M R chemical shifts showed conclu(1) R. Buchet and C. Sandorfy, J . Phys. Chem., 87, 275 (1983).
sively that the H-bond equilibrium is tilted in favor of the free species. The effect of phenobarbital on the base pair was spectacular: the l-cyclohexyluracil/9-ethyladenine complex was destroyed in favor of the 9-ethyladenine/phenobarbitalcomplex. This, of course, implies the dissociation of the H bonds in the original pair and the formation of new ones. In order to investigate the way in which the H bonds in the adenine/thymine and guanine/cytosine pairs are affected by external agents, and to approach biological reality, we have extended this work to the 8-bromo-2’,3’,5’-tri-O-acetyladenosine/ 1-methylthymine pair (Figure 1). This is in a Watson-Crick type model, the presence of the bromine atom preventing the formation of Hoogsteen type H bonds. It has been established that 92% of
0022-3654/84/2088-3274$01.50/0 0 1984 American Chemical Society
The Journal of Physical Chemistry, Vol. 88, No. 15, 1984 3275
Effect of Barbiturates on Nucleotide Base Pairs y 2
&IC OAc
0
2',3',5'
H
TRI-0-ACETYL
GUANOSINE
H2N &
O
in the following way: A solution of the salts in 10% HCl was prepared and the acid form extracted by the addition of ether; it was then purified through repeated recrystallization in chloroform. The infrared spectra were recorded in a Perkin-Elmer Model 621 spectrophotometer using 1.148-mm NaCl cells in the 40002000-cm-' region of the spectrum. The spectral resolution was of the order of 2 cm-'. The concentrations of the nucleotide bases were about 20 mM. The N M R spectrometer was a Bruker WH-90 instrument operating at 90 MHz in the Fourier-transform mode. The chemical shifts are reported in 6 (ppm) with tetramethylsilane as a standard. The near-infrared spectra were recorded on a Cary-17 instrument using 5- and 10-cm cells and concentrations of 20-30 mM. The resolution was of the order of 2 cm-'. Calculation of Equilibrium Constants. Let us consider the following equilibria:
COACG OAC d
a: y 2
I
0
CH3
I - M E T Y L CYTOSINE
I- M E T Y L T H Y M I N E
R,
0
i
R p : CH3CH2-
i"'
R2
2
CH2= CKH2 :
CH3(CH,)2
-
Rl
CH3(CH2)2 FHCH
R2 i "R2 l E = CCH3CH26 H5-
3
Figure 1. Structures of compounds dealt with in this paper.
the complex formed is of the Watson-Crick The external agents used in the study presented here were all barbiturates. They were chosen for two reasons. First, barbiturates form selective H bonds with adenine derivatives4 which are stronger than the adenine-uracil or adeninethymine H bonds. Second, according to our previous observation phenobarbital is able to replace uracil in an adenine/uracil complex. The barbiturates used were phenobarbital, secobarbital, allobarbital, pentobarbital, and barbital (Figure 1). Infrared, near-infrared, and proton N M R spectra were used to monitor the changes that occured in the H bonds in the base pairs when they were put into contact with a barbiturate. Experimental Section
8-Bromo-2',3',5'-tri-O-acetyladenosine (A), 1-methylthymine (T), and 1-methylcytosine (Vega Biochemicals), 2',3',5'-tri-Oacetylguanosine (Sigma Chemical Co.), and the barbiturates (B) (May and Baker) were high-purity products and were used without further purification. Secobarbital and pentobarbital were purchased in salt form. They were transformed into the acid form (2) H. Iwahashi, H. Sugeta, and Y . Kyogoku, Biochemistry, 21, 631 (1982). ( 3 j M . Watanabc, H. Sugeta, H. Iwahashi, Y.Kyogoku, and M. Kainosho, Eur. J . Biochem., 117, 553 (1981). (4) Y.Kyogoku, R. C. Lord, and A. Rich, Nature (London), 218, 69 (1968).
.-
(3) [AB1
-
- 2[A21)([BT1 -
- 2[B21)
- (b2 - 4~)'/,)/2
[AB] = (-b
(5)
where
= -([AT] +
+
iBTI
/ K - 2[A21 - 2[B21)
and = ([ATIIBTI - 2[AT1[B21 - 2[A21[BT1 + 4[A21[B21)
PENTOBAR~TAL
PHENOBARBITAL
B+B+B2
(4) where [AT] and [BT] stand for the total concentrations of A and B while [AB], [A,], and [B2] represent the concentrations of the complex AB and of the self-associated H-bonded dimers of A and B. [A] and [B] are the concentrations of free A and B. The concentration of the complex AB is determined by
SECOBARBITAL
3
(2)
-
$H CH3
R I * CH3CH2R2
K = - - [AB1
ALLOBARBITAL
I
(1)
A+A+A2 (1) is governed by the equilibrium constant
BARBITAL
R, = R = CH2=CHCH2-
A+B+AB
The method of calculation has been described by Long and Drago,s who used UV absorbances, but naturally it can be based on IR absorbances as well. We adapted the method to the use of N M R data, replacing the absorbances by chemical shifts. Futhermore, we took account of the self-association of A and B. Both IR and N M R data were used in our calculations. The observed chemical shift, dobsd, is given by
where dA2and dA stand for the chemical shift of the dimer and monomer species whie dABrepresents the chemical shift of complex AB. By rearrangement we can obtain the equation (7) with
A6 =
(dobsd
=
(dA,
- dA) - dA)
= (dAB - dA) In order to obtain the equilibrium constant K , we use a nonlinear least-squares type method. The quantity A? has to be minimized in the following expression: = Ej(A6 - A6caicd)? (8) where ABcalcdcorrespond to A6 computed from eq 7 and A6 is ( 5 ) J. R. Long and R. S. Drago, J. Chem. Educ., 59, 1037 (1982).
3276 The Journal of Physical Chemistry, Vol. 88, No. 15, 1984
Buchet and Sandorfy
3400 3300 3200 3100
3500
km-’)
icm-‘1 Figure 2. (---) Part of the infrared spectrum of an equimolar (15 mM) solution of 8-bromo-2’,3’,5’-tri-U-acetyladenosine and 1-methylthymine (A/T) in CDC13. (-) The same, containing in addition 15 mM of
secobarbital.
Figure 3. (---) Part of the Infrared spectrum of an equimolar (15 mM) (A) and secobarbital solution of 8-bromo-2’,3’,5’-tri-U-acetyladenosine in CDCI,. (-) The same, containing in addition 15 mM of l-methyl-
thymine (T).
obtained experimentally. The summation is over all the measurements which were carried out. Minimization leads to
=
-
First, the self-association constant values of A, and Bz molecules are computed separately, and subsequently the values of A6, and d , are obtained (do is obtained by extrapolating to zero concentration). Then, a trial value of K is injected into expression 5 to compute [AB] by neglecting the values of [A,] and [B,]. This [AB] value is needed to evaluate the [A2] and [B2] values by means of the expression [A,] = ( b - (b2 - ~ ) ‘ / ’ ) / 2
(10)
with
b = [AT] - [AB] 4- 1/(4K,) and c = ([At1 where K , is the self-association constant of A. We can obtain equivalent expressions for computing [B2] (replace [AT] by [BT] and K , for Kb). A new value of [AB] is computed by means of eq 5 with the [A,] and [B2] values. This process is repeated until a good value of [AB] is obtained. (Generally 10 iterations were needed). A& is then computed from (9) with the values of [AB] and [A2]. Subsequently, using (7), we obtain and by using (8). The trial values of K are varied until a minimal value of $ is obtained.
Results Infrared Spectra. Figure 2 shows the IR spectrum of the 8-bromo-2’,3’,5’-tri-O-acetyladenosine/l-methylthymine pair (dashed lines) (henceforth abbreviated by A/T for the sake of simplicity). This spectrum has a great resemblance to the one of the 9-ethyladenine/ 1-cyclohexyluracil pair which was discussed in our previous paper.l The assignment of the bands is, of course, also the same. The free NH, stretching bands of A are at 3526 cm-’ (antisym) and 3413 cm-’ (sym), respectively, while the corresponding association bands are at 3489 and 3319 cm-I. The free N H band of T coincides with the 3413-cm-’ band and cannot be distinguished. Association bands are found at 3273 and 3210 cm-’. They may correspond to the A/T complex or to the T dimer. The effect of the barbiturate secobarbital on the spectrum of this base pair is seen on the solid curve in Figure 2 (15 mM for each component). First, we observe a decrease in the intensity of the free NH, bands of A at 3526 cm-’. The band at 3413 cm-’
is overlapped by the intense free NH band of secobarbital at 3389 cm-l. However, we observe that the association NH, band at 3489 cm-’ becomes more intense. This can only mean that secobarbital formed H bonds with the nucleotides. Since the 3526- and 3489-cm-’ bands are due to N H 2 vibrations of the adenine derivative and since they are affected by secobarbital, we may conclude that secobarbital forms H bonds with A. The symmetrical association band shifted from 3319 to 3325 cm-’. Absorption in the region of the association bands at 3325, 3273, and 3210 cm-’ is boosted by the absorption of the barbiturate, but it is likely to gain intensity from Afsecobarbital complex formation as well. In order to obtain more insight into these conditions, we have recorded the spectrum of a 1:1 mixture (15 mM of each component) of secobarbital and A (Figure 3, dashed curve). The solid curve is that of a mixture of A, T , and secobarbital. The two spectra in Figure 3 are similar, certainly more so than the two spectra shown in Figure 2 where the difference between the curves of A / T and A / T secobarbital is shown. This leads to the conclusion that a part of the H bonds between the adenine and the thymine derivatives has been dissociated in favor of a new association between the secobarbital and the adenine derivative (confirmed by the N M R results; see below). The same applies for all five barbiturates which were used in this study. In order to ascertain if thymine derivatives can associate with barbiturates, we recorded the spectrum of an equimolar solution of barbital and T and compared it with that of barbital alone. Only a slight decrease in the intensity of the free N H band at 3383 cm-’ was observed, showing that the association between T and barbital is weak. (See the next section). Proton NMR. H bonding involving purine and pyrimidine bases has been the subject of a number of studies.6-’0 In order to follow the perturbing effect of barbiturates on the H bonds in the A/T pair, the chemical shifts of some of the protons have been determined. The diagrams shown in Figure 4 represent the variation of these chemical shifts as a function of the concentration of a barbiturate (B), barbital. As is seen, the signals of H 2 and NH2 of the adenine part shift toward lower fields with increasing barbiturate concentration. This can only mean that A is engaging in H-bond formation with the barbiturate. This result is not surprising; it was previously demonstrated by K y ~ g o k u .The ~ H2(A) proton is also sensitive to this change in the H-bonding pattern but much less so. The chemical shifts of NH(T) and NH(B) (barbiturate) are of interest. That of NH(T) goes toward higher fields as the concentration of the barbiturate increases. It reaches a minimum (6) R. R. Shoup, H.T. Miles, and E. D. Becker, Biochem. Biophys. Res. Commun., 23, 194 (1966). (7) L. Katz and S. Penman, J . Mol. Biol., 15, 220 (1966). (8) L. Katz, J . Mol. Biol., 44, 279 (1969). (9) T. Morishima, T. Inubushi, T. Yonezawa, and Y. Kyogoku, J . A m . Chem. Soc., 99, 4299 (1977). (10) G. Govil and R. V. Hosur, “NMR 20. Conformation of Biological Molecules. New Results from NMR”; Springer-Verlag, West Berlin, 1982.
.
The Journal of Physical Chemistry, Vol. 88, No. 15, 1984 3217
Effect of Barbiturates on Nucleotide Base Pairs 301 K
301 K
2 0 mM AT 2 0 mM A T
NH (81 9 50
NH ( T I
H2 ( A )
H 2 (A)
8.25
8.25
IO
20
30
40
IO
mM
A L L OBA R BI TA L
ppml
30
20
40
mM
SE C 0BAR BI TAL
PENTOBARBITAL
301 K
301 K 10.751
.:1""OL _2 10.2 5
10501 Y
2 7 m M AT
1O.OOf
I O 00
NH
9 75
(El
NH (TI
H 2 (AI
7 2 L NH2 (A)
6 25
6 25 I
20
40
60
BAR 81TAL
BO
mM
IO
20
30
* mM
PHENOBARBITAL
Figure 4. Variation, at 301 K, of proton chemical shifts with the concentration of a barbiturate: A, 8-bromo-2',3',5'-tri-O-acetyladenosine; T, 1-methylthymine; B, barbiturate, as indicated. The A/T concentraiton was kept constant.
at about 30 mM of barbital (Figure 4), and then it turns slightly in the opposite direction. This, we believe, is an important observation and can be interpreted in the following way. Below 30 m M the main event is the dissociation of the A / T pair and the association of the adenine derivative with the barbiturate whereby the thymine derivative is set free. Subsequently, as the concentration of the barbiturate increases, other associations became increasingly important. These can be the self-association of thymine, the association between thymine and barbital, and the formation of trimers of type A/B/T. All these can compete with the liberation of T; this can make it understandable that as the concentration of the barbiturate reaches a certain level, the trend in the chemical shift of NH(T) is reversed. The trends observed in the chemical shift of the NH(B) proton (Figure 4) are also significant. With increasing concentration of the barbiturate (and at constant concentration of A/T), the chemical shift is toward higher fields. This simply means that a t low B concentrations the quasi-totality of B molecules is H bonded to A, but when the concentration of B exceeds that of A, an increasing number of B molecules remain "free". It also shows that H bonding between the barbiturate and the adenine derivative dominates over all other H-bond formation that might be involved, in particular the self-association of the barbiturate. Increasing self-association of the latter would, naturally, cause a shift to lower fields in NH(B). Similar results were obtained with phenobarbital,
pentobarbital, secobarbital, and allobarbital (Figure 4). The substituents at C5in the barbiturate have no significant influence on the extent in which they perturb the H bonds in the nucleotide base pair, but phenobarbital seems to be slightly more efficient than the others. Low-temperature N M R spectra also furnish some interesting observations. As is known,, at room temperature the two NH, protons give only one peak in the proton N M R spectrum; at low temperature, however, it is possible to resolve two peaks. An example is given in Figure 5 ; it is the spectrum of a solution containing 20 m M of the A / T pair and 10 mM of pentobarbital. The signals from the H-bonded protons shift toward lower fields as temperature decreases since associations are then more and more favored. The rate of shifting with temperature differs from molecule to molecule. Thus, the signal of the N H protons of barbiturates (which coincides at room temperature) shifts much faster than that of thymine. Therefore, the NH(B) and NH(T) signals get increasingly better separated as temperature decreases. This is likely to be due to the dissociation of the A / T and the formation of A/B complexes in which the H bonds are stronger. These low-temperature N M R spectra help in understanding the mechanism of dissociation of the A / T pair and give evidence for trimer formation. Figure 6 represents the spectrum of a 20 mM solution of the A / T pair at 213 f 1 K and the variations of the chemical shift
3278 The Journal of Physical Chemistry, Vol. 88, No. 15, 1984
Buchet and Sandorfy
TABLE I: Eguilibrium Constants Obtained from Proton NMR Datao
system allobarbital, self-associated methylthymine, self-associated methylthymine (4 L mol-’) + allobarbital (4 L mol-’) allobarbital (4 L mol-I) + 8-bromo-2’,3’,5’tri-0-acetyladenosine(2 L mol-I) methylthymine (4 L mol-I) + 8-bromo-2’,3’,5’tri-0-acetyladenosine(2 L mol-’)
4,ppm 3.12 f 0.08 3.20 f 0.13 3.12 f 0.08 3.12 f 0.08
4.50
44 f 4
d,, ppm 7.86 f 0.02 8.05 f 0.04 7.86 f 0.02 7.86 f 0.02
4.73 f 0.15
proton NH, allobarbital NH, methylthymine NH, allobarbital NH, allobarbital
29 f 4
8.05 f 0.04
3.20 f 0.13
4.83 f 0.09
NH, methylthymine
K,L mol-’ 4 f 0.5 4 f 0.5 7 f 2.8
A&, ppm
* 0.11
“Numbers in parentheses indicate self-association constants used for estimation of equilibrium constants. CHC13
H2 (AI
213 K
While this is, of course, somewhat speculative, is appears that trimer formation may lead to a change in the relative proportions of the Hoogsteen and Watson-Crick type pairs. Higher trimer concentration would increase the probability of formation of Hoogsteen pairs. This is followed by a shift of the NH2 syn signal toward higher fields, indicating a decreased proportion of Watson-Crick pairs, and a shift of the N H 2 anti signal toward lower fields, indicating formation of Hoogsteen type H bonds. This is actually what is observed (Figure 6) at high barbiturate concentrations. The other barbiturates studied gave similar results. In order to substantiate the above hypothesis, we carried out similar measurements on A barbiturate mixtures, in the absence of T. The results are shown in Figure 8. We observe a decrease in the variation of the chemical shift of the syn (A) proton similar to that of A/T, although it is somewhat less pronounced. There is also a slight decrease in dissociation of the H bonds at the syn “,(A) side when the concentration of the trimer becomes relatively high. At even higher barbiturate concentration this trend disappears and the variation of the chemical shift of the syn “,(A) proton comes to a standstill. Then the site syn(A) is completely saturated (Figure 8). While this reasoning is based on slight changes in the chemical shifts, these shifts are entirely reproducible. Equilibrium Constants. Approximate values of the equilibrium constants were determined by using the methods outlined above. The entries in Table I were obtained at room temperature (21 f 1 “C) when determined from IR data and at 28 f when from NMR, from the average of seven or eight measurements. They give an approximate idea of the relative strengths of the associations dealt with in this paper. Equilibrium constants for similar systems found in the literature3 are close to ours. For systems containing adenine derivatives instead of the 8-bromo d e r i ~ a t i v e the , ~ equilibrium constant is larger by a factor of about 5-10. The effect of this on the AGO values is moderate, however, and does not affect our conclusions. The Guanosine/Cytosine Pair. After having observed sizeable effects on our model of the adenosine/thymine pair, it was logical IQ inquire about the possible effect of barbiturates on the guandne/cytosine pair. As a model, we chose the 2’,3f,5f-tri-0acetylguanosine/ 1-methylcytosine pair (Figure 1) which was previously used by Pitha, Jones, and Pithova.” First, we used MezSO for solvent but found that the barbiturates were completely associated to Me,SO. N o variation of the chemical shift with concentration was observed. Subsequently, we recorded the IR spectra in CDC13solutions at concentrations of about 1 mM. The bands due to the H-bonded base pair could be clearly distinguished at this concentration. None of the five barbiturates we used affected the free/association ratio, however. So, the result is simply that neither guanosine nor cytosine formed complexes with the barbiturates under the given conditions; that is, the H bonds in the G/C pair were not affected. This is in conformity with previous results by Kyogoku et aL4 Near-Infrared Spectra. The near-infrared region is the part of the electromagnetic spectrum comprised between the 14 000 and 4000 cm-’. From the point of view of the present work the 7000-4000-~m-~ region is important. Between about 7000 and 6500 cm-’ we find the first overtones of the NH, and N H stretching vibrations, and between 6000 and 5600 cm-’ we find those of the CH3 and CH, stretching vibrations; so, we shall call
+
223 K
228 K
233 K
243 K
12
11
1 0 9
8
7
6
5
PPm
Figure 5. Temperature variation of a part of the proton NMR spectrum of a solution in CDCI3of 8-bromo-2’,3’,5’-tri-O-acetyladenosine(A, 20 mM), I-methylthymine (T, 20 mM), and pentobarbital (B, 10 mM). as a function of the concentration of allobarbital. The interpretation of those related to protons NH(B), NH(T), and H2(A) is essentially the same as before. The ”,(A) protons give some new insight, however. The two protons can be designated syn and anti (Figure 7). Proton syn is the more strongly associated one; it is H bonded to T, forming a Watson-Crick type pair. (For the nomenclature and assignment of bands syn and anti, see ref 2.) Between 0 and 20 mM of allobarbital the variation of the chemical shift of proton syn is greater than that of proton anti. This means that the barbiturate binds preferentially to A. Subsequently, when most of the syn protons became engaged in H bonds, the anti proton starts forming its H bonds. This is accompanied by the shift of the anti signal toward lower fields when the concentration of allobarbital is increased, implying trimer formation (Figure 7). (Two different trimers can be thought of: B-A-T or B-A-B). It is interesting to observe (at 20 m M of allobarbital) that while trimers are already being formed, the NH(T) signal is still shifting toward higher fields, showing that T is still being liberated. At high barbiturate concentrations a slight decrease in the rate of variation of the syn ”,(A) signal is observed (Figure 6). This could be due to a slight degree of dissociation at the syn side according to one of the following mechanisms: B A-T(syn) * B-A-T * B-A(anti) + T Watson-Crick pair Hoogsteen pair
+
or B
+ A-B(syn) * B-A-B Watson-Crick pair
* B-A(anti) + B Hoogsteen pair
( 1 1 ) J. Pitha, R. N. Jones,and P. Pithova, Can.J. Chem., 44, 1045 (1966).
The Journal of Physical Chemistry, Vol. 88, No. 15, 1984 3279
Effect of Barbiturates on Nucleotide Base Pairs
1
ppm
213 K
213
K
228 K 15 m M AT
12 0
2 0 mM A T
130
H 2 (A)
S Y N NH2 ( A )
75 70
A N I NH2 (AI 60
U -7-
40
20
60
20
mM
4
20
60 mM
.
40
. 60
1
80
100
mM
SECOBARBITAL
PENTOBARBITAL
ALLOBARBITAL
ppm
40
213 K
NH ( T )
H 2 (AI
-
S Y N NH2 LA)
-
S Y N NH2
20
40
60
mM
20
BARB1 T A L
40
60
mM
PHENOBARBITAL
Figure 6. Variation, at 21 3 K, of proton chemical shifts with the concentration of a barbiturate: A, 8-bromo-2’,3’,5’-tri-O-acetyladenosine; T, 1-methylthymine;B, barbiturate, as indicated. The A/T concentration was kept constant. ICH3
I R 3
Figure 7. syn and anti hydrogen bonds on a trimer formed by a thymine derivative, an adenine derivative, and a barbiturate.
this the “overtone region”. At lower frequencies, 5000-4000 cm-’, the most characteristic bands are due to combination bands involving one quantum of an NH stretching and one quantum of
an N H in-plane bending vibration, and at somewhat lower frequency, similar combinations of CH vibrations. The appearance of the near-infrared spectra for H-bonded systems is governed by the fact the free/association intensity ratio, which in the region of the fundamentals is highly in favor of the association bands, is highly in favor of the free bands for the ~ v e r t o n e s ’ ~and, J ~ to a lesser extent, for the combination t o n e ~ . ’ ~ - l ~ The pertaining observations and the reasons for these facts were presented several years ago. Basically, they are connected with the signs and values of the mechanical and electrical anharmonicities of the vibrations and the high polarity of the H (12) (13) (1968). (14) (15)
W. A. P. Luck and W. Ditter, J . Mol. Struct., 1, 261 (1967-1968). W. A. P. Luckand W. Ditter, Ber. Bunsenges. Phys. Chem., 72, 365
0. D. Bonner and Y .S. Choi, J . Phys. Chem., 78, 1723 (1974). G. Trudeau, K. C. Cole, R. Massuda, and C. Sandorfy, Can. J . Chem,, 56, 1681 (1978), (16) T. Di Paolo, C. BourdBron, and C. Sandorfy, Can. J . Chem., 50,3161 (1972).
3280 The Journal of Physical Chemistry, Vol. 88, No. 15, 1984
pvm
f
Buchet and Sandorfy TABLE II: Band Frequencies in the Near-InfraredSpectra of 8-Bromo-2’,3’,5’-tri-O-acetyladenosine(A), 1-Methylthymine (T), and Phenobarbital in CHCL or CDCl,
213 K
freq, cm-’ H2
8ol 7
7 5
’IN
IAI
6662
2X
assignment 1-Methylthymine u(NH) free
“2
combinations involving u(NH) and 6(NH) 20
40
60
ALLOBARBITAL
Phenobarbital 6649
213 K
N H IBI
SYN NH2(AI
NH
I1 0
SIN
75
(81 LA1
40
60
EO
mY
70
100
disappear if D 2 0 is added
6889 6812 6743 6609 6557
””}
2 X u(CH)
2 X u(NHz) (a) free
A N T I NHZ (A1
50
combinations involving 4”) or W H )
8-Bromo-2’,3’,5’-tri-O-acetyladenosine disappear if DzO is added to the solution 2 X u(NHz) (a) associated u(NHZ) (a) + u(NH2) (s) free 2 X u(NH,) (s) free 2 X u(NH2) associated 2 X u(NH2) associated
NH2LAl
65
PH ENOEPRBITAL
disappears if D20is added
4679 4624 4580
6997 ANTI NH2 LA1
2 X u(NH) free
5784 5685
H2
20
disappear if D 2 0 is added
4716
M-
80 m M
disappears if DzO is added
1%
200 mM
SECOEARBITAL
Figure 8. Variation, at 213 K, of proton chemical shifts with the concentration of a barbiturate: A, 8-bromo-2’,3’,5’-tri-O-acetyladenosine; B, barbiturate, as indicated. The A concentration was kept constant.
Table I1 lists the frequencies and their assignments for A, T, and phenobarbital, respectively. For the adenine derivative three N H 2 bands are found in the overtone region, at 6997, 6812, and 6743 cm-’ for a 30 mM solution in chloroform (Figure 9). The first one and the last one are readily assigned the first overtone of the free antisymmetrical (v3) and symmetrical (vl) NH2stretching vibrations, respectively. The fundamentals are at 3526 and 3413 crn-’, yielding -28 and -42 cm-l for the respective anharmonicity constants. The band a t 6812 cm-’ is almost certainly the (vl + vg) combination tone, corresponding to a coupling constant equal to -1 27 cm-’ (68 12 - 3526 - 3413 cm-I). All these bands are allowed under the C,, total symmetry of the N H 2 group and even more so in A which has no symmetry a t all. All these anharmonicity constants fit in well with previous observations on N H 2 and ”.I8 The anharmonicity constants X = vol - vo2/2 (where vol and vo2 are the observed frequencies of the fundamental and the first overtone, respectively) for the free N H stretching vibration are of the order of -50-70 cm-’. When there are two hydrogens attached to the central atom, the constant is generally divided about half and half between the antisymmetrical and symmetrical modes as is known, for example, fot the similar case of O H vs. H20.19 Also, the anharmonic (17) C Sandorfy, “The Hydrogen Bond”, Vol. 2,P. Schuster, G.Zundel, and C. Sandorfy, Eds., North-Holland Publishing Co., Amsterdam, 1976,p 613. (18)M.C.Bernard-Houplain and C. Sandorfy, J . Chem. Phys., 56,3412 (1972).
5813 5656
disappear if D 2 0 is added to the solution
5153 5096 5051 4994 4754
4694 4640 4586
u(”)
I
+ 6(NHz)?
combinations involving ’(”2) Or ‘(“2)
weaken if DzOis added
J
“a = asymmetrical; s = symmetrical; u = stretching; 6 = in-plane bending. coupling between the antisymmetrical and symmetrical modes in known to be large for Hz0.20 The free overtone for T is found at 6662 crn-’, and that for phenobarbital, at 6649 an-’.The related anharmonicity constants are -55 and -56 cm-I. There exists an alternative assignment for the 68 12-cm-I band of NH,: (v3 2 4 , the combination of one quantum of u3 and two quanta of v2, the in-plane N H 2 deformation vibration. While this combination might contribute some intensity to the observed band, as a ternary combination it is less eligible for the assignment 4. than (vl
+
+
(19) A.Burneau and J. Corset, J . Chim. Phys. Phys.-Chim. Biol., 69,153, 171 (1972). , (20)B. T. Darling and D. M. Dennison, Phys. Rev., 57, 128 (1940).
The Journal of Physical Chemistry, Vol. 88, No. 15, 1984 3281
Effect of Barbiturates on Nucleotide Base Pairs
00. O0
7000
6900
6800 6700 (crn-1)
6600
6500
iI
I 7000
6900
6800
6700
6600
6500
6600
6500
6600
6500
(cm-1)
IO-
05 6800
6700
6600
6500
(crn-0
Figure 9. The overtone region in the near-infrared spectrum of 8bromo-2',3',5'-tri-O-acetyladenosine (A 30 mM), 1-methylthymine(T 30
OO i
7000
mM), and phenobarbital (B 30 mM) in CHC13or CDC1. Figure 9 shows the overtones of the free bands of T and phenobarbital. At 30 mM the association bands cannot be identified. The assignment of the above-mentioned bands has been confirmed by deuteration and by recording the spectra at low temperatures. The latter procedure is helpful in distinguishing free from association bands. The C H stretching vibrations are located in the 6000-5600-cm~' region and are of no particular interest for the present study. For this region CDC13 was used as a solvent, in order to avoid interference with the C H stretching band of CHC13. Only the overtones have been used for the purposes of this study. The proliferation of bands of the region of the combination tones is a severe limitation to their usefulness. Figure 10 shows the overtone spectrum of the A / T pair (30 mM of each). All three free bands of the adenine derivative (6997, 6812, and 6743 cm-l) decrease in intensity, indicating H-bond formation between A and T. Figure 10 also shows the spectrum of the complex A/phenobarbital (30 m M of each). We again observe the decrease in intensity of the free NH, bands of A. In addition, a new band appears at 6889 cm-'; we assign it to a H bond formed by the adenine derivatives NH2 and phenobarbital. The strong band at 6649 cm-' receives most of its intensity from the N H stretching vibration of the barbitrate. The spectrum of the triple mixture of A, T , and phenobarbital (Figure 10) is similar to the spectra of the A/B complex, showing that we have Alphenobarbital association. This confirms our results obtained by I R and N M R studies in a n equimolar mixture of adenine, thymine, and phenobarbital. The barbiturate takes the place of thymine. Table I11 gives the values of the related equilibrium constants determined from the intensities of the (free) overtones. The method of calculation was as described above. They are in fair agreement with their values obtained form our N M R data. The band at 6889 cm-' was not used in the calculations because of its weakness. In view of the general weakness of association N H overtones, however, the appearance of this band is a strong indication of adenine/barbiturate complex formation.
Discussion and Conclusions The infrared and N M R studies described in this paper clearly establish that the H-bond equilibrium within our model of the
6900
6800
6700
(cm-1)
7000
6900
6800
6700 (cm-1)
Figure 10. The overtone region in the near-infrared spectrum of the complexes A/T, A/B, and A/B/T in CDCl3 (concentration of the constituents 30 mM). See Figure 9 for abbreviations.
TABLE III: Equilibrium Constants Obtained from the Intensities of the Free NH Overtone Bands in the Near-Infrared Region' K, B, L 'v used, system L mol-' mol-' cm-' cm-'
I-brorno-2',3',5'-tri O-acetyladenosine (2.0 L mol-')" + phenobarbital (7.0 L mol-')' mean: I-bromo-2',3',5'-tri-O-acetyladenosine (2.0 L mol-')' + 1-methylthymine (4.0 L mol-')'
mean:
138 70
0.53 f 0.02 1.09 i 0.02
6812 6743
0.30 f 0.01 0.62 f 0.01 1.05 f 0.01
6997 6812 6743
104 f 34 21 19 24 22 f 3
'Numbers in parentheses indicate self-association constants used for the estimation of equilibrium constants. adenine/thymine pair can be perturbed by barbiturates. The generalization of this to the adenine/thymine pair is straightforward. Indeed, because of the similarity of the structures, the H bonds in our model can only differ very slightly from those in
3282
J. Phys. Chem. 1984, 88, 3282-3287
the true A / T pair. The obvious thought that follows from this is that, if barbiturates succeed in penetrating to molecular distance from the base pairs, an "opening" of some of the A / T H bonds might occur. This, in turn, might initiate cell division which then might be conducive to cancer. Similar ideas have been put forward by Hobza and SandorfyZ1 concerning the interaction of the H bonds in the A / T base pair and a positive ion like Na+. The question to ask is, in both cases, if the perturber can penetrate close enough to the base pair to exert an effective perturbation. Now, the nucleotide base pair is protected by a water layer22which is expected to reduce the probability for this to happen to a very small value. This would apply to Na+ as well as to barbiturates. If, however, a carcinogen like for example an aromatic hydrocarbon perturbs the surrounding water structure, this might facilitate the penetration of the ion or of the barbiturate; then, dissociation of the H bonds in the base pair might occur with serious consequences for the stability of the DNA helix. (21) P.Hobza and C. Sandorfy, Proc. Natl. Acad. Sei. U.S.A.,80, 2859 ( 1983).
(22) E. Clementi and G. Corongiu, Biopolymers, 21, 763 (1982).
Another point of interest is the possibility of trimer formation as a part of the mechanism whereby a barbiturate replaces thymine in the A/T pair. As suggested above, this can lead to an increase of the proportion of the Hoogsteen type pairs which could impair the normal functioning of DNA and, eventually, lead to cancer. The perturbation might affect the mechanism of recognition of given sites in DNA by enzymes and proteins, in genera1.23,24 It is believed that the models used in this study are sufficiently representative to make the above ideas reasonable. The results obtained by IR, near-IR, and proton N M R techniques are concordant in this respect.
Acknowledgment. Financial assistance from the Natural Sciences and Engineering Research Council of Canada and from the Ministhe de 1'Education du Quabec is gratefully acknowledged. We thank Robert Mayer for help in measuring the low-temperature N M R spectra. Registry No. Phenobarbital, 50-06-6; secobarbital, 76-73-3; allobarbital, 52-43-7;pentobarbital, 76-74-4; barbital, 57-44-3. (23) N. C. Seeman, J. M. Rosenberg, and A. Rich, Proc. Natl. Acad. Sei. U.S.A.,73, 804 (1976). (24) C . HBlEne, FEBS Lett., 74, 10 (1977).
Mechanism of the Direct Current Plasma Discharge Decomposition of Disilane P. A. Longeway,* H. A. Weakliem, and R. D. Estes RCA Laboratories, Princeton, New Jersey 08540 (Received: October I I , 1983)
We have studied the static pressure disilane dc discharge using mass spectrometric techniques both in the absence and presence of nitric oxide (NO), a known free-radical scavenger. The observed products of the discharge are H2, SiH4, Si3HB,Si4HL0, and an amorphous hydrogenated silicon film, a-Si:H. The disilane depletion rate and product formation rates were seen to be linear with the discharge current and exhibited a weak pressure dependence. The introduction of NO to the discharge reduced the yields of the products Si3H8and Si4H10to 70% and 21%, respectively, of their values in the absence of NO. The yield of SiH4was unaffected by NO introduction, and the formation of a-Si:H film was totally suppressed. Additionally, the steady-state yield of Si4HI0was reduced by increasing the temperature of surfaces exposed to the discharge, suggesting a surface reaction for the generation of a fraction of this product. The results of these experiments indicate that the relative yields of the reactive fragments SiH2,SiH3, SiH3SiH, and Si2H5are 30%, 34%, 2%, and 34%, respectively. We also conclude that the film precursors from the disilane dc discharge are SiH3and SizH5and that the difference in film properties between silane disilane discharge prepared films can, in part, be explained by noting that the film precursor in the silane case is primarily SiH3, with no significant contributions from Si2H5.
Introduction Disilane has been used to deposit hydrogenated amorphous silicon films, a-Si:H, having desirable photovoltaic properties by both chemical vapor deposition' (CVD) and glow discharge2 (GD) methods. The decomposition of disilane by the CVD method proceeds via a thermal process which may take place in the gas phase or at the surface. The decomposition of disilane in a glow discharge proceeds via electron-impact dissociation, a much more energetic process than the thermal case. Nevertheless, by choosing suitable operating conditions, we may prepare films having comparable properties by either method. One advantage in using disilane rather than silane in the G D method is that the former has deposition rates that are 2-4 times greater, for a specific discharge current, although the hydrogen content of the film is approximately twice that of a film deposited from the silane GD for the same substrate temperatures3 (1) S. C. Gau, B. R. Weinberger, M. Akhtar, 2.Kiss, and A. G. MacDiarmid, Appl. Phys. Lett., 39, 436 (1981). (2) B. A.,Scott,,M. H. Brodsky, D. C. Green, P. B. Kirby, R. M. Plecenik, and E. E. Simonyi, Appl. Phys. Lett., 37, 725 (1980).
0022-3654/84/2088-3282$01.50/0
We have recently studied the decomposition kinetics of the silane glow discharge under both static-pressure4 and flowing-gas5 conditions and concluded from the results that the silyl radical, SiH,, is the primary film-forming intermediate. Furthermore, a surface temperature dependence was observed for the yield of disilane generated from the recombination of two SiH3 units, indicating that a large fraction of the higher silanes in the gas phase may be formed by surface reaction^.^ With these results in mind, we undertook a similar study of the disilane dc discharge. The results of N O quenching experiments indicate that the monoradicals SiH3 and SizHSare primarily responsible for film growth, being generated by the discharge in nearly equal amounts. While the calculated yield of SiH2 from the disilane discharge (30%) is higher than that calculated for the silane discharge4 (
